Intel(R) C++ Compiler 9.0 for Linux --------------------------------------------- Intel(R) C++ compiler options. Used for SGI Altix 350, 3000, 3700 Bx2, and 4700 submissions. ICC(1) Intel(R) C++ Compiler Options ICC(1) NAME icc - invokes the Intel(R) C++ compiler SYNOPSIS icc [ options ] file1 [ file2 ...] where: options represents zero or more compiler options. filen is a C/C++ source (.C .c .cc .cpp .cxx .c++ .i), assembly (.s), object (.o), static library (.a), or other linkable file. NOTE: The icpc command uses the same compiler options as the icc command. Invoking the compiler using icpc com­ piles .c, and .i files as C++. Invoking the compiler using icc compiles .c and .i files as C. Using icpc always links in C++ libraries. Using icc only links in C++ libraries if C++ source is provided on the command line. DESCRIPTION The Intel(R) C++ Compiler is designed to preprocess, com­ pile, assemble, and link C and C++ programs on systems based on Intel architecture. This document explains how information and instructions apply differently to IA-32 applications, Intel(R) EM64T applications, and Itanium(R) architecture-based applica­ tions. If a description does not explicitly state a spe­ cific architecture, it is applicable to all. See the Intel C++ Compiler User's Guide for more complete informa­ tion than is presented here. Some compiler features (options, predefined macros, etc.) are only available on certain systems. In this document, such compiler features are labeled. The label "i32" means the compiler feature is defined on IA-32-based systems; label "i32em" means the compiler feature is defined on Intel(R) Extended Memory 64 Technology (Intel(R) EM64T) systems; label "i64" means the compiler feature is defined on Itanium(R)-based systems. If no label is present, the compiler feature is available on all supported systems. If "only" appears in the label, that compiler feature is only available on the identified system(s). gcc* Interoperability C++ compilers are interoperable if they can link object files and libraries generated by one compiler with object files and libraries generated by the second compiler, and the resulting executable runs successfully. The Intel(R) C++ Compiler options that affect gcc* interoperability include: ⋠-cxxlib- ⋠-gcc-name ⋠-gcc-version ⋠-fabi-version ⋠-no-gcc (see gcc Predefined Macros for more informa­ tion) The Intel(R) C++ Compiler is interoperable with GNU gcc* compiler versions greater than or equal to 3.2. See the Intel C++ Compiler User's Guide for more information. OPTION SUMMARIES Option Alphabetic Summary All of the Intel C++ Compiler options are listed alphabet­ ically in this section. After each listing is a reference to the Functional area where a detailed description of the option appears. -A- see Preprocessor Options -A[] see Preprocessor Options -[no-]alias-args see Advanced Performance Options -[no]align see Language Options -ansi see Language Options -[no-]ansi-alias see Advanced Performance Options -auto-ilp32 see Advanced Performance Options -ax{K|W|N|B|P} see Performance Options, Opti­ mization Levels, Automatic Pro­ cessor-specific Optimization, Vectorization Options -B see Miscellaneous Options -c see Output, Debug, and PCH Options -C see Preprocessor Options -[no-]c99 see Language Options -[no-]complex-limited-range see Advanced Performance Options -create-pch see Output, Debug, and PCH Options; Precompiled Headers -cxxlib- see Linking or Linker Options -D[=] see Preprocessor Options -[no-]debug [keyword] see Output, Debug, and PCH Options -dM see Preprocessor Options -dryrun see Miscellaneous Options -dynamic-linker see Linking or Linker Options -E see Preprocessor Options -EP see Preprocessor Options -export see Language Options -export-dir see Language Options -F see Preprocessor Options -falias see Performance Options -fargument-alias see Advanced Performance Options -fargument-noalias see Advanced Performance Options -fargument-noalias-global see Advanced Performance Options -fast see Performance Options, Opti­ mization Levels -fcode-asm see Output, Debug, and PCH Options -f[no-]common see Miscellaneous Options -fdata-sections see Performance Options -f[no-]exceptions see Performance Options -ffnalias see Performance Options -ffunction-sections see Performance Options -finline-functions see Performance Options -f[no-]math-errno see Miscellaneous Options -fminshared see Miscellaneous Options -fno-alias see Performance Options -fno-builtin see Miscellaneous Options -fno-builtin- see Performance Options, Miscel­ laneous Options -fno-fnalias see Performance Options -fno-gnu-keywords see Language Options -fno-implicit-inline-templates see Language Options -fno-implicit-templates see Language Options -f[no-]omit-frame-pointer see Performance Options -fno-operator-names see Language Options -fnsplit[-] see Advanced Performance Options -fp see Performance Options -fpack-struct see Miscellaneous Options -fpermissive see Language Options -f[no-]pic see Miscellaneous Options -f[no-]PIC see Miscellaneous Options -fp-model see Advanced Performance Options -[no-]fp-port see Performance Options -fpstkchk see Performance Options -fr32 see Miscellaneous Options -f[no-]rtti see Language Options -freg-struct-return see Miscellaneous Options -fshort-enums see Language Options -fsource-asm see Output, Debug, and PCH Options -fstack-security-check see Miscellaneous Options -fsyntax-only see Language Options -ftemplate-depth- see Language Options -ftls-model= see Advanced Performance Options -ftrapuv see Output, Debug, and PCH Options -[no-]ftz see Advanced Performance Options -funroll-loops see Advanced Performance Options -funsigned-bitfields see Language Options -funsigned-char see Language Options -f[no]verbose-asm see Output, Debug, and PCH Options -fvisibility=[extern|default|protected|hid­ den|internal] see Miscellaneous Options -fvisibility-default= see Miscellaneous Options -fvisibility-extern= see Miscellaneous Options -fvisibility-hidden= see Miscellaneous Options -fvisibility-internal= see Miscellaneous Options -fvisibility-protected= see Miscellaneous Options -g see Output, Debug, and PCH Options -g0 see Output, Debug, and PCH Options -gcc-name= see Miscellaneous Options -gcc-version= see Miscellaneous Options -[no-]global-hoist see Miscellaneous Options -H see Preprocessor Options -help see Miscellaneous Options -I see Preprocessor Options -idirafter see Preprocessor Options -i-dynamic see Linking or Linker Options -imacros see Preprocessor Options -inline-debug-info see Output, Debug, and PCH Options -ip see Advanced Performance Options, Interprocedural Optimizations (IPO) -[no-]IPF-fltacc see Advanced Performance Options -IPF-flt-eval-method0 see Advanced Performance Options -[no-]IPF-fma see Advanced Performance Options -[no-]IPF-fp-relaxed see Advanced Performance Options -IPF-fp-speculation{fast|safe|strict|off} see Advanced Performance Options -ip-no-inlining see Advanced Performance Options -ip-no-pinlining see Advanced Performance Options -ipo[value] see Advanced Performance Options, Interprocedural Optimizations (IPO) -ipo-c see Advanced Performance Options, Interprocedural Optimizations (IPO) -[no-]ipo-obj see Advanced Performance Options, Interprocedural Optimizations (IPO) -ipo-S see Advanced Performance Options, Interprocedural Optimizations (IPO) -ipo-separate see Advanced Performance Options, Interprocedural Optimizations (IPO) -iprefix see Preprocessor Options -i-static see Linking or Linker Options -isystem see Preprocessor Options -ivdep-parallel see Advanced Performance Options -iwithprefix see Preprocessor Options -iwithprefixbefore see Preprocessor Options -Kc++ see Language Options -kernel see Miscellaneous Options -L see Linking or Linker Options -long_double see Miscellaneous Options -M see Preprocessor Options -malign-double see Language Options -map-opts see Output, Debug, and PCH Options -march={pentiumpro|pentiumii|pentiumiii|pentium4} see Performance Options -mcpu={pentium|pentiumpro|pentium4|itanium|ita­ nium2} see Performance Options -MD see Preprocessor Options -MF see Preprocessor Options -mfixed-range=f12-f15,f32-f127 see Miscellaneous Options -MG see Preprocessor Options -m[no-]ieee-fp see Performance Options -MM see Preprocessor Options -MMD see Preprocessor Options -mno-relax see Linking or Linker Options -mno-serialize-volatile see Advanced Performance Options -mp see Performance Options -MP see Preprocessor Options -mp1 see Performance Options -MQ see Preprocessor Options -mrelax see Linking or Linker Options -mserialize-volatile see Advanced Performance Options -msse see Performance Options -msse2 see Performance Options -msse3 see Performance Options -MT see Preprocessor Options -[no-]multibyte-chars see Miscellaneous Options -nobss-init see Miscellaneous Options -no-cpprt see Linking or Linker Options -nodefaultlibs see Linking or Linker Options -no-gcc see Preprocessor Options -nostartfiles see Linking or Linker Options -nostdinc see Preprocessor Options -nostdlib see Linking or Linker Options -mtune= see Performance Options -o see Output, Debug, and PCH Options -O0 see Performance Options, Opti­ mization Levels -O1 see Performance Options, Opti­ mization Levels -O2 see Performance Options, Opti­ mization Levels -O3 see Performance Options, Opti­ mization Levels -Ob see Performance Options -openmp see Advanced Performance Options, Parallelization with OpenMP* -openmp-profile see Advanced Performance Options, Parallelization with OpenMP* -openmp-report{0|1|2} see Advanced Performance Options, Parallelization with OpenMP* -openmp-stubs see Advanced Performance Options, Parallelization with OpenMP* -opt-report see Advanced Performance Options -opt-report-file see Advanced Performance Options -opt-report-help see Advanced Performance Options -opt-report-level[{min|med|max}] see Advanced Performance Options -opt-report-phase{ipo|hlo|ilo|ecg|omp|all} see Advanced Performance Options, Parallelization with OpenMP* -opt-report-routine see Advanced Performance Options -Os see Performance Options -p see Advanced Performance Options -P see Preprocessor Options -parallel see Advanced Performance Options, Auto Parallelization Options -par-report{0|1|2|3} see Advanced Performance Options, Auto Parallelization Options -par-threshold[] see Advanced Performance Options, Auto Parallelization Options -pc{32|64|80} see Performance Options -pch see Output, Debug, and PCH Options; Precompiled Headers -pch-dir see Output, Debug, and PCH Options; Precompiled Headers -[no-]prec-div see Performance Options -[no-]prec-sqrt see Performance Options -[no-]prefetch see Advanced Performance Options -print-multi-lib see Output, Debug, and PCH Options -prof-dir see Advanced Performance Options, Profile-guided Optimizations (PGO) -prof-file see Advanced Performance Options, Profile-guided Optimizations (PGO) -prof-gen[x] see Advanced Performance Options, Profile-guided Optimizations (PGO) -prof-gen-sampling see Advanced Performance Options, Profile-guided Optimizations (PGO) -prof-use see Advanced Performance Options, Profile-guided Optimizations (PGO) -Qinstall see Component Control Options -Qlocation,, see Component Control Options -Qoption,, see Component Control Options -qp see Advanced Performance Options -rcd see Performance Options -reserve-kernel-regs see Miscellaneous Options -[no]restrict see Language Options -S see Output, Debug, and PCH Options -shared see Linking or Linker Options -shared-libcxa see Linking or Linker Options -[no-]sox see Miscellaneous Options -ssp see Preprocessor Options -static see Linking or Linker Options -static-libcxa see Linking or Linker Options -std=c99 see Language Options -strict-ansi see Language Options -T see Linking or Linker Options -tcheck see Advanced Performance Options -[no]traceback see Compiler Diagnostics Options -u see Linking or Linker Options -U see Preprocessor Options -unroll see Advanced Performance Options -unroll0 see Advanced Performance Options -use-asm see Output, Debug, and PCH Options -use-msasm see Output, Debug, and PCH Options -use-pch see Output, Debug, and PCH Options; Precompiled Headers -V see Miscellaneous Options -v see Miscellaneous Options -vec-report[{0|1|2|3|4|5}] see Advanced Performance Options --version see Miscellaneous Options -w see Compiler Diagnostics Options -w see Compiler Diagnostics Options -Wa[,,...] see Linking or Linker Options -W[no-]abi see Compiler Diagnostics Options -Wall see Compiler Diagnostics Options -Wbrief see Compiler Diagnostics Options -Wcheck see Compiler Diagnostics Options -W[no-]comment[s] see Compiler Diagnostics Options -Wcontext-limit= see Compiler Diagnostics Options -wd[,,...] see Compiler Diagnostics Options -W[no-]deprecated see Compiler Diagnostics Options -we[,,...] see Compiler Diagnostics Options -Werror see Compiler Diagnostics Options -Winline see Compiler Diagnostics Options -Wl,[,,...] see Linking or Linker Options -W[no-]main see Compiler Diagnostics Options -W[no-]missing-prototypes see Compiler Diagnostics Options -wn see Compiler Diagnostics Options -Wp64 see Compiler Diagnostics Options -Wp,[,,...] see Linking or Linker Options, Precompiled Headers -W[no-]pointer-arith see Compiler Diagnostics Options -wr[,,...] see Compiler Diagnostics Options -W[no-]return-type see Compiler Diagnostics Options -W[no-]shadow see Compiler Diagnostics Options -W[no-]uninitialized see Compiler Diagnostics Options -W[no-]unknown-pragmas see Compiler Diagnostics Options -W[no-]unused-function see Compiler Diagnostics Options -ww[,,...] see Compiler Diagnostics Options -X see Preprocessor Options -x see Miscellaneous Options -x{K|W|N|B|P} see Performance Options, Opti­ mization Levels, Optimizing Exclusively for Specific Proces­ sors, Vectorization Options -Xlinker see Linking or Linker Options -Zp[n] see Language Options Option Functional Summary All the options listed in the following summary are described in more detail in the sections that follow. Performance Options -ax{K|W|N|B|P} -falias -fast -fdata-sec­ tions -f[no-]exceptions -ffnalias -ffunc­ tion-sections -finline-functions -fno-alias -fno-builtin- -fno-fna­ lias -f[no-]omit-frame-pointer -fp -[no-]fp-port -fpstkchk -march={pen­ tiumpro|pentiumii|pentiumiii|pentium4} -mcpu={pentium|pentiumpro|pentium4|ita­ nium|itanium2} -m[no-]ieee-fp -mp -mp1 -msse -msse2 -msse3 -mtune= -O0 -O1 -O2 -O3 -Ob -pc{32|64|80} -[no-]prec-div -[no-]prec-sqrt -rcd -x{K|W|N|B|P} Advanced Performance Options -[no-]alias-args -[no-]ansi-alias -auto-ilp32 -[no-]complex-limited-range -fargument-alias -fargument-noalias -far­ gument-noalias-global -fnsplit[-] -fp-model -ftls-model= -[no-]ftz -funroll-loops -ip -[no-]IPF-fltacc -IPF-flt-eval-method0 -[no-]IPF-fma -[no-]IPF-fp-relaxed -IPF-fp-speculation{fast|safe|strict|off} -ip-no-inlining -ip-no-pinlining -ipo[value] -ipo-c -[no-]ipo-obj -ipo-S -ipo-separate -ivdep-parallel -mno-serial­ ize-volatile -mserialize-volatile -openmp -openmp-profile -openmp-report{0|1|2} -openmp-stubs -opt-report -opt-report-file -opt-report-help -opt-report-level[{min|med|max}] -opt-report-phase{ipo|hlo|ilo|ecg|omp|all} -opt-report-routine -p -parallel -par-report{0|1|2|3} -par-threshold[] -[no-]prefetch -prof-dir -prof-file -prof-gen[x] -prof-gen-sampling -prof-use -qp -tcheck -unroll -unroll0 -vec-report[{0|1|2|3|4|5}] Output, Debug, and PCH Options -c -create-pch -[no-]debug[keyword] -fcode-asm -fsource-asm -ftrapuv -f[no]verbose-asm -g -g0 -inline-debug-info -map-opts -o -pch -pch-dir -print-multi-lib -S -use-asm -use-msasm -use-pch Preprocessor Options -A- -A[] -C -D[=] -dM -E -EP -H -I -idirafter -imacros -iprefix -isystem -iwithpre­ fix -iwithprefixbefore -M -MD -MF -MG -MM -MMD -MP -MQ -MT -no-gcc -nostdinc -P -ssp -U -X Component Control Options -Qinstall -Qlocation,, -Qoption,, Language Options -[no]align -ansi -[no-]c99 -export -export-dir -fno-gnu-keywords -fno-implicit-inline-templates -fno-implicit-templates -fno-operator-names -fpermissive -f[no-]rtti -fshort-enums -fsyntax-only -ftemplate-depth- -fun­ signed-bitfields -funsigned-char -Kc++ -malign-double -[no]restrict -std=c99 -strict-ansi -Zp[n] Compiler Diagnostics Options -[no]traceback -w -w -W[no-]abi -Wall -Wbrief -Wcheck -W[no-]comment[s] -Wcon­ text-limit= -wd[,,...] -W[no-]deprecated -we[,,...] -Wer­ ror -Winline -W[no-]main -W[no-]miss­ ing-prototypes -wn -Wp64 -W[no-]pointer-arith -wr[,,...] -W[no-]return-type -W[no-]shadow -W[no-]uninitialized -W[no-]unknown-pragmas -W[no-]unused-function -ww[,,...] Miscellaneous Options -B -dryrun -f[no-]common -f[no-]math-errno -fminshared -fno-builtin -fno-builtin- -fpack-struct -f[no-]pic -f[no-]PIC -fr32 -freg-struct-return -fstack-security-check -fvisibility=[extern|default|protected|hid­ den|internal] -fvisibility-default= -fvisibility-extern= -fvisibil­ ity-hidden= -fvisibility-inter­ nal= -fvisibility-protected= -gcc-name= -gcc-version= -[no-]global-hoist -help -kernel -long_double -mfixed-range=f12-f15,f32-f127 -[no-]multibyte-chars -nobss-init -reserve-kernel-regs -[no-]sox -V -v --version -x Linking or Linker Options -cxxlib- -dynamic-linker -i-dynamic -i-static -L -mno-relax -mrelax -no-cpprt -nodefaultlibs -nos­ tartfiles -nostdlib -shared -shared-libcxa -static -static-libcxa -T -u -Wa[,,...] -Wl,[,,...] -Wp,[,,...] -Xlinker OPTION DESCRIPTIONS Performance Options The performance options act using general criteria to optimize code. The optimizations listed in this section are available for Intel architectures or are compatible with gcc compiler options. -ax (i32, i32em only) Generate code specialized for processor extensions specified by while also generating generic IA-32 code. includes one or more of the following characters: K -- Intel Pentium III processors and com­ patible Intel processors W -- Intel Pentium 4 processors and compati­ ble Intel processors N -- Intel Pentium 4 processors and compati­ ble Intel processors. Enables new optimiza­ tions in addition to Intel processor-spe­ cific optimizations. B -- Intel Pentium M processors and compati­ ble Intel processors. Enables new optimiza­ tions in addition to Intel processor-spe­ cific optimizations. P -- Intel Pentium 4 Processors with Streaming SIMD Extensions 3 (SSE3) instruc­ tion support. Enables new optimizations in addition to Intel processor-specific opti­ mizations. NOTE: The only -ax options available on Intel(R) EM64T-based systems are -axW and -axP. -falias Assume aliasing in the program (DEFAULT). -fast The -fast option maximizes speed across the entire program. It sets command options that can improve run-time performance, as follows: i32, i32em: The -fast option turns on -O3, -ipo, -static, and -xP. IA-32 programs com­ piled with -fast will execute only on Intel(R) Pentium(R) 4 processors with Streaming SIMD Extensions 3 (SSE3). Pro­ grams compiled with this option will detect non-compatible processors and generate an error message during execution. i64: The -fast option turns on -O3, -ipo, and -static. -fdata-sections Separate functions for the linker (COMDAT). Same as -ffunction-sections. -f[no-]exceptions The -f[no-]exceptions option turns off exception handling table generation, result­ ing in smaller code. Any use of exception handling constructs - try blocks, throw statements will produce an error. Exception specifications are parsed but ignored. A preprocessor symbol __EXCEPTIONS is defined when this option is not used. It is unde­ fined when this option is present. -ffnalias Assume aliasing within functions (DEFAULT). -ffunction-sections Separate functions for the linker (COMDAT). Same as -fdata-sections. -finline-functions Inline any function at the compiler's dis­ cretion. Same as -ip. -fno-alias Assume no aliasing in program. -fno-builtin- Disable the intrinsic. -fno-fnalias Assume no aliasing within functions, but assume aliasing across calls. -f[no-]omit-frame-pointer The negative version is the same as -fp. -fp (i32, i32em only) Disable using EBP as general purpose regis­ ter. -[no-]fp-port (i32, i32em only) Round floating-point results at assignments and casts (some speed impact). -fpstkchk (i32, i32em only) Generate extra code after every function call to assure that the FP stack is in the expected state. Generally, when the FP stack overflows, a NaN value is put into FP calcu­ lations, and the program's results differ. Unfortunately, the overflow point can be far away from the point of the actual bug. The -fpstkchk option places code that would access violate immediately after an incor­ rect call occurred, thus making it easier to locate these issues. -march= (i32 only) Generate code exclusively for a given cpu, where is one of the following: pentiumpro -- Intel Pentium Pro processors pentiumii -- Intel Pentium II processors pentiumiii -- Intel Pentium III processors pentium4 -- Intel Pentium 4 processors -mcpu= Optimize for a specific cpu, where is one of the following: pentium -- (i32 only) Optimize for Intel Pentium processor. pentiumpro -- (i32 only) Optimize for Intel Pentium Pro, Intel Pentium II and Intel Pen­ tium III processors. pentium4 -- (i32 only; DEFAULT on IA-32) Optimize for Intel Pentium 4 processors. itanium -- (i64 only) Optimize for Intel Itanium processor itanium2 -- (i64 only) Optimize for Intel Itanium 2 processor. -m[no-]ieee-fp -mp Maintain floating-point precision (disables some optimizations). The -mp option restricts optimization to maintain declared precision and to ensure that floating-point arithmetic conforms more closely to the ANSI and IEEE standards. For most programs, spec­ ifying this option adversely affects perfor­ mance. If you are not sure whether your application needs this option, try compiling and running your program both with and with­ out it to evaluate the effects on both per­ formance and precision. -mp1 Improve floating-point precision. -mp1 dis­ ables fewer optimizations and has less impact on performance than -mp. -msse (i32 only) Generate code for Intel Pentium III and com­ patible Intel processors. -msse2 (i32 only) Generate code for Intel Pentium 4 and com­ patible Intel processors. -msse3 (i32, i32em only) Generate code for Intel Pentium 4 processors with SSE3 extensions. -mtune= Optimize for a specific processor. -O0 Disable optimizations. -O Same as -O2 -O1 Optimize to favor code size and code local­ ity. Disables loop unrolling. -O1 may improve performance for applications with very large code size, many branches, and execution time not dominated by code within loops. In most cases, -O2 is recommended over -O1. i32: Disable intrinsics inlining to reduce code size. i64: Disable software pipelining and global code scheduling. -O2 (DEFAULT) Optimize for code speed. This is the gener­ ally recommended optimization level. i64: Turn software pipelining ON. -O3 Enable -O2 optimizations and in addition, enable more aggressive optimizations such as loop and memory access transformation, and prefetching. The -O3 option optimizes for maximum speed, but may not improve perfor­ mance for some programs. The -O3 optimiza­ tions may slow down code in some cases com­ pared to -O2 optimizations. Recommended for applications that have loops with heavy use of floating point calculations and process large data sets. Turn on high-level optimizations. Enable -O2 plus more aggressive optimizations, such as loop transformation. i32, i32em: In conjunction with -ax{K|W|N|B|P} and -x{K|W|N|B|P} options, this option causes the compiler to perform more aggressive data dependency analysis than for -O2. This may result in longer com­ pilation times. -Ob Control inline expansion, where is one of the following values: 0 -- Disables inlining. 1 -- (DEFAULT) Enables inlining of functions declared with the __inline keyword. Also enables inlining according to the C++ lan­ guage. 2 -- Inlines any function, at the compiler's discretion. Enables interprocedural opti­ mizations and has the same effect as -ip. -Os Enable speed optimizations, but disable some optimizations that increase code size for small speed benefit. -pc (i32, i32em only) Enable floating-point significand precision control. Some floating-point algorithms are sensitive to the accuracy of the signifi­ cand, or fractional part of the float­ ing-point value. For example, iterative operations like division and finding the square root can run faster if you lower the precision with the -pc option. Set to one of the following values to round the significand to the indicated number of bits: 32: 24 bits (single precision) -- Caution: A change of the default precision control or rounding mode (for example, by using the -pc32 flag or by user intervention) may affect the results returned by some of the mathematical functions. 64: 53 bits (single precision) 80: 64 bits (double precision) -- DEFAULT -[no-]prec-div (i32, i32em only) Improve precision of floating-point divides (some speed impact). With some optimizations the Intel C++ Compiler changes float­ ing-point division computations into multi­ plication by the reciprocal of the denomina­ tor. For example, A/B is computed as A x (1/B) to improve the speed of the computa­ tion. However, for values of B greater than 2126, the value of 1/B is "flushed" (changed) to 0. When it is important to maintain the value of 1/B, use -[no-]prec-div to disable the floating-point division-to-multiplication optimization. The result of -[no-]prec-div is more accurate, with some loss of performance. -[no-]prec-sqrt (i32, i32em only) Determine if certain square root optimiza­ tions are enabled. -rcd (i32, i32em only) Enable fast float-to-int conversions. The Intel compiler uses the -rcd option to improve the performance of code that requires floating-point-to-integer conver­ sions. The system default floating point rounding mode is round-to-nearest. However, the C language requires floating point val­ ues to be truncated when a conversion to an integer is involved. To do this, the com­ piler must change the rounding mode to trun­ cation before each floating-point-to-integer conversion and change it back afterwards. The -rcd option disables the change to trun­ cation of the rounding mode for all floating point calculations, including floating point-to-integer conversions. Turning on this option can improve performance, but floating point conversions to integer will not conform to C semantics. -x (i32, i32em only) Generate specialized code to run exclusively on processors supporting the extensions indicated by . includes one or more of the following characters: K -- Intel Pentium III processors and com­ patible Intel processors W -- Intel Pentium 4 processors and compati­ ble Intel processors N -- Intel Pentium 4 processors and compati­ ble Intel processors. Enables new optimiza­ tions in addition to Intel processor-specific optimizations. B -- Intel Pentium M processors and compati­ ble Intel processors. Enables new optimiza­ tions in addition to Intel processor-spe­ cific optimizations. P -- Intel Pentium 4 Processors with Streaming SIMD Extensions 3 (SSE3) instruc­ tion support. Enables new optimizations in addition to Intel processor-specific opti­ mizations. Note: The only options available on Intel(R) EM64T-based systems are -xW and -xP. You can use more than one of the -x options by combining the characters that denote the processor type. For example, you can specify -xKW to generate code for Intel(R) Pentium(R) III and Intel Pentium 4 proces­ sors (though the -xW option does not support i32em features). Note: Programs compiled using option -xN, -xB, or -xP will display a fatal run-time error if executed by an unsupported proces­ sor. For more information, see the Intel C++ Compiler User's Guide. Advanced Performance Options The advanced performance options allow fine tuning of compilation or allow control over advanced fea­ tures of the compiler. -[no-]alias-args Assume arguments may be aliased. (DEFAULT) [not aliased]. -[no-]ansi-alias -[no-]ansi_alias The -[no-]ansi-alias option directs the com­ piler to assume that the program adheres to the rules defined in the ISO C Standard. If your program adheres to these rules, then this option will allow the compiler to opti­ mize more aggressively. If it doesn't adhere to these rules, then it can cause the compiler to generate incorrect code. -auto-ilp32 (i32em, i64 only) Specifies that the application should run within a 32-bit address space. Also tells the compiler to use 32-bit pointers whenever possible. To use this option, you must spec­ ify -ipo. Note: The -auto-ilp32 option applies to i64 and i32em, but with i32em it has no effect unless option -xP or option -axP is also specified. It does not apply to i32. -[no-]complex-limited-range Tell the compiler to use the highest perfor­ mance formulations of complex arithmetic operations, which may not produce acceptable results for input values near the top or bottom of the legal range. Without this option, the compiler uses a better formula­ tion of complex arithmetic operations, thus producing acceptable results for the full range of input values, though at some loss in performance. -fargument-alias Same as -alias-args. -fargument-noalias Same as -no-alias-args. -fargument-noalias-global Arguments do not alias each other and do not alias global storage. -fnsplit[-] Enable[disable] function splitting. Function splitting is enabled by -prof-use in Phase 3 to improve code locality by splitting rou­ tines into different sections: one section to contain the cold or very infrequently executed code, and one section to contain the rest of the code (hot code). You can use -fnsplit[-] to disable function splitting for the following reasons: ⋠Most importantly, to get improved debugging capability. In the debug sym­ bol table, it is difficult to represent a split routine, that is, a routine with some of its code in the hot code section and some of its code in the cold code section. ⋠The -fnsplit[-] option disables the splitting within a routine but enables function grouping, an optimization in which entire routines are placed either in the cold code section or the hot code section. Function grouping does not degrade debugging capability. ⋠When the profile data does not repre­ sent the actual program behavior, that is, when the routine is actually used frequently rather than infrequently. -fp-model Enable floating point model variation. -ftls-model= Change thread-local storage model, where can be the following: ⋠global-dynamic ⋠local-dynamic ⋠initial-exec ⋠local-exec -ftz (i64 only) Flush denormal results to zero. -funroll-loops Unroll loops based on default heuris­ tics. -ip Enable single-file IP optimizations (within files). With this option, the compiler performs inline function expansion for calls to functions defined within the current source file. -IPF-flt-eval-method0 (i64 only) Direct the compiler to evaluate the expressions involving floating-point operands in the precision indicated by the variable types declared in the pro­ gram. -[no-]IPF-fltacc (i64 only) Enable [disable] optimizations that affect floating point accuracy. By default (-IPF-fltacc-) the compiler may apply optimizations that reduce float­ ing-point accuracy. You may use -IPF-fltacc or -mp to improve float­ ing-point accuracy, but at the cost of disabling some optimizations. -[no-]IPF-fma (i64 only) Enable [disable] the combining of floating point multiplies and add/sub­ tract operations. -IPF-fma[-] enables [disables] the contraction of float­ ing-point multiply and add/subtract operations into a single operation. Unless -mp is specified, the compiler contracts these operations whenever possible. The -mp option disables the contractions. -IPF-fma and -IPF-fma- can be used to override the default compiler behavior. -[no-]IPF-fp-relaxed (i64 only) -[no-]IPF_fp_relaxed (i64 only) Enable [disable] use of faster but slightly less accurate code sequences for math functions, such as divide and square root. -IPF-fp-speculation{fast|safe|strict|off} (i64 only) Enable floating point speculations with the following conditions: fast -- Speculate floating point opera­ tions safe -- Speculate only when safe strict -- Same as off off -- Disables speculation of float­ ing-point operations -ip-no-inlining Disable inlining that would result from the -ip interprocedural optimization, but has no effect on other interproce­ dural optimizations. -ip-no-pinlining (i32, i32em only) Disable partial inlining (requires -ip or -ipo) -ipo[value] Enables multifile IP optimizations (between files). When you specify this option, the compiler performs inline function expansion for calls to func­ tions defined in separate files. The [value] argument is an optional integer that specifies the number of object files the compiler should cre­ ate. Any integer greater than or equal to 0 is valid. If [value] is 0, the compiler decides whether to create one or more object files based on an estimate of the size of the object file. It generates one object file for small applications, and two or more object files for large applications. If [value] is positive, the compiler generates [value] object files, unless [value] exceeds the number of source files (m), in which case the compiler generates only m object files. If you do not specify [value], the default is 1. -ipo-c Generate a multi-file object file (ipo-out.o) that can be used in further link steps. -[no-]ipo-obj Force the compiler to create real object files when used with -ipo (requires -ipo). -ipo-S Generate a multi-file assembly file (ipo-out.s) that can be used in further link steps. -ipo-separate Creates one object file for every source file. This option overrides -ipo[value]. -ivdep-parallel (i64 only) Indicate there is absolutely no loop-carried memory dependency in any loop where IVDEP directive is speci­ fied. This technique is useful for some sparse matrix applications. -mno-serialize-volatile (i64 only) Memory access ordering for volatile data object references may be sup­ pressed. -mserialize-volatile (i64 only) Enable strict memory access ordering for volatile data object references. -openmp Enable the parallelizer to generate multi-threaded code based on the OpenMP* directives. The code can be executed in parallel on both uniproces­ sor and multiprocessor systems. The -openmp option works with both -O0 (no optimization) and any optimization level of -O1, -O2 (default) and -O3. Specifying -O0 with -openmp helps to debug OpenMP applications. -openmp-profile Enables analysis of OpenMP* applica­ tions with Intel(R) Thread Profiler, which is required to use this option. -openmp-report[] Control the OpenMP parallelizer's level of diagnostic messages, where is one of the following: 0 -- no diagnostic information is dis­ played. 1 -- Display diagnostics indicating loops, regions, and sections success­ fully parallelized. (DEFAULT) 2 -- same as -openmp-report1 plus diag­ nostics indicating MASTER constructs, SINGLE constructs, CRITICAL constructs, ORDERED constructs, ATOMIC directives, etc. successfully handled. -openmp-stubs Enable the user to compile OpenMP pro­ grams in sequential mode. The openmp directives are ignored and a stub OpenMP library is linked (sequential). -opt-report (i32em, i64 only) Generate an optimization report and direct it to stderr. DEFAULT: The compiler does not generate optimization reports. -opt-report-file (i32em, i64 only) Specify the filename for the generated report -opt-report-help (i32em, i64 only) List the logical names of optimizers available for report generation. -opt-report-level[] (i32em, i64 only) Specify the level of report verbosity, where is one of the following: min -- The min argument provides the minimal summary (DEFAULT) med -- The med argument produces an intermediate-level report. max -- The max argument produces the full report. -opt-report-phase (i32em, i64 only) Specify the optimizer phase against which reports are generated. The com­ piler generates reports for the opti­ mizer you specify in the . This option can be used multiple times on the same command line to generate reports for multiple optimizers. Cur­ rently, the following optimizer reports are supported: ipo -- Interprocedural Optimizer hlo -- High Level Optimizer ilo -- Intermediate Language Scalar Optimizer ecg -- Code Generator omp -- OpenMP all -- All phases When one of the above logical names for optimizers is specified for , all reports from that optimizer are gener­ ated. -opt-report-routine (i32em, i64 only) Generate reports from all routines with names containing as part of their name. If is not spec­ ified, reports from all routines are generated. DEFAULT: The compiler generates reports for all routines. -p Compile and link for function profiling with Linux gprof* tool. Same as -qp. -parallel Enable the auto-parallelizer to gener­ ate multi-threaded code for loops that can be safely executed in parallel. The -parallel option enables the auto-par­ allelizer if either the -O2 or -O3 optimization option is also on (the default is -O2). -par-report[] Control the level of auto-parallelizer diagnostic messages, where is one of the following: 0 -- no diagnostic information is dis­ played. 1 -- indicates loops successfully auto-parallelized (DEFAULT). Issues a "LOOP AUTO-PARALLELIZED" message for parallel loops. 2 -- indicates successfully auto-paral­ lelized loops as well as unsuccessful loops. 3 -- same as 2 plus additional informa­ tion about any proven or assumed depen­ dencies inhibiting auto-parallelization (reasons for not parallelizing). -par-threshold[] Set a threshold for the auto-paral­ lelization of loops based on the proba­ bility of profitable execution of the loop in parallel. This option is used for loops whose computation work volume cannot be determined at compile-time. The threshold is usually relevant when the loop trip count is unknown at com­ pile-time. =0-100. (DEFAULT: =75) The compiler applies a heuristic that tries to balance the overhead of creat­ ing multiple threads versus the amount of work available to be shared amongst the threads. -[no-]prefetch Enables [disables] the insertion of software prefetching by the compiler. Default is -prefetch. -prof-dir Specify directory for profiling output files (*.dyn and *.dpi). Use the -prof-dir option with prof-gen as rec­ ommended for most programs, especially if the application includes the source files located in multiple directories. -prof-dir ensures that the profile information is generated in one consis­ tent place. -prof-file Specify file name for profiling summary file -prof-gen[x] Instruct the compiler to produce instrumented code in your object files in preparation for instrumented execu­ tion. With the x qualifier, extra information is gathered. This option is used in Phase 1 of PGO to instruct the compiler to produce instrumented code in your object files in preparation for instrumented execution. Parallel make is automatically supported for -prof-genx compilations. -prof-gen-sampling Prepare code for use with the profrun sample gathering tool. -prof-use Instruct the compiler to produce a pro­ file-optimized executable and merge available dynamic information (.dyn) files into a pgopti.dpi file. Use the -prof-use option in Phase 3 of PGO. -qp Compile and link for function profiling with Linux gprof* tool. Same as -p. -tcheck The -tcheck compiler option enables analysis of threaded applications with Intel(R) Thread Checker, which is required to use this option. -unroll (i32, i32em only) Set maximum number of times to unroll loops. This applies only to loops that the compiler determines should be unrolled. Omit n to let the compiler decide whether to perform unrolling or not. Use n=0 to disable loop unrolling. -unroll0 (i64 only) Disable loop unrolling (i64-only). For i64: the only allowed value is 0. -vec-report[] (i32, i32em only) Control the vectorizer's level of diag­ nostic messages, where is one of the following: 0 -- No diagnostic information 1 -- Indicate vectorized loops (DEFAULT) 2 -- Indicate vectorized/non-vectorized loops 3 -- Indicate vectorized/non-vectorized loops and prohibiting data dependence information 4 -- Indicate non-vectorized loops 5 -- Indicate non-vectorized loops and prohibiting data dependence information If you use -c, -ipo with -vec-report{n} option or -c, -x{K|W|N|B|P} or -ax{K|W|N|B|P} with -vec-report[], the compiler issues a warning and no report is generated. To produce a report when using the aforementioned options, you need to add the -ipo-obj option. The combination of -c and -ipo-obj produces a single file compilation, and hence does generate object code, and eventually a report is generated. Output, Debug, and PCH Options The output, debug, and PCH options provide control over compiler output, setup for debugging, and use of the precompiled header features of the compiler. -c Compile to object (.o) only, do not link. -create-pch Manual creation of precompiled header . Use the -create-pch option if you want the compiler to create a PCH file called . The filename param­ eter must be specified. The filename param­ eter can be a full path name. The full path to filename must exist. The .pchi extension is not automatically appended to filename. This option cannot be used in the same com­ pilation as -use-pch filename. The -create-pch filename option is supported for single source file compilations only. -[no-]debug[keyword] Enable debug information and control output of enhanced debug information. The sup­ ported [keywords] are extended, [no]vari­ able-locations, [no]semantic-stepping, [no-]expr-source-pos, and [no]inline_debug_info}. The debug switches control emission of enhanced debug informa­ tion. They must be used in conjunction with the basic -g switches that request debug information. extended -- Turns on the -debug options; ⋠-debug [no-]expr-source-pos, ⋠-debug [no]variable-locations, ⋠-debug [no]semantic-stepping, and ⋠-debug [no]inline_debug_info. [no-]expr-source-pos -- Determines whether source position information at the expression level of granularity is produced. [no]variable-locations -- Determines whether enhanced debug information use­ ful in finding scalar local variables is produced. [no]semantic-stepping -- Determines whether enhanced debug information use­ ful for breakpoints and stepping is produced. [no]inline_debug_info -- Determines whether enhanced debug information is produced for inlined code. -fcode-asm Produce assembly file with optional code annotations. -fsource-asm Produce assembly file with optional code annotations. -ftrapuv Initializes stack local variables to an unusual value to help error detection. -f[no]verbose-asm Produce assembly file with compiler comments (DEFAULT). -g Produce symbolic debug information in object file. The compiler does not sup­ port the generation of debugging infor­ mation in assemblable files. If you specify the -g option, the resulting object file will contain debugging information, but the assemblable file will not. The -g option changes the default optimization from -O2 to -O0. If you specify -g with -O1, -O2, or -O3, then -fp is disabled and allows the compiler to use the EBP register as a general-purpose register in optimiza­ tions. However, most debuggers expect EBP to be used as a stack frame pointer, and cannot produce a stack backtrace unless this is so. Using the -fp option can result in slightly less efficient code. i32 only: Specifying the -g or -O0 option automatically enables the -fp option. -g0 Disable generation of symbolic debug information. -inline-debug-info Preserve the source position of inlined code instead of assigning the call-site source position to inlined code. -map-opts Enable option mapping tool. -o Name output file. -pch Automatic processing for precompiled headers. Direct the compiler to use appropriate PCH files. If none are available, they are created as source­ file.pchi. This option supports multi­ ple source files. Note: The -pch option uses PCH files created from other sources if the head­ ers files are the same. For example, if you compile source1.cpp using -pch, then source1.pchi is created. If you then compile source2.cpp using -pch, the compiler will use source1.pchi if it detects the same headers. -pch-dir Directs the compiler to find and/or create a file for pre-compiled headers in dirname. Use the -pch-dir option to specify the path (dirname) to the PCH file. You can use this option with -pch, -create-pch filename, and -use-pch filename. -print-multi-lib Print information about libraries being used. -S Compile to an assemblable file (.s), then stop the compilation. -use-asm Produce objects through the assembler. -use-msasm (i32, i32em only) Support Microsoft* style assembly lan­ guage insertion using MASM style syntax and, if requested, output assembly in MASM format. -use-pch Manual use of precompiled header (file­ name.pchi). This option directs the compiler to use the PCH file specified by filename. It cannot be used in the same compilation as -create-pch . The -use-pch option supports full path names and supports multiple source files when all source files use the same .pchi file. Preprocessor Options The preprocessor options listed here control pre­ processing operations for the compiler. -A- Remove all predefined macros. Causes all predefined macros and assertions to be inac­ tive. -A[] Associate a symbol with the specified sequence of . Equivalent to an #assert preprocessing directive. -C Preserve comments in preprocessed source output. Comments following preprocessing directives are not preserved. -D[=] Define the macro and associate it with the specified . Equivalent to a #define preprocessing directive. DEFAULT: -D defines the macro with a of 1. -dM Output macro definitions in effect after preprocessing (use with -E). -E Direct the preprocessor to expand your source module and write the result to stan­ dard output. The preprocessed source con­ tains #line directives, which the compiler uses to determine the source file and line number. -EP Direct the preprocessor to expand your source module and write the result to stan­ dard output. Does not include #line direc­ tives in the output. -EP is equivalent to -E -P. -H Print "include" file order and continue com­ pilation. -I Add directory to include file search path. For multiple search directories, mul­ tiple I commands must be used. The compiler searches directories for include files in the following order: 1. Directory of the source file that contains the include. 2. Directories specified by the -I option. -idirafter Add directory to the second include file search path (after -I). -imacros Treat as an #include file, but throw away all preprocessing while the macros that are defined remain defined. -iprefix Use with -iwithprefix as a prefix. -isystem Add directory to the start of the sys­ tem include path. -iwithprefix Append to the prefix passed in by -iprefix and put it on the include search path at the end of the include directories. -iwithprefixbefore Similar to -iwithprefix except the include directory is placed in the same place as the -I command line include directories. -M Generate makefile dependency lines for each source file, based on the #include lines found in the source file. -MD Preprocess and compile. Generate output file (.d extension) containing dependency infor­ mation. -MF Generate makefile dependency information in . Must specify -M or -MM. -MG Similar to -M, but treats missing header files as generated files. -MM Similar to -M, but does not include system header files. -MMD Similar to -MD, but does not include system header files. -MP Add a phony target for each dependency. -MQ Same as -MT, but quotes special Make charac­ ters. -MT Change the default target rule for depen­ dency generation. -no-gcc Do not predefine the __GNUC__, __GNUC_MINOR__, and __GNUC_PATCHLEVEL__ macros. -nostdinc Same as X. -P Direct the preprocessor to expand your source module and store the result in a .i file in the current directory. Unlike the -E option, the output from -P does not include #line number directives. By default, the preprocessor creates the name of the output file using the prefix of the source file name with a .i extension. Caution: When you use the -P option, any existing files with the same name and exten­ sion are overwritten. -ssp (i32 only) Enable software-based speculative pre-compu­ tation. -U Remove predefined macro. Equivalent to a #undef preprocessing directive. -X Remove standard directories from include file search path. You can use the -X option with the -I option to prevent the compiler from searching the default path for include files and direct it to use an alternate path. Component Control Options You can direct the compiler to specify alternate tools for preprocessing, compilation, assembly, and linking. Further, you can invoke options specific to your alternate tools on the command line. These are the component control options. -Qinstall Set as the root of the compiler installation. -Qlocation,, Set as the location of the tool spec­ ified by . may be cpp, c, asm, or ld. is the complete path to the tool. For example: -Qlocation,gas, Specifies the GNU assembler. -Qlocation,gld, Specifies the GNU linker. -Qoption,, Pass options to the tool specified by . may be cpp, c, asm, or ld. The argument indicates one or more valid argument strings for the designated program. If the argument is a command-line option, you must include the hyphen. If the argument contains a space or tab character, you must enclose the entire argument in quo­ tation characters (""). You must separate multiple arguments with commas. Language Options The language options define parameters and stan­ dards for the compiler with reference to the C/C++ source files. -[no]align (i32, i32em only) Analyze and reorder memory layout for vari­ ables and arrays. -ansi Support all ANSI standard C programs, equiv­ alent to the -ansi option of gcc. -[no-]c99 Enable [disable] C99 support for C programs. -export Enable recognition of exported templates. Supported in C++ mode only. -export-dir Specifies a directory name for the exported template search path. -fno-gnu-keywords Do not recognize 'typeof' as a keyword. -fno-implicit-inline-templates Do not emit code for implicit instantiations of inline templates. -fno-implicit-templates Never emit code for non-inline templates which are instantiated implicitly (i.e. by use); only emit code for explicit instantia­ tions. -fno-operator-names Disable support for operator name keywords. -fpermissive Allow for non-conformant code. -f[no-]rtti Enable [disable] RTTI support. -fshort-enums Allocate as many bytes as needed for enumer­ ated types. -fsyntax-only Same as -syntax. -ftemplate-depth- Control the depth to which recursive tem­ plates are expanded. -funsigned-bitfields Change default bitfield type to unsigned. -funsigned-char Change default char type to unsigned. -Kc++ Compile all source or unrecognized file types as C++ source files. -malign-double Same as -align. -[no]restrict Enable [disable] the "restrict" keyword for disambiguating pointers. -std=c99 Enable C99 support for C programs -strict-ansi Select strict ANSI C/C++ conformance dialect -Zp[n] Specify alignment constraint for structure and union types, where n is one of the fol­ lowing: 1,2,4,8,16. Compiler Diagnostics Options The compiler diagnostics options define and control aspects of the compilation associated with diagnos­ tics. -[no]traceback Tells the compiler to generate [not gener­ ate] extra information in the object file to allow the display of source file traceback information at run time when a severe error occurs -w Disable all warnings. Displays error mes­ sages only. -w Control diagnostics, where is one of the following: 0 -- Display errors (same as -w) 1 -- Display warnings and errors (DEFAULT) 2 -- Display remarks, warnings, and errors -W[no-]abi Warn if generated code is not C++ ABI com­ pliant (DEFAULT). -Wall Enable all warnings. -Wbrief Print brief one-line diagnostics. When enabled, the original source line is not displayed and the error message text is not wrapped when too long to fit on a single line. -Wcheck Enable more strict diagnostics. Performs compile-time code checking for code that exhibits non-portable behavior, represents a possible unintended code sequence, or possi­ bly affects operation of the program because of a quiet change in the ANSI C Standard. -W[no-]comment[s] Warn when /* appears in the middle of a /* */ comment. -Wcontext-limit= Set maximum number of template instantiation contexts shown in diagnostic. -wd[,,...] Disable diagnostics L1 through LN. -W[no-]deprecated Print warnings related to deprecated fea­ tures. -we[,,...] Change severity of diagnostics L1 through LN to error. -Werror Force warnings to be reported as errors. -Winline Enable inline diagnostics. -W[no-]main Warn if return type of main is not expected. -W[no-]missing-prototypes Warn for missing prototypes. -wn Print a maximum of errors displayed before the compiler aborts. By default, if more than 100 errors are displayed, compila­ tion aborts. Remarks and warnings do not count towards this limit. -Wp64 (i64, i32em only) Print diagnostics for 64-bit porting. -W[no-]pointer-arith Warn for questionable pointer arithmetic. -wr[,,...] Change severity of diagnostics L1 through LN to remark. -W[no-]return-type Warn when a function uses the default int return type and warn when a return statement is used in a void function. -W[no-]shadow Warn if a variable declaration hides a pre­ vious variable declaration. -W[no-]uninitialized Warn if a variable is used before being ini­ tialized. -W[no-]unknown-pragmas Warn if an unknown #pragma directive is used (DEFAULT). -W[no-]unused-function Warn if declared function is not used. -ww[,,...] Change severity of diagnostics L1 through LN to warning. Miscellaneous Options The miscellaneous options cover a variety of unre­ lated tasks related to the compiler. -B Specifies where to find libraries, headers, and executables for the compiler itself. -dryrun Show driver tool commands but do not execute tools. -f[no-]common Enables the compiler to treat common vari­ ables as if they were defined, allowing the use of gprel addressing of common data vari­ ables. -f[no-]math-errno Set ERRNO after calling standard math library functions. -fminshared Compilation is for the main executable. Absolute addressing can be used and non-position independent code generated for symbols that are at least protected. -fno-builtin Disable inline expansion of intrinsic func­ tions. -fno-builtin- Disable the intrinsic. -fpack-struct Pack structure members together. -f[no-]pic, -f[no-]PIC i32: This option generates position indepen­ dent code. By default this option is OFF. i64: This option generates code allowing full symbol preemption. By default this option is OFF. -fr32 (i64 only) Disable use of high floating point regis­ ters. Use only lower 32 floating-point reg­ isters. -freg-struct-return Return struct and union values in registers when possible. -fstack-security-check (i32, i32em only) Enable overflow security checks. -fvisibility=[extern|default|protected|hid­ den|internal] Global symbols (data and functions) will get the visibility attribute given by default. Symbol visibility attributes explicitly set in the source code or using the symbol visi­ bility attribute file options will override the -fvisibility setting. -fvisibility-default= Space separated symbols listed in the argument will get visibility set to default. -fvisibility-extern= Space separated symbols listed in the argument will get visibility set to extern. -fvisibility-hidden= Space separated symbols listed in the argument will get visibility set to hidden. -fvisibility-internal= Space separated symbols listed in the argument will get visibility set to inter­ nal. -fvisibility-protected= Space separated symbols listed in the argument will get visibility set to pro­ tected. -gcc-name= Use this option to specify the location of g++ when compiler cannot locate gcc C++ libraries. For use with -cxxlib- con­ figuration. Use this option when referencing a non-standard gcc installation. -gcc-version= This option provides compatible behavior with gcc, where indicates the gcc ver­ sion. The required values for are: not set -- if gcc version is older than 3.2 320 -- if gcc version is 3.2 330 -- if gcc version is 3.3 340 -- if gcc version is 3.4 -[no-]global-hoist Enables [disables] hoisting and speculative loads of global variables. -help Print list of compiler options. -kernel (i64 only) Generates code for inclusion in the kernel. Prevents generation of speculation as sup­ port may not be available when code runs. Suppresses software pipelining. -long_double Enable 80-bit 'long double'. -mfixed-range=f12-f15,f32-f127 (i64 only) Reserves certain registers (f12-f15, f32-f127) for use by the kernel. -[no-]multibyte-chars Provide support for multi-byte characters. -nobss-init Place variables that are initialized with zeroes in the DATA section instead of the BSS section. -reserve-kernel-regs (i64 only) Reserves registers f12-f15 and f32-f127 for use by the kernel. These will not be used by the compiler. -[no-]sox Enable (DEFAULT) [disable] saving of com­ piler options and version in the executable. -V Display compiler version information. -v Show driver tool commands and execute tools. --version Display GCC style version information. Note that two '-'s are required. -x All source files found subsequent to -x will be recognized as one of the fol­ lowing types: c -- C source file c++ -- C++ source file c-header -- C header file cpp-output -- C pre-processed file c++-cpp-output -- C++ pre-processed file assembler -- Assembly file assembler-with-cpp -- Assembly file that needs to be preprocessed none -- Disable recognition, and revert to file extension Linking or Linker Options The linking/linker options group all link-related options together. -cxxlib- Tells the compiler to link using certain C++ runtime libraries. You can specify one of the following values for : gcc[=] -- Link using the C++ run-time libraries provided with the gcc compiler. The parameter is an optional top-level location for the gcc binaries and libraries. icc -- Link using the C++ run-time libraries provided by Intel. -dynamic-linker Select a dynamic linker (filename) other than the default. -i-dynamic Link Intel provided libraries dynamically. -i-static -L Instruct the linker to search for libraries. -mno-relax (i64 only) Do not pass -relax to the linker. -mrelax (i64 only) Pass -relax to the linker. -no-cpprt Do not link in C++ runtime libraries. -nodefaultlibs Do not use standard libraries when linking. -nostartfiles Do not use standard startup files when link­ ing. -nostdlib Do not use standard libraries and startup files when linking. -shared Produce a shared object. -shared-libcxa Link Intel libcxa C++ library dynamically, overriding the default behavior when -static is used. This option has the opposite effect of -static-libcxa. When this option is used, the Intel-provided libcxa C++ library is linked in dynamically, allowing the user to override the static linking behavior when the -static option is used. -static Prevent linking with shared libraries. Causes the executable to link all libraries statically, as opposed to dynamically. -static-libcxa Link Intel libcxa C++ library statically. By default, the Intel-provided libcxa C++ library is linked in dynamically. Use -static-libcxa on the command line to link libcxa statically, while still allowing the standard libraries to be linked in by the default behavior. -T Direct linker to read link commands from . -u Pretend the is undefined. -Wa[,,...] -Wl,[,,...] -Wl,[,,...] Pass options o1, o2, etc. to the linker for processing. -Xlinker Pass directly to the linker for pro­ cessing. Optimization Levels The Intel C++ Compiler applies the following opti­ mizations when you invoke the -O1, -O2, or -O3 options: ⋠Constant propagation ⋠Copy propagation ⋠Dead-code elimination ⋠Global register allocation ⋠Instruction scheduling ⋠Loop unrolling (-O2, -O3 only) ⋠Loop-invariant code movement ⋠Partial redundancy elimination ⋠Strength reduction/induction variable simplifi­ cation ⋠Variable renaming ⋠Exception handling optimizations ⋠Tail recursions ⋠Peephole optimizations ⋠Structure assignment lowering and optimizations ⋠Dead store elimination ⋠Loop-invariant code motion. Depending on the Intel architecture, optimization options can have different effects. To specify optimizations for your target architecture, refer to the -O[0|1|2|3], -fast, -ax, -x, and related options. Optimizing Exclusively for Specific Processors (i32, i32em only) The -x{K|W|N|B|P} options target your program to run on a specific Intel processor. The resulting code might contain unconditional use of features that are not supported on other processors. If these options are used on a non-compatible proces­ sor, the program might fail with an illegal instruction exception, or it might display other unexpected behavior. Do not specify this option if the program will be executed on x86 processors not provided by Intel Corporation. For more infor­ mation, see the Intel C++ Compiler User's Guide. Automatic Processor-specific Optimization (i32, i32em Only) The -ax{K|W|N|B|P} options direct the compiler to find opportunities to generate separate versions of functions that use instructions supported on the specified processors. If the compiler finds such an opportunity, it first checks whether generating a processor-specific version of a function is likely to result in a performance gain. If so, the com­ piler generates both a processor-specific version of a function and a generic version of the func­ tion. The generic version will run on any IA-32 processor. At run time, one of the versions is chosen to exe­ cute, depending on the current processor. In this way, the program can benefit from performance gains on more advanced processors, while still working properly on older processors. The disadvantages of using -ax{K|W|N|B|P} are: ⋠The size of the compiled binary increases because it contains both a processor-specific version of some of the code and a generic ver­ sion of the code. ⋠Performance is affected by the run-time checks to determine which code to run. Interprocedural Optimizations (IPO) Use -ip and -ipo[value] to enable interprocedural optimizations (IPO), which allow the compiler to analyze your code to determine where to apply the following optimizations: inline function expansion, interprocedural constant propagation, monitoring module-level static variables, dead code elimina­ tion, propagation of function characteristics, and multifile optimization. For IA-32, IPO also passes arguments in registers. Inline function expansion is one of the main opti­ mizations performed by the interprocedural opti­ mizer. For function calls that the compiler believes are frequently executed, the compiler might decide to replace the instructions of the call with code for the function itself (inline the call). See the Intel C++ Compiler User's Guide for more complete information on IPO. Profile-guided Optimizations (PGO) Profile-guided optimizations (PGO) tell the com­ piler which areas of an application are most fre­ quently executed. By knowing these areas, the com­ piler is able to use feedback from a previous com­ pilation to be more selective in optimizing the application. For example, the use of PGO often enables the compiler to make better decisions about function inlining, thereby increasing the effec­ tiveness of interprocedural optimizations. PGO creates an instrumented program from your source code and special code from the compiler. Each time this instrumented code is executed, the instrumented program generates a dynamic informa­ tion file. When you compile a second time, the dynamic information files are merged into a summary file. Using the profile information in this file, the compiler attempts to optimize the execution of the most heavily traveled paths in the program. The PGO methodology requires three phases: Phase 1: Instrumentation compilation and linking with -prof-gen[x] Phase 2: Instrumented execution by running the executable Phase 3: Feedback compilation with -prof-use See the Intel C++ Compiler User's Guide for more complete information on PGO. High-level Language Optimizations (HLO) High-level language optimizations (HLO) exploit the properties of source code constructs, such as loops and arrays, in applications developed in high-level programming languages, such as C++. They include loop interchange, loop fusion, loop unrolling, loop distribution, unroll-and-jam, blocking, data prefetch, scalar replacement, data layout optimiza­ tions, and others. Vectorization Options (i32 only) The vectorizer is a component of the Intel C++ Com­ piler that automatically uses SIMD instructions in the MMX, SSE, and SSE2 instruction sets. The vec­ torizer detects operations in the program that can be executed in parallel, and then converts the sequential program to process 2, 4, 8, or 16 ele­ ments in one operation, depending on the data type. The -x{K|W|N|B|P} and -ax{K|W|N|B|P} options enable the vectorizer. See the Intel C++ Compiler User's Guide for more complete information on the vector­ izer. Auto Parallelization Options The auto-parallelization feature of the Intel C++ Compiler automatically translates serial portions of the input program into equivalent multithreaded code. The auto-parallelizer analyzes the dataflow of the program's loops and generates multithreaded code for those loops which can be safely and effi­ ciently executed in parallel. This enables the potential exploitation of the parallel architecture found in symmetric multiprocessor (SMP) systems. The parallel run-time support provides the same run-time features found in OpenMP*, such as han­ dling the details of loop iteration modification, thread scheduling, and synchronization. While OpenMP directives enable serial applications to transform into parallel applications quickly, the programmer must explicitly identify specific portions of the application code that contain par­ allelism and add the appropriate compiler direc­ tives. Auto-parallelization triggered by the -par­ allel option automatically identifies those loop structures that contain parallelism. During compi­ lation, the compiler automatically attempts to decompose the code sequences into separate threads for parallel processing. No other effort by the programmer is needed. Parallelization with OpenMP* The Intel C++ Compiler supports the OpenMP* C++ version 2.0 API specification. The compiler per­ forms transformations to generate multithreaded code based on the user's placement of OpenMP direc­ tives in the source program making it easy to add threading to existing software. The Intel compiler supports all of the current industry-standard OpenMP directives, except WORKSHARE, and compiles parallel programs annotated with OpenMP directives. In addition, the compiler provides Intel-specific extensions to the OpenMP C++ version 2.0 specifica­ tion including run-time library routines and envi­ ronment variables. Note: As with many advanced features of compilers, you must properly understand the functionality of the OpenMP directives in order to use them effec­ tively and avoid unwanted program behavior. Precompiled Headers The Intel C++ Compiler supports precompiled header (PCH) files to significantly reduce compile times using the options described below. If many of your source files include a common set of header files, place the common headers first, followed by the #pragma hdrstop directive. This pragma instructs the compiler to stop generating PCH files. For example, if source1.cpp, source2.cpp, and source3.cpp all include common.h, then place #pragma hdrstop after common.h to optimize compile times. See the Intel C++ Compiler User's Guide for more complete information on Precompiled Headers. Caution: Depending on how you organize the header files listed in your sources, these options might increase compile time. PREDEFINED MACROS The predefined macros available for the Intel C++ Compiler are described below. Intel C++ Compiler Predefined Macros __ECC Value on IA-32 -- NA Value on Itanium Architecture -- 800 Notes -- Assigned value refers to the com­ piler (e.g., 800 is 8.00). Supported for legacy reasons. Use __INTEL_COMPILER instead. __EDG__ Value on IA-32 -- 1 Value on Itanium Architecture -- 1 __EDG_VERSION__ Value on IA-32 -- 302 Value on Itanium Architecture -- 302 __ELF__ Value on IA-32 -- 1 Value on Itanium Architecture -- 1 __EXCEPTIONS Value on IA-32 -- Defined when -fno-excep­ tions is not used. Value on Itanium Architecture -- NA __extension__ Value on IA-32 -- no value Value on Itanium Architecture -- no value __GNUC__ Values on IA-32 or Itanium Architecture -- 2 -- if gcc version is less than 3.2 3 -- if gcc version is 3.2, 3.3, or 3.4 __gnu_linux__ Value on IA-32 -- 1 Value on Itanium Architecture -- 1 __GNUC_MINOR__ Values on IA-32 or Itanium Architecture -- 95 -- if gcc version is less than 3.2 2 -- if gcc version is 3.2 3 -- if gcc version is 3.3 4 -- if gcc version is 3.4 __GNUC_PATCHLEVEL__ Value on IA-32 -- 0 Value on Itanium Architecture -- 0 __GXX_ABI_VERSION Value on IA-32 -- 102 Value on Itanium Architecture -- 102 __HONOR_STD Value on IA-32 -- 1 __i386 Value on IA-32 -- 1 Value on Itanium Architecture -- NA __i386__ Value on IA-32 -- 1 Value on Itanium Architecture -- NA i386 Value on IA-32 -- 1 Value on Itanium Architecture -- NA __ia64 Value on IA-32 -- NA Value on Itanium Architecture -- 1 __ia64__ Value on IA-32 -- NA Value on Itanium Architecture -- 1 ia64 Value on IA-32 -- NA Value on Itanium Architecture -- 1 __ICC Value on IA-32 -- 800 Value on Itanium Architecture -- NA Notes -- Assigned value refers to the com­ piler (e.g., 800 is 8.00). Supported for legacy reasons. Use __INTEL_COMPILER instead. __INTEL_COMPILER Value on IA-32 -- 800 Value on Itanium Architecture -- 800 Notes -- Defines the compiler version. Defined as 800 for the Intel C++ Compiler 8.0. __INTEL_COMPILER_BUILD_DATE= Notes -- Defines the compiler build date. This date should correspond to the date on the compiler version banner. Applies to both IA-32 and Itanium Architecture. __INTEL_RTTI__ Notes -- Is defined (default) when the -frtti option is specified for the Intel Linux C++ compiler (i.e., when RTTI support is enabled). It is disabled when the the -fno-rtti option is specified for the Intel Linux C++ compiler. Applies to both IA-32 and Itanium Architecture. __INTEL_STRICT_ANSI__ Notes -- Enabled when the -strict_ansi option is specified for the Intel Linux C++ compiler (i.e., strict ANSI conformance dialect). Applies to both IA-32 and Itanium Architecture. _INTEGRAL_MAX_BITS Value on IA-32 -- NA Value on Itanium Architecture -- 64 Notes -- Indicates support for the __int64 type. __itanium__ Value on Itanium Architecture -- 1 __linux Value on IA-32 -- 1 Value on Itanium Architecture -- 1 __linux__ Value on IA-32 -- 1 Value on Itanium Architecture -- 1 linux Value on IA-32 -- 1 Value on Itanium Architecture -- 1 __LONG_DOUBLE_SIZE__ Value on IA-32 -- 80 Value on Itanium Architecture -- NA __LONG_MAX__ Value on IA-32 -- NA Value on Itanium Architecture -- 9223372036854775807L __lp64 Value on IA-32 -- NA Value on Itanium Architecture -- 1 __LP64__ Value on IA-32 -- NA Value on Itanium Architecture -- 1 __LP64 Value on IA-32 -- NA Value on Itanium Architecture -- 1 _OPENMP Value on IA-32 -- 200203 Value on Itanium Architecture -- 200203 Notes -- Defined when -openmp is used. __OPTIMIZE__ Value on IA-32 -- 1 Value on Itanium Architecture -- 1 Notes -- Not enabled if all optimizations are turned off. _PGO_INSTRUMENT Value on IA-32 -- 1 Value on Itanium Architecture -- 1 Notes -- Defined when compiled with either -prof_gen or -prof_genx. __PTRDIFF_TYPE__ Value on IA-32 -- int Value on Itanium Architecture -- long __QMSPP_ Value on IA-32 -- 1 Value on Itanium Architecture -- NA __REGISTER_PREFIX__ Value on IA-32 -- no value Value on Itanium Architecture -- no value __SIGNED_CHARS__ Value on IA-32 -- 1 Value on Itanium Architecture -- 1 __SIZE_TYPE__ Value on IA-32 -- unsigned Value on Itanium Architecture -- unsigned long __STDC_HOSTED__ Value on IA-32 -- 1 Value on Itanium Architecture -- 1 __unix Value on IA-32 -- 1 Value on Itanium Architecture -- 1 __unix__ Value on IA-32 -- 1 Value on Itanium Architecture -- 1 unix Value on IA-32 -- 1 Value on Itanium Architecture -- 1 __USER_LABEL_PREFIX__ Value on IA-32 -- (no value) Value on Itanium Architecture -- (no value) __WCHAR_TYPE__ Value on IA-32 -- long int Value on Itanium Architecture -- int __WINT_TYPE__ Value on IA-32 -- unsigned int Value on Itanium Architecture -- unsigned int ANSI/ISO Macros Included with the Compiler The ANSI/ISO standard for C language requires that certain predefined macros be supplied with conform­ ing compilers. The Intel C++ Compiler supplies the following macros in accordance with this standard, as well as additional predefined macros: __cplusplus The name __cplusplus is defined when compil­ ing a C++ translation unit. __DATE__ The date of compilation as a string literal in the form Mmm dd yyyy. __FILE__ A string literal representing the name of the file being compiled. __LINE__ The current line number as a decimal con­ stant. __STDC__ The name __STDC__ is defined when compiling a C translation unit. __TIME__ The time of compilation. As a string literal in the form hh:mm:ss. gcc Predefined Macros This version of the Intel C++ Compiler includes the following new predefined macros also supported by gcc: __GNUC__ __GNUC_MINOR__ __GNUC_PATCHLEVEL__ You can specify the -no-gcc option if you do not want these macros defined. If you need gcc interoperability (-cxxlib-gcc), do not use the -no-gcc compiler option. Suppress Macro Definition Use the -Uname option to suppress any macro defini­ tion currently in effect for the specified name. The -U option performs the same function as an #undef preprocessor directive. ENVIRONMENT VARIABLES You can customize your environment by setting the following environment variables. You can specify paths where the compiler can search for special files such as libraries and include files. LD_LIBRARY_PATH Specifies the location for all Intel-pro­ vided libraries. The LD_LIBRARY_PATH envi­ ronment variable contains a colon-separated list of directories in which the linker will search for library (.a) files. If you want the linker to search additional libraries, you can add their names to LD_LIBRARY_PATH, to the command line, to a response file, or to the configuration file. In each case, the names of these libraries are passed to the linker before the names of the Intel libraries that the driver always specifies. PATH Specifies the directories the system searches for binary executable files. ICCCFG Specifies the configuration file for cus­ tomizing compilations with the icc compiler. ICPCCFG Specifies the configuration file for cus­ tomizing compilations with the icpc com­ piler. TMP Specifies the directory in which to store temporary files. If the directory specified by TMP does not exist, the compiler places the temporary files in the current direc­ tory. IA32ROOT (i32 only) Points to the directory containing the bin, lib, include and substitute header directo­ ries. IA64ROOT (i64 only) Points to the directory containing the bin, lib, include and substitute header directo­ ries. GNU* Environment Variables The Intel C++ Compiler supports the following GNU environment variables: CPATH Specifies a list of directories to search following the directories specified by -I. C_INCLUDE Specifies a list of directories to search following the directories specified by -isystem. CPLUS_INCLUDE_PATH Same as C_INCLUDE. DEPENDENCIES_OUTPUT Specifies how to output dependencies for Make based on preprocessed, non-system header files. SUNPRO_DEPENDENCIES Same as DEPENDENCIES_OUTPUT, except that system header files are not ignored. Compilation Environment Options The Intel C++ Compiler installation includes shell scripts that you can use to set environment vari­ ables. See the Intel C++ User's Guide for more information Standard OpenMP Environment Variables OMP_SCHEDULE Specifies the type of runtime scheduling. DEFAULT: static OMP_NUM_THREADS Sets the number of threads to use during execution. DEFAULT: Number of processors currently installed in the system while generating the executable OMP_DYNAMIC Enables (TRUE) or disables (FALSE) the dynamic adjustment of the number of threads. DEFAULT: FALSE OMP_NESTED Enables (TRUE) or disables (FALSE) nested parallelism. DEFAULT: FALSE Intel Extensions to OpenMP Environment Variables KMP_LIBRARY Selects the OpenMP run-time library through­ put. The options for the variable value are: serial, turnaround, or throughput indicating the execution mode. The default value of throughput is used if this variable is not specified. DEFAULT: throughput (execution mode) KMP_STACKSIZE Sets the number of bytes to allocate for each parallel thread to use as its private stack. Use the optional suffix b, k, m, g, or t, to specify bytes, kilobytes, megabytes, gigabytes, or terabytes. DEFAULT: i32: 2m i64: 4m PGO Environment Variables The following environment values determine the directory in which to store dynamic information files or whether to overwrite pgopti.dpi. PROF_DIR Specifies the directory in which dynamic information files are created. This variable applies to all three phases of the profiling process. PROF_NO_CLOBBER Alters the feedback compilation phase slightly. By default, during the feedback compilation phase, the compiler merges the data from all dynamic information files and creates a new pgopti.dpi file if .dyn files are newer than an existing pgopti.dpi file. When this variable is set, the compiler does not overwrite the existing pgopti.dpi file. Instead, the compiler issues a warning and you must remove the pgopti.dpi file if you want to use additional dynamic information files. PROF_DUMP_INTERVAL Initiate Interval Profile Dumping in an instrumented application. The _PGOPTI_Set_Interval_Prof_Dump(int interval) function activates Interval Profile Dumping and sets the approximate frequency at which dumps will occur. The interval parameter is measured in milliseconds and specifies the time interval at which profile dumping will occur. An alternative method of initiating Interval Profile Dumping is by setting this environment variable. Set this environment variable to the desired interval value prior to starting the application. EXAMPLES The following examples demonstrate optimizing across multiple input files: 1. icc ax.cpp This command compiles the C++ file 'ax.cpp' producing executable file 'a.out'. Optimizations occur by default. 2. icc -o abc ax.cpp bx.cpp cx.cpp This command uses option -o to name the executable file 'abc' and compiles 'ax.cpp', 'bx.cpp', and 'cx.cpp' as one program. 3. icc -c ax.cpp bx.cpp cx.cpp This command uses option -c to suppress linking and produce individual object files 'ax.o', 'bx.o', and 'cx.o'. Interproce­ dural optimizations are prevented. 4. icc -c ay.c This command compiles the C file 'ay.c' and the -c option suppresses linking to produce of an object file 'ay.o'. 5. icpc -c by.cpp This command compiles the C++ file 'by.cpp' and the -c option suppresses linking to produce an object file 'by.o'. 6. icpc ay.o by.o This command links the two object files 'ay.o' and 'by.o' and produces an executable file 'a.out'. 7. icc -c -O1 sub2.cpp icc -c -O1 sub3.cpp icc -o main.exe -g -O0 main.cpp sub2.o sub3.o The first two commands show incremental compilation with minimal optimization. The first command generates an object file of 'sub2'; the second generates an object file of 'sub3'. The last command uses option -O0 to disable all compiler default optimizations. It uses option -g to generate symbolic debugging information and line numbers in the object code, which can be used by a source-level debugger. TECHNICAL SUPPORT The Intel C++ Compiler product web site offers timely and comprehensive product information, including product features, white papers, and tech­ nical articles. For the latest information, visit http://developer.intel.com/software/products/. Intel also provides a support web site that con­ tains a rich repository of self help information, including getting started tips, known product issues, product errata, license information, user forums, and more. Registering your product entitles you to one year of technical support and product updates through Intel Premier Support. Intel Premier Support is an interactive issue management and communication web site that enables you to submit issues and review their status, and to download product updates any­ time of the day. To register your product, contact Intel, or seek product support, please visit: http://www.intel.com/software/products/support. SEE ALSO The Intel C++ Compiler User's Guide is the defini­ tive source for detailed information on using the Intel C++ Compiler, including a complete listing of all compiler options, and much more. In addition, see these other documents: ⋠Product Release Notes ⋠Intel Itanium Assembler User's Guide ⋠Intel Itanium Architecture Assembly Language Reference Guide ⋠Enhancing Performance with the Intel Compiler (training) You can access these documents from /doc/doc_index.htm or from http://developer.intel.com/software/products/com­ pilers/ . COPYRIGHT INFORMATION Copyright (C) 2002 - 2005, Intel Corporation. All rights reserved. * Other brands and names are the property of their respective owners. Copyright(C) 2002 - 2005Intel Corporation ICC(1) Intel(R) Fortran Compiler 9.0 for Linux --------------------------------------------- Intel(R) Fortran compiler options. Used for SGI Altix 350, 3000, 3700 Bx2, and 4700 submissions. ifort(1) ifort(1) NAME ifort - invokes the Intel(R) Fortran Compiler SYNOPSIS ifort [ options ] file1 [ file2 ]... options Are zero or more compiler options. fileN Is a Fortran source file, assembly file, object file, object library, or other linkable file. DESCRIPTION The ifort command invokes the Intel(R) Fortran Compiler that is designed to preprocess, compile, assemble, and link Fortran programs on Intel(R) IA-32-based systems, Intel IA-32-based systems with Intel(R) Extended Memory 64 Technology (Intel(R) EM64T), or Intel(R) Itanium(R)-based systems. For more information on this compiler, see the Intel(R) Fortran Compiler for Linux* Systems User's Guide. The ifort command interprets input files by their filename suffix as follows: ⋠Filenames with the suffix .f90 are interpreted as free-form Fortran 95/90 source files. ⋠Filenames with the suffix .f, .for, or .ftn are interpreted as fixed-form Fortran source files. ⋠Filenames with the suffix .fpp, .F, .FOR, .FTN, or .FPP are interpreted as fixed-form Fortran source files, which must be preprocessed by the fpp pre­ processor before being compiled. ⋠Filenames with the suffix .F90 are interpreted as free-form Fortran source files, which must be pre­ processed by the fpp preprocessor before being com­ piled. ⋠Filenames with the suffix .s are interpreted as assembler files and are passed to the assembler. ⋠Filenames with the suffix .S are interpreted as assembler files and are preprocessed before being passed to the assembler. ⋠Filenames with the suffix .a are interpreted as object libraries and are passed to ld(1). ⋠Filenames with the suffix .o are interpreted as compiled object files and are passed to ld(1). You can override some options specified on the command line by using the OPTIONS statement in your Fortran source program. An OPTIONS statement affects only the program unit in which the statement occurs. For more information, see the Intel(R) Fortran Language Reference. Some language features (options, predefined symbols, etc.) are only available on certain systems. Such language fea­ tures are labeled, as follows: i32 Means the feature is available on IA-32-based sys­ tems i32em Means the feature is available on IA-32-based sys­ tems with Intel(R) Extended Memory 64 Technology (Intel(R) EM64T) i64 Means the feature is available on Itanium(R)-based systems If no label appears, the language feature is available on all supported systems. Performance Enhancing Options The following command line options can be used to increase the run-time performance of code generated by the Intel Fortran compiler: ⋠On IA-32-based systems (including those with Intel(R) EM64T): -ax

, -ftz, -ip, -ipo, -O[n], -openmp, -par­ allel, -prof-use, -tpp, -x

. ⋠On Itanium-based systems: -fnsplit, -ftz, -ip, -ipo, -O[n], -openmp, -parallel, -prof-use, -tpp. Configuration and Indirect Files Command options to be used whenever the compiler is invoked can be put into a system configuration file named ifort.cfg, which resides in the same area as the compiler. The text in this file is processed by ifort before the text on the command line. To use a personal configuration file, set the environment variable IFORTCFG to point to the path and filename to be used. An indirect file contains text that can be included on the ifort command line. Precede the filename with an at sym­ bol (@) on the command line at the point where the options are to be inserted. For example, assume file double_size contains options "-i8 -r8" and file my_includes contains options "-I/bld/inc -I/bld/headers". In this case, the following command line: ifort -O3 @double_size myprog.f90 @my_includes passes "-O3 -i8 -r8 myprog.f90 -I/bld/inc -I/bld/headers" to the compiler. OPTIONS Some compiler options have the form -name keyword. -name must be spelled out completely, but keyword can be abbre­ viated to its shortest unique prefix (4 characters are recommended). For example, -assume buffered_io can be specified as -assume buff. For information on linker and load-time options, see ld(1). For some options, you can (or must) specify additional information, such as a keyword, a directory, a file name, a number, and so forth. When this information is required, it is shown in angle brackets (<>); when it is optional, it is shown in square brackets ([]). For exam­ ple, in option -align , keyword is required; in option -unroll[n], n (a number) is optional. The ifort command takes the following options: -1 Tells the compiler to execute at least one itera­ tion of DO loops (same as the -onetrip option). This option has the same effect as -f66. -66 Enforces FORTRAN 66 semantics (same as the -f66 option). -72 Specifies that the statement field of each fixed- form source line ends at column 72. This is the default. -80 Specifies that the statement field of each fixed- form source line ends at column 80. -132 Specifies that the statement field of each fixed- form source line ends at column 132 (same as the -extend_source option). -align [keyword] Tells the compiler how to align certain data items. The following are -align options: ⋠-align all Tells the compiler to add padding bytes whenever possible to obtain the natural alignment of data items in common blocks, derived types, and record structures. Specifies -align nocommons, -align dcommons, -align records, -align nosequence. This is the same as specifying -align with no keyword. ⋠-align commons Aligns all common block entities on natural boundaries up to 4 bytes, by adding padding bytes as needed. The default is -align nocommons, which adds no padding to common blocks. ⋠-align dcommons Aligns all common block entities on natural boundaries up to 8 bytes, by adding padding bytes as needed. If you specify any -std option or the -stand f90 or -stand f95 option, this option is ignored. The default is -align nodcommons, which adds no padding to common blocks. ⋠-align none Tells the compiler not to add padding bytes any­ where in common blocks or structures. This is the same as specifying -noalign. The default is to add no padding to common blocks but to add padding to structures. ⋠-align norecords Aligns components of derived types and fields within record structures on arbitrary byte bound­ aries with no padding. The default is -align records, which is the same as specifying -align rec16byte. ⋠-align recbyte Aligns fields of records and components of derived types on the smaller of the size boundary specified or the boundary that will naturally align them. can be 1, 2, 4, 8, or 16. The default is -align rec16byte, which is the same as specifying -align records. ⋠-align sequence Aligns components of a derived type declared with the SEQUENCE statement (sequenced components) according to the alignment rules that are cur­ rently in use. The default alignment rules are to align unsequenced components on natural bound­ aries. The default is -align nosequence, which requests that sequenced components be packed regardless of any other alignment rules. If you specify any -std option or the -stand f90 or -stand f95 option, this option is ignored. -no-ansi-alias -no-ansi_alias Tells the compiler to assume the program does not adhere to the Fortran 95 Standard type aliasability rules. The default is -ansi-alias, which tells the compiler to assume that the program adheres to these aliasability rules. -arch (i32 and i32em) Determines the version of the architecture for which the compiler generates instructions. The fol­ lowing are -arch options: ⋠-arch pn1 Optimizes for the Intel(R) Pentium(R) processor. ⋠-arch pn2 Optimizes for the Intel(R) Pentium(R) Pro, Intel(R) Pentium(R) II, and Intel(R) Pentium(R) III processors. ⋠-arch pn3 Optimizes for the Intel(R) Pentium(R) Pro, Intel(R) Pentium(R) II, and Intel(R) Pentium(R) III processors. This is the same as specifying the -arch pn2 option. ⋠-arch pn4 Optimizes for the Intel(R) Pentium(R) 4 proces­ sor. This is the default. ⋠-arch SSE Optimizes for Intel Pentium 4 processors with Streaming SIMD Extensions (SSE). ⋠-arch SSE2 Optimizes for Intel Pentium 4 processors with Streaming SIMD Extensions 2 (SSE2). The only options available on Intel(R) EM64T sys­ tems are -arch pn1, -arch pn2, -arch pn3, and -arch pn4. -assume Tells the compiler to make certain assumptions. The following are -assume options: ⋠-assume bscc Tells the compiler to treat the backslash charac­ ter (\) as a C-style control (escape) character syntax in character literals. The default is -assume nobscc, which tells the compiler to treat the backslash character as a normal character instead of a control character in character lit­ erals. ⋠-assume buffered_io Tells the compiler to accumulate records in a buffer. This sets the default for opening sequen­ tial output files to BUFFERED='YES', which also occurs if the FORT_BUFFERED run-time environment variable is specified. The default is -assume nobuffered_io, which means that data will be immediately written to disk. ⋠-assume byterecl Specifies that the units for the OPEN statement RECL specifier (record length) value are in bytes for unformatted data files, not longwords (four- byte units). For formatted files, the RECL unit is always in bytes. The default is -assume nobyterecl. INQUIRE returns RECL in bytes if the unit is not open. INQUIRE returns RECL in longwords if the file is open for unformatted data (and -assume byterecl is not specified); otherwise, it returns RECL in bytes. ⋠-assume cc_omp Enables conditional compilation as defined by the OpenMP Fortran API. That is, when "!$space" appears in free-form source or "c$spaces" appears in column 1 of fixed-form source, the rest of the line is accepted as a Fortran line. If -openmp is specified, the default is -assume cc_omp; oth­ erwise, the default is -assume nocc_omp. ⋠-assume dummy_aliases Tells the compiler that dummy (formal) arguments to procedures share memory locations with other dummy arguments (aliases) or with variables shared through use association, host association, or common block use. These program semantics slow performance and do not strictly obey the Fortran 95/90 Standards. The default is -assume nodummy_aliases. ⋠-assume minus0 Tells the compiler to use Fortran 95 standard semantics for the treatment of the IEEE* floating value -0.0 in the SIGN intrinsic, which distin­ guishes the difference between -0.0 and +0.0, and to write a value of -0.0 with a negative sign on formatted output. The default is -assume nominus0, which tells the compiler to use Fortran 90/77 standard semantics in the SIGN intrinsic to treat -0.0 and +0.0 as 0.0, and to write a value of 0.0 with no sign on formatted output. ⋠-assume none Disables all the -assume options. ⋠-assume noprotect_constants Tells the compiler to pass a copy of a constant actual argument. This copy can be modified by the called routine, even though the Fortran standard prohibits such modification. The calling routine does not see any modification to the constant. The default is -assume protect_constants, which passes the constant actual argument. Any attempt to modify it results in an error. ⋠-assume nosource_include Tells the compiler to search the default direc­ tory for module files specified by a USE state­ ment or source files specified by an INCLUDE statement. The default is -assume source_include, which tells the compiler to search the directory the source file is in for any INCLUDE files or modules. ⋠-assume nounderscore Tells the compiler not to append an underscore character to external user-defined names: the main program name, named common blocks, BLOCK DATA blocks, global data names in MODULEs, and names implicitly or explicitly declared EXTERNAL. The name of a blank (unnamed) common block remains _BLNK__, and Fortran intrinsic names are not affected. The default is -assume underscore. ⋠-assume 2underscores Tells the compiler to append two underscore char­ acters to external user-defined names that con­ tain an embedded underscore: the main program name, named common blocks, BLOCK DATA blocks, global data names in MODULEs, and names implic­ itly or explicitly declared EXTERNAL. The name of a blank (unnamed) common block remains _BLNK__, and Fortran intrinsic names are not affected. The default is -assume no2underscores. This option does not affect external names that do not contain an embedded underscore. By default, the compiler only appends one underscore to those names. -auto Causes all local, non-SAVEd variables to be allo­ cated on the run-time stack (same as -automatic or -nosave). The default is -auto-scalar. However, if you specify -recursive or -openmp, the default is -automatic. -auto-scalar Causes allocation of scalar variables of intrinsic types INTEGER, REAL, COMPLEX, and LOGICAL to the run-time stack. This is the default. However, if you specify -recursive or -openmp, the default is -automatic. You cannot specify -save, -auto, or -automatic with this option. -autodouble Makes default real and complex variables 8 bytes long. REAL declarations are treated as DOUBLE PRE­ CISION (REAL(KIND=8)) and COMPLEX declarations are treated as DOUBLE COMPLEX (COMPLEX(KIND=8)). This option is the same as specifying -real_size 64 or -r8. -automatic Causes all local, non-SAVEd variables to be allo­ cated on the run-time stack (same as -auto or -nosave). The default is -auto-scalar. However, if you specify -recursive or -openmp, the default is -automatic. -ax

(i32 and i32em) Generates processor-specific code if there is a performance benefit, while also generating generic IA-32 code. The characters K, W, N, B, and P denote the processor types (

). The only options avail­ able on Intel(R) EM64T systems are -axW and -axP. The following are -ax options: ⋠-axK Generates code for Intel Pentium III processors and compatible Intel processors. ⋠-axW Generates code for Intel Pentium 4 processors and compatible Intel processors. ⋠-axN Generates code for Intel Pentium 4 processors and compatible Intel processors. Also enables new optimizations in addition to Intel processor-spe­ cific optimizations. ⋠-axB Generates code for Intel Pentium M processors and compatible Intel processors. Also enables new optimizations in addition to Intel processor-spe­ cific optimizations. ⋠-axP Generates code for Intel(R) Pentium(R) 4 processors with Streaming SIMD Extensions 3 (SSE3) instruction support. Also enables new optimizations in addition to Intel processor-spe­ cific optimizations. You can use more than one of the -ax options by combining the characters that denote the processor type. For example, you can specify -axNB to gener­ ate code for Intel(R) Pentium(R) 4 processors and Intel Pentium M processors. If you specify both the -ax and -x options, the generic code will only execute on processors com­ patible with the processor type specified by the -x option. -B

Specifies a directory that can be used to find include files, libraries, and executables. The com­ piler uses as a prefix. For include files, the is converted to -I//include. This command is added to the front of the includes passed to the preprocessor. For libraries, the is converted to -L/. This command is added to the front of the standard -L inclusions before system libraries are added. For executables, if contains the name of a tool, such as ld or as, the compiler will use it instead of those found in the default directories. The compiler looks for include files in /include while library files are looked for in . -Bdynamic Enables dynamic linking of libraries at run time. Smaller executables are created than with static linking. -Bstatic Enables static linking of a user's library. -c Prevents linking. It causes the compiler to com­ pile to an object (.o) file only. -C Enables all checks on run-time conditions (same as the -check all option). -CB Performs run-time checks on whether array subscript and substring references are within declared bounds (same as the -check bounds option). -ccdefault Specifies the type of carriage control used when a file is displayed at a terminal screen (units 6 and *). The following are -ccdefault options: ⋠-ccdefault default Specifies that the compiler is to use the default carriage-control setting. This is the default. The default setting can be affected by the -vms option: if "-vms -ccdefault default" is speci­ fied, carriage control defaults to FORTRAN if the file is formatted, and the unit is connected to a terminal; if "-novms -ccdefault default" is spec­ ified, carriage control defaults to LIST. ⋠-ccdefault fortran Specifies normal Fortran interpretation of the first character. ⋠-ccdefault list Specifies one line feed between records. ⋠-ccdefault none Specifies no carriage control processing. -check [keyword] Checks for certain conditions at run time. The following are -check options: ⋠-check all Enables all -check options. This is the same as specifying -check with no keyword. ⋠-check arg_temp_created Generates code to check if actual arguments are copied into temporary storage before routine calls. If a copy is made at run time, an infor­ mative message is displayed. The default is -check noarg_temp_created. ⋠-check bounds Performs run-time checks on whether array sub­ script and substring references are within declared bounds. The default is -check nobounds. ⋠-check format Issues the run-time FORVARMIS fatal error when the data type of an item being formatted for out­ put does not match the format descriptor being used (for example, a REAL*4 item formatted with an I edit descriptor). If -vms is specified, the default is -check format; otherwise, the default is -check noformat. With -check noformat, the data item is formatted using the specified descriptor unless the length of the item cannot accommodate the descriptor (for example, it is still an error to pass an INTEGER*2 item to an E edit descriptor). ⋠-check none Disables all -check options. This is the default. This is the same as specifying -nocheck. ⋠-check output_conversion Issues the run-time OUTCONERR continuable error message when a data item is too large to fit in a designated format descriptor field. The field is filled with asterisks (*) and execution contin­ ues. If -vms is specified, the default is -check output_conversion; otherwise, the default is -check nooutput_conversion. -cm Suppresses all messages about questionable program­ ming practices (same as the -warn nousage option). -common-args Tells the compiler that dummy (formal) arguments to procedures share memory locations with other dummy arguments or with variables shared through use association, host association, or common block use. This is the same as specifying -assume dummy_aliases. -complex-limited-range Enables the use of basic algebraic expansions of some arithmetic operations involving data of type COMPLEX. This can cause some performance improve­ ments in programs that use a lot of COMPLEX arith­ metic, but values at the extremes of the exponent range may not compute correctly. The default is -no-complex-limited-range, which disables this option. -convert Specifies the format for unformatted files contain­ ing numeric data. The following are -convert options: ⋠-convert big_endian Specifies that the format will be big endian for INTEGER*1, INTEGER*2, INTEGER*4, or INTEGER*8, and big endian IEEE floating-point for REAL*4, REAL*8, REAL*16, COMPLEX*8, COMPLEX*16, or COM­ PLEX*32. ⋠-convert cray Specifies that the format will be big endian for INTEGER*1, INTEGER*2, INTEGER*4, or INTEGER*8, and CRAY* floating-point for REAL*8 or COM­ PLEX*16. ⋠-convert fdx Specifies that the format will be little endian for INTEGER*1, INTEGER*2, INTEGER*4, or INTE­ GER*8, VAX processor floating-point format F_floating for REAL*4 or COMPLEX*8, D_floating for REAL*8 or COMPLEX*16, and X_floating for REAL*16 or COMPLEX*32. ⋠-convert fgx Specifies that the format will be little endian for INTEGER*1, INTEGER*2, INTEGER*4, or INTE­ GER*8, VAX processor floating-point format F_floating for REAL*4 or COMPLEX*8, G_floating for REAL*8 or COMPLEX*16, and X_floating for REAL*16 or COMPLEX*32. ⋠-convert ibm Specifies that the format will be big endian for INTEGER*1, INTEGER*2, or INTEGER*4, and IBM* Sys­ tem\370 floating-point format for REAL*4 or COM­ PLEX*8 (IBM short 4) and REAL*8 or COMPLEX*16 (IBM long 8). ⋠-convert little_endian Specifies that the format will be little endian for INTEGER*1, INTEGER*2, INTEGER*4, or INTEGER*8 and little endian IEEE floating-point for REAL*4, REAL*8, REAL*16, COMPLEX*8, COMPLEX*16, or COM­ PLEX*32. ⋠-convert native Specifies that unformatted data should not be converted. This is the default. ⋠-convert vaxd Specifies that the format will be little endian for INTEGER*1, INTEGER*2, INTEGER*4, or INTE­ GER*8, VAX processor floating-point format F_floating for REAL*4 or COMPLEX*8, D_floating for REAL*8 or COMPLEX*16, and H_floating for REAL*16 or COMPLEX*32. ⋠-convert vaxg Specifies that the format will be little endian for INTEGER*1, INTEGER*2, INTEGER*4, or INTE­ GER*8, VAX processor floating-point format F_floating for REAL*4 or COMPLEX*8, G_floating for REAL*8 or COMPLEX*16, and H_floating for REAL*16 or COMPLEX*32. -cpp Runs the Fortran preprocessor on source files prior to compilation (same as the -fpp option). -cxxlib- Tells the compiler to link using certain C++ run- time libraries. You can specify one of the follow­ ing values for : ⋠gcc[=] Tells the compiler to link using the C++ run-time libraries provided with the gcc compiler. is an optional top-level location for the gcc binaries and libraries. ⋠icc Tells the compiler to link using the C++ run-time libraries provided by Intel. -D[=] Specifies as a definition (symbol) to use with conditional compilation directives or the For­ tran preprocessor (-fpp). can be an inte­ ger or it can be a character string delimited by double quotes; for example, -Dname="string". If is not specified, is defined as "1". -DD Compiles debug statements (indicated by D in column 1); this is the same as specifying -d_lines. -d_lines Compiles debug statements (indicated by D in column 1); this is the same as specifying -DD. The default is -nod_lines. -debug [keyword] Specifies settings that enhance debugging. The following are -debug options: ⋠-debug inline_debug_info Produces enhanced debug information for inlined code. It provides more information to debuggers for function call traceback. The default is -debug noinline_debug_info. ⋠-debug semantic_stepping Produces enhanced debug information useful for breakpoints and stepping. It tells the debugger to stop only at machine instructions that achieve the final effect of a source statement. For exam­ ple, in the case of an assignment statement, this might be a store instruction that assigns a value to a program variable; for a function call, it might be the machine instruction that executes the call. Other instructions generated for those source statements are not displayed during step­ ping. The default is -debug nosemantic_stepping. ⋠-debug variable_locations Produces enhanced debug information useful in finding scalar local variables. It uses a fea­ ture of the Dwarf object module known as "loca­ tion lists". This feature allows the run-time locations of local scalar variables to be speci­ fied more accurately; that is, whether, at a given position in the code, a variable value is found in memory or a machine register. The default is -check novariable_locations. ⋠-debug extended Sets the debug options: inline_debug_info, seman­ tic_stepping, and variable_locations. -debug-parameters [keyword] Tells the compiler to generate debug information for PARAMETERs used in a program. The following are -debug-parameters options: ⋠-debug-parameters none Generates no debug information for any PARAMETERs used in the program. This is the default; it is the same as specifying -nodebug-parameters. ⋠-debug-parameters used Generates debug information for only PARAMETERs that have actually been referenced in the pro­ gram. This is the same as specifying -debug-parameters with no keyword. ⋠-debug-parameters all Generates debug information for all PARAMETERs defined in the program. Note that if a .mod file contains PARAMETERs, debug information is only generated for the PARAMETERs that have actually been referenced in the program, even if you specify -debug-parameters all. -double_size Defines the default KIND for DOUBLE PRECISION and DOUBLE COMPLEX declarations, constants, functions, and intrinsics. can be 64 (KIND=8) or 128 (KIND=16). The default is -double_size 64. -dryrun Tells the driver that tool commands should be shown but not executed. See also -v. -dynamic-linker Specifies a dynamic linker () other than the default. -dyncom "common1,common2,common3" Enables dynamic allocation of the specified COMMON blocks at run time. The quotes must be present. -E Causes the Fortran preprocessor to send output to stdout. -e90 Causes the compiler to issue errors instead of warnings for nonstandard Fortran 90 (same as the -warn stderrors option). No such errors or warn­ ings are issued by default. -e95 Causes the compiler to issue errors instead of warnings for nonstandard Fortran (same as the -warn stderrors option). No such errors or warnings are issued by default. -EP Causes the Fortran preprocessor to send output to stdout, omitting #line directives. -error_limit Specifies the maximum number of error-level or fatal-level compiler errors allowed for a file specified on the command line. If you specify -noerror_limit, there is no limit to the number of errors that are allowed. The default is -error_limit 30 (a maximum of 30 error-level and fatal-level messages before the compiler stops the compilation). -extend_source [size] Specifies the column number to use to end the statement field in fixed-form source files. [size] can be 72, 80, or 132. The default behavior is -noextend_source, which implies column 72. If you do not specify size, it is the same as specifying -extend_source 132. -F Causes the Fortran preprocessor to send output to a file (same as the -preprocess_only and -P options). Note that the source file is not compiled. -f66 Tells the compiler to apply FORTRAN 66 semantics: the execution of at least one iteration of DO loops, different EXTERNAL statement syntax and semantics, and different behavior of the BLANK= and STATUS= specifiers on the OPEN statement. This is the same as specifying the -66 option. By default, the compiler applies Fortran 95 semantics. -f77rtl Tells the compiler to use the run-time behavior of Fortran 77 instead of Intel(R) Fortran. This affects some INQUIRE specifiers when the unit is not connected to a file, PAD= defaults to 'NO' for formatted input, NAMELIST input format is differ­ ent, and NAMELIST and list-directed input of char­ acter strings must be delimited by apostrophes or quotes. The default is -nof77rtl. -fast Maximizes speed across the entire program. On Ita­ nium(R)-based systems, this option sets options -ipo, -O3, and -static. On IA-32 and Intel(R) EM64T systems, this option sets options -ipo, -O3, -no-prec-div, -static, and -xP. Note that programs compiled with the -xP option will detect non-compatible processors and generate an error message during execution. -fcode-asm Produces an assembly file with optional machine code annotations. To use this option, you must also specify -S. -FI Specifies source files are in fixed format (same as the -fixed option). -fixed Specifies source files are in fixed format. By default, source file format is determined by the file suffix. -fltconsistency Enables improved floating-point consistency. Float­ ing-point operations are not reordered and the result of each floating-point operation is stored in the target variable rather than being kept in the floating-point processor for use in a subse­ quent calculation. This is the same as specifying -mp or -mieee-fp. The default, -nofltconsistency, provides better accuracy and run-time performance at the expense of less consistent floating-point results. -fmath-errno Tells the compiler to assume that the program tests errno after calls to math library functions. This restricts optimization because it causes the com­ piler to treat most math functions as having side effects. The default, -fno-math-errno, tells the compiler to assume that the program does not test errno after calls to math library functions. This frequently allows the compiler to generate faster code. Floating-point code that relies on IEEE exceptions instead of errno to detect errors can safely use this option to improve performance. -fminshared Specifies that a compilation unit is a component of a main program and will not be linked as part of a shareable object. -fno-alias Specifies that aliasing should not be assumed in the program. The default is -falias. -fno-fnalias Specifies that aliasing should not be assumed within functions, but should be assumed across calls. The default is -ffnalias. -fno-omit-frame-pointer (i32 and i32em) Disables use of EBP as a general purpose register so it can be used as a stack frame pointer. This is the same as specifying -fp. The default, -fomit- frame-pointer, enables EBP to be used as a general purpose register. -fnsplit Enables function splitting if -prof-use is also specified. Otherwise, the default is -fnsplit-, which disables the splitting within a routine but leaves function grouping enabled. -fp (i32 and i32em) Disables use of EBP as a general purpose register so it can be used as a stack frame pointer. By default, EBP is used as a general purpose register. This is the same as specifying -fno-omit-frame- pointer. -fpconstant Tells the compiler to extend the precision to dou­ ble precision for single-precision constants assigned to double-precision variables. The default is -nofpconstant. -fpe Specifies floating-point exception handling for the main program at run-time. You can specify one of the following values for : 0 - Floating underflow results in zero; all other floating-point exceptions abort execution. 1 - Floating underflow results in zero; all other floating-point exceptions produce exceptional val­ ues (signed Infinities or NaNs) and execution con­ tinues. 3 - All floating-point exceptions produce excep­ tional values (signed Infinities, denormals, or NaNs) and execution continues. This is the default; it provides full IEEE support. (Also see -ftz.) -fpic Generates position-independent code. On Itanium- based systems and Intel EM64T systems, this option must be used when building shared objects. This option can also be specified as -fPIC. The default is -fno-pic. -fpp Runs the Fortran preprocessor on source files prior to compilation. -fpscomp [keyword] Specifies the compatibility with Microsoft* Fortran PowerStation or Intel Fortran. The following are -fpscomp options: ⋠-fpscomp all Specifies that all options should be used for Fortran PowerStation compatibility. This is the same as specifying -fpscomp with no keyword. The default is -fpscomp libs. ⋠-fpscomp filesfromcmd Specifies that Fortran PowerStation behavior is used when the OPEN file specifier is blank. The default is -fpscomp nofilesfromcmd. ⋠-fpscomp general Specifies that Fortran PowerStation semantics are used when differences exist with Intel Fortran. The default is -fpscomp nogeneral. ⋠-fpscomp ioformat Specifies that Fortran PowerStation semantics and record format for list-directed formatted and unformatted I/O should be used. The default is -fpscomp noioformat. ⋠-fpscomp ldio_spacing Specifies that a blank should not be inserted after a numeric value before a character value (undelimited character string). This representa­ tion is used by Intel Fortran releases before Version 8.0 and by Fortran PowerStation. If you specify -fpscomp general, it sets -fpscomp ldio_spacing. The default is -fpscomp noldio_spacing, which conforms to the Fortran 95 standard by inserting a blank after a numeric value before a character value. ⋠-fpscomp nolibs Prevents the portability library from being passed to the linker. The default is -fpscomp libs. ⋠-fpscomp logicals Specifies that the integer values 1 and 0 are used to represent the LOGICAL values .TRUE. and .FALSE. respectively. This representation is used by Intel Fortran releases before Version 8.0 and by Fortran PowerStation. The default is -fpscomp nologicals, which speci­ fies that odd integer values are treated as true and even integer values are treated as false; specifically .TRUE. and .FALSE. are -1 and 0 respectively. This representation is used by Compaq Visual Fortran. ⋠-fpscomp none Specifies that no options should be used for For­ tran PowerStation compatibility. This is the same as specifying -nofpscomp. -fpstkchk (i32 and i32em) Generates extra code after every function call to ensure that the FP (floating-point) stack is in the expected state. By default, there is no checking. So when the FP stack overflows, a NaN value is put into FP calculations, and the program's results differ. Unfortunately, the overflow point can be far away from the point of the actual bug. The -fpstkchk option places code that causes an access violation exception immediately after an incorrect call occurs, thus making it easier to locate these issues. -FR Specifies source files are in free format (same as the -free option). -fr32 (i64 only) Disables use of high floating-point registers. Uses only the lower 32 floating-point registers. -free Specifies source files are in free format. By default, source file format is determined by the file suffix. -fsource-asm Produces an assembly file with optional source code annotations. To use this option, you must also specify -S. -ftrapuv Initializes stack local variables to an unusual value to aid error detection. Normally, these local variables should be initialized in the appli­ cation. The option sets any uninitialized local variables that are allocated on the stack to a value that is typically interpreted as a very large integer or an invalid address. References to these variables are then likely to cause run-time errors that can help you detect coding errors. -ftz (i32 and i64) Enables flush denormal results to zero. The default is -no-ftz. On Itanium-based systems, option -O3 sets the -ftz option. -funroll-loops Tells the compiler to use default heuristics to unroll loops. This is the same as specifying -unroll. -fverbose-asm Produces an assembly file with compiler comments, including options and version information. To use this option, you must also specify -S, which sets -fverbose-asm. If you do not want this default when you specify -S, specify -fnoverbose-asm. -fvisibility= -fvisibility-= The first form specifies the default visibility for global symbols. The second form specifies the vis­ ibility for symbols that are in a file (this form overrides the first form). is the pathname of a file containing the list of symbols whose vis­ ibility you want to set; the symbols are separated by whitespace (spaces, tabs, or newlines). specifies the visibility setting; it can be any of the following: default - This setting means other components can reference the symbol, and the symbol definition can be overridden (preempted) by a definition of the same name in another component. extern - This setting means the symbol is treated as though it is defined in another component. It also means that the symbol can be overridden by a definition of the same name in another component. hidden - This setting means other components cannot directly reference the symbol. However, its address might be passed to other components indi­ rectly. internal - This setting means the symbol cannot be referenced outside its defining component, either directly or indirectly. protected - This setting means other components can reference the symbol, but it cannot be overridden by a definition of the same name in another compo­ nent. -g Produces symbolic debug information in the object file. The compiler does not support the generation of debugging information in assemblable files. If you specify the -g option, the resulting object file will contain debugging information, but the assemblable file will not. On IA-32 systems, specifying the -g or -O0 option automatically sets the -fp option. -gen-interfaces Tells the compiler to generate an interface block for each routine (that is, for each SUBROUTINE and FUNCTION statement) defined in the source file. The compiler generates two files for each routine, a .mod file and a .f90 file, and places them in the current directory or in the directory specified by the include or module option. The .f90 file is the text of the interface block; the .mod file is the interface block compiled into binary form. The default is -nogen-interfaces. -help Displays the list of compiler options. -I Specifies a directory to add to the include path, which is used to search for module files (USE statement) and include files (INCLUDE statement). -i-dynamic Links Intel-provided libraries dynamically. It is the opposite of -i-static. -i-static Links Intel-provided libraries statically. It is the opposite of -i-dynamic. -i2 Makes default integer and logical variables 2 bytes long (same as the -integer_size 16 option). The default is -integer_size 32. -i4 Makes default integer and logical variables 4 bytes long (same as the -integer_size 32 option). This is the default. -i8 Makes default integer and logical variables 8 bytes long (same as the -integer_size 64 option). The default is -integer_size 32. -implicitnone Sets the default type of a variable to undefined (IMPLICIT NONE). This is the same as specifying the -u option. -inline [keyword] Specifies the level of inline function expansion. The following are -inline options: ⋠-inline none Disables inlining of user-defined functions. This is the same as specifying -inline manual. ⋠-inline manual Disables inlining of user-defined functions. For­ tran statement functions are always inlined. ⋠-inline size Enables inlining of any function. However, the compiler decides which functions are inlined. This option enables interprocedural optimizations and most speed optimizations. ⋠-inline speed Enables inlining of any function. This is the same as specifying -inline all. ⋠-inline all Enables inlining of any function. However, the compiler decides which functions are inlined. This option enables interprocedural optimizations and all speed optimizations. If you do not specify a keyword for -inline, it is the same as specifying -inline all. -inline-debug-info Produces enhanced source position information for inlined code. This leads to greater accuracy when reporting the source location of any instruction. It also provides enhanced debug information useful for function call traceback. To use this option for debugging, you must also specify a debug enabling option, such as -g. -intconstant Tells the compiler to use Fortran 77 semantics, rather than Fortran 95/90 semantics, to determine the KIND for integer constants. The default is -nointconstant. -integer_size Defines the size of INTEGER and LOGICAL variables. can be 16, 32, or 64. The default is -inte­ ger_size 32. -ip Enables more interprocedural optimizations for sin­ gle file compilation. If you specify this option, the compiler performs inline function expansion for calls to functions defined within the current source file. -ip-no-inlining Disables full and partial inlining enabled by -ip or -ipo. -ip-no-pinlining (i32 and i32em) Disables partial inlining. To use this option, you must specify -ip or -ipo. -IPF-flt-eval-method0 (i64 only) Tells the compiler to evaluate the expressions involving floating-point operands in the precision indicated by the variable types declared in the program. By default, intermediate floating-point expressions are maintained in higher precision. -IPF-fltacc (i64 only) Disables optimizations that affect floating-point accuracy. If the default setting is used (-IPF- fltacc-), the compiler may apply optimizations that reduce floating-point accuracy. You can use -IPF- fltacc or -fltconsistency to improve floating-point accuracy, but at the cost of disabling some opti­ mizations. -IPF-fp-relaxed (i64 only) -IPF_fp_relaxed (i64 only) Enables use of faster but slightly less accurate code sequences for math functions, such as divide and sqrt. When compared to strict IEEE* precision, this option slightly reduces the accuracy of float­ ing-point calculations performed by these func­ tions, usually limited to the least significant digit. -IPF-fp-speculation (i64 only) Tells the compiler to speculate on floating-point (FP) operations in one of the following s: fast - Speculate on floating-point operations. This is the default. safe - Speculate on floating-point operations only when safe. strict - This is the same as specifying off. off - Disables speculation of floating-point opera­ tions. -ipo[n] Enables multifile interprocedural (IP) optimiza­ tions (between files). When you specify this option, the compiler performs inline function expansion for calls to functions defined in sepa­ rate files. n is an optional integer that specifies the number of object files the compiler should create. Any integer greater than or equal to 0 is valid. If n is 0, the compiler decides whether to create one or more object files based on an estimate of the size of the application. It generates one object file for small applications, and two or more object files for large applications. If n is greater than 0, the compiler generates n object files, unless n exceeds the number of source files (m), in which case the compiler generates only m object files. If you do not specify n, the default is 0. -ipo-c Tells the compiler to generate a multifile object file (ipo_out.o) that can be used in further link steps. -ipo-S Tells the compiler to generate a multifile assembly file (ipo_out.s) that can be used in further link steps. -ipo-separate Tells the compiler to generate one object file per source file. This option overrides any -ipo[n] specification. -ivdep-parallel (i64 only) Tells the compiler that there is no loop-carried memory dependency in any loop following an IVDEP directive. -Kpic This is a deprecated option; it can also be speci­ fied as -KPIC. Use -fpic instead. -L Tells the linker to search for libraries in before searching the standard directories. -logo Displays the compiler version information (same as the -V option). The default is -nologo. -lowercase Causes the compiler to ignore case differences in identifiers and to convert external names to lower­ case (same as the -names lowercase option). This is the default. -map-opts Maps one or more Linux compiler options to their equivalent on a Windows system and outputs the result to stdout. The tool can be invoked from the compiler command line or it can be used directly. No compilation is performed when the option mapping tool is used. Compiler options are mapped to their equivalent on the architecture you are using. -mieee-fp Enables improved floating-point consistency. Float­ ing-point operations are not reordered and the result of each floating-point operation is stored in the target variable rather than being kept in the floating-point processor for use in a subse­ quent calculation. This is the same as specifying -fltconsistency or -mp. The default, -mno-ieee-fp, provides better accuracy and run-time performance at the expense of less consistent floating-point results. -mixed_str_len_arg Tells the compiler that the hidden length passed for a character argument is to be placed immedi­ ately after its corresponding character argument in the argument list. The default is -nomixed_str_len_arg, which places the hidden lengths in sequential order at the end of the argu­ ment list. -module Specifies the directory where module (.mod) files should be placed when created and where they should be searched for (USE statement). -mp Maintains floating-point precision (while disabling some optimizations). The -mp option restricts opti­ mization to maintain declared precision and to ensure that floating-point arithmetic conforms more closely to the ANSI* and IEEE standards. This is the same as specifying -fltconsistency or -mieee- fp. For most programs, specifying this option adversely affects performance. If you are not sure whether your application needs this option, try compiling and running your program both with and without it to evaluate the effects on both performance and precision. -mp1 Improves floating-point precision and consistency. This option disables fewer optimizations and has less impact on performance than -fltconsistency or -mp. -mtune= Performs optimizations for a specified CPU. On IA-32 systems, you can specify one of the fol­ lowing values for : ⋠pentium Optimizes for Intel(R) Pentium(R) processors. ⋠pentiumpro Optimizes for Intel(R) Pentium(R) Pro, Intel Pen­ tium II, and Intel Pentium III processors. ⋠pentium4 Optimizes for Intel Pentium 4 processors. This is the default. ⋠pentium-mmx Optimizes for Intel(R) Pentium(R) with MMX(TM) technology. On Itanium(R)-based systems, you can specify one of the following values for : ⋠itanium Optimizes for Intel(R) Itanium(R) processors. ⋠itanium2 Optimizes for Intel(R) Itanium(R) 2 processors. This is the default. -names Specifies how source code identifiers and external names are interpreted. The following are -names options: ⋠-names as_is Causes the compiler to distinguish case differ­ ences in identifiers and to preserve the case of external names. ⋠-names lowercase Causes the compiler to ignore case differences in identifiers and to convert external names to low­ ercase. This is the default. ⋠-names uppercase Causes the compiler to ignore case differences in identifiers and to convert external names to uppercase. -nbs Tells the compiler to treat a backslash (\) as a normal character in character literals, not an escape character (same as the -assume nobscc option). This is the default. -no-cpprt Prevents linking of the C++ run-time libraries. -no-fp-port (i32 and i32em) Tells the compiler to keep results of floating- point operations in higher precision. This pro­ vides better performance but less consistent floating-point results. The default is -fp-port, which rounds floating-point results after floating- point operations so rounding to user-declared pre­ cision occurs at assignments and type conversions. This has some impact on speed. -no-global-hoist Disables certain optimizations, such as load hoist­ ing and speculative loads, that can move memory loads to a point earlier in the program execution than where they appear in the source. This option is useful for some applications, such as those that use shared or dynamically mapped mem­ ory, which can fail if a load is moved too early in the execution stream (for example, before the mem­ ory is mapped). In most cases, these optimizations are safe and can improve performance. The default, -global-hoist, enables these optimizations. -no-IPF-fma (i64 only) Disables the combining of floating-point multiplies and add/subtract operations. It also disables the contraction of floating-point multiply and add/sub­ tract operations into a single operation. The default is -IPF-fma, which tells the compiler to combine and contract these operations whenever pos­ sible. However, if you specify -mp and do not explicitly specify -IPF-fma, the default is -no- IPF-fma. -no-prefetch (i32 only) Disables prefetch insertion optimization. To use this option, you must also specify -O3. The default is -prefetch. -noalign Prevents the alignment of data items. This is the same as specifying -align none. The default is -align. -noaltparam Specifies that the alternate form of parameter con­ stant declarations (without parentheses) should not be recognized (same as the -nodps option). This form has no parentheses surrounding the list, and the form of the constant, rather than implicit or explicit typing, determines the data type of the variable. The default is -altparam. -nobss-init Places any variables that are explicitly initial­ ized with zeros in the DATA section. By default, variables explicitly initialized with zeros are placed in the BSS section. -nodefaultlibs Prevents the compiler from using standard libraries when linking. -nodefine Specifies that all preprocessor definitions apply only to -fpp and not to Intel Fortran conditional compilation directives. This is the same as specifying -noD. -nodps Specifies that the alternate form of parameter con­ stant declarations (without parentheses) should not be recognized (same as the -noaltparam option). The default is -dps. -nofor_main Specifies the main program is not written in For­ tran, and prevents the compiler from linking for_main.o into applications. The default is -for_main. -noinclude Prevents the compiler from searching in /usr/include for files specified in an INCLUDE statement. You can specify the -I option along with this option. This option does not affect the fpp preprocessor behavior, and is not related to the Fortran 95/90 USE statement. -nostartfiles Prevents the compiler from using standard startup files when linking. -nostdinc Removes standard directories from the include file search path (same as the -X option). -nostdlib Prevents the compiler from using standard libraries and startup files when linking. -nus Prevents the compiler from appending an underscore character to external user-defined names. This option is the same as the -assume nounderscore option, and is the opposite of -us. -o Specifies the name () for an output file as follows: ⋠If -c is specified, -o specifies the name of an object file. ⋠If -S is specified, -o specifies the name of an assembly listing file. ⋠Otherwise, -o specifies the name of the exe­ cutable file. -O0 Disables all -O optimizations. On IA-32 and Intel(R) EM64T systems, this option sets the -fp option. -O1 On IA-32 and Intel(R) EM64T systems, enables opti­ mizations for speed. Also disables intrinsic recog­ nition and the -fp option. This option is the same as the -O2 option. On Itanium-based systems, the -O1 option enables optimizations for server applications (straight- line and branch-like code with a flat profile). Enables optimizations for speed, while being aware of code size. For example, this option disables software pipelining and loop unrolling. -O2 or -O This option is the default for optimizations. How­ ever, if -g is specified, the default is -O0. On Itanium-based systems, the -O2 option enables optimizations for speed, including global code scheduling, software pipelining, predication, and speculation. It also enables: ⋠Inlining of intrinsics ⋠Intra-file interprocedural optimizations, which include: inlining, constant propagation, forward substitution, routine attribute propagation, variable address-taken analysis, dead static function elimination, and removal of unreferenced variables. ⋠The following capabilities for performance gain: constant propagation, copy propagation, dead-code elimination, global register allocation, global instruction scheduling and control speculation, loop unrolling, optimized code selection, partial redundancy elimination, strength reduction/induc­ tion variable simplification, variable renaming, exception handling optimizations, tail recur­ sions, peephole optimizations, structure assign­ ment lowering and optimizations, and dead store elimination. -O3 Enables -O2 optimizations plus more aggressive optimizations, such as prefetching, scalar replace­ ment, and loop transformations. Enables optimiza­ tions for maximum speed, but does not guarantee higher performance unless loop and memory access transformations take place. On IA-32 and Intel(R) EM64T systems, when the -O3 option is used with the -ax and -x options, it causes the compiler to perform more aggressive data dependency analysis than for -O2, which may result in longer compilation times. On Itanium-based systems, the -O3 option enables optimizations for technical computing applications (loop-intensive code): loop optimizations and data prefetch. -Ob Controls inline expansion. You can specify one of the following values for : 0 - Disables inlining. However, statement functions are always inlined. 1 - Enables inlining of functions if option -ip is specified, but expands only INLINE directives. This is the default. 2 - Enables inlining of any routine, at the com­ piler's discretion. Enables interprocedural opti­ mizations (has the same effect as the -ip option). -onetrip Tells the compiler to execute at least one itera­ tion of DO loops (same as the -1 option). This option has the same effect as -f66. -openmp Enables the parallelizer to generate multithreaded code based on OpenMP* directives. The code can be executed in parallel on both uniprocessor and mul­ tiprocessor systems. The -openmp option works with both -O0 (no optimization) and any optimization level of -O. Specifying -O0 with -openmp helps to debug OpenMP applications. -openmp-profile Enables analysis of OpenMP* applications. To use this option, you must have Intel(R) Thread Profiler installed, which is one of the Intel(R) Threading Tools. If this threading tool is not installed, this option has no effect. -openmp-report[n] Controls the OpenMP parallelizer's level of diag­ nostic messages. You can specify one of the follow­ ing values for [n]: 0 - Displays no diagnostic information. 1 - Displays diagnostics indicating loops, regions, and sections successfully parallelized. This is the default. 2 - Displays the diagnostics specified by -openmp- report1 plus diagnostics indicating successful han­ dling of MASTER constructs, SINGLE constructs, CRITICAL constructs, ORDERED constructs, ATOMIC directives, etc. -openmp-stubs Tells the compiler to generate sequential OpenMP code. The OpenMP directives are ignored and a stub OpenMP library is linked. -opt-report Tells the compiler to generate an optimization report to stderr. -opt-report-file Tells the compiler to generate an optimization report named . -opt-report-help Displays the logical names of optimizer phases available for report generation (using -opt-report- phase). -opt-report-level[level] Specifies the detail level of the optimization report. The [level] can be min, med, or max. The default is -opt-report-levelmin. -opt-report-phase Specifies the optimizer phase () to generate reports for. This option can be used multiple times on the same command line to generate reports for multiple optimizers. Currently, the following optimizer phases are supported: ipo - Interprocedural Optimizer hlo - High Level Optimizer ilo - Intermediate Language Scalar Optimizer ecg - Code Generator (Itanium(R)-based systems only) omp - OpenMP all - All phases When one of these logical names for optimizers is specified for , all reports from that opti­ mizer are generated. -opt-report-routine[string] Generates a report on the routines containing the specified as part of their name. If is not specified, reports from all rou­ tines are generated. -p Compiles and links for function profiling with gprof(1). This is the same as specifying -pg or -qp, except that -pg is not available on Itanium- based systems. -P Causes the Fortran preprocessor to send output to a file (same as the -preprocess_only and -F options). Note that the source file is not compiled. -pad Enables the changing of the variable and array mem­ ory layout. The default is -nopad. -pad-source Specifies that fixed-form source records shorter than the statement field width are to be padded with spaces (on the right) to the end of the state­ ment field. This affects the interpretation of character and Hollerith literals that are continued across source records. The default is -nopad- source. -par-report[n] Controls the auto-parallelizer's level of diagnos­ tic messages. You can specify one of the following values for [n]: 0 - Displays no diagnostic information. 1 - Displays diagnostics indicating loops success­ fully auto-parallelized. This is the default. Issues a "LOOP AUTO-PARALLELIZED" message for par­ allel loops. 2 - Displays diagnostics indicating loops success­ fully auto-parallelized, as well as unsuccessful loops. 3 - Displays the diagnostics specified by -par- report2 plus additional information about any proven or assumed dependencies inhibiting auto-par­ allelization (reasons for not parallelizing). -par-threshold[n] Sets a threshold for the auto-parallelization of loops based on the probability of profitable execu­ tion of the loop in parallel. This option is used for loops whose computation work volume cannot be determined at compile-time. The threshold is usu­ ally relevant when the loop trip count is unknown at compile-time. [n] is an integer from 0 to 100. The default value is 100. The compiler applies a heuristic that tries to bal­ ance the overhead of creating multiple threads ver­ sus the amount of work available to be shared amongst the threads. -parallel Tells the auto-parallelizer to generate multi­ threaded code for loops that can be safely executed in parallel. To use this option, you must also specify -O2 or -O3. -pc (i32 and i32em) Enables control of floating-point significand pre­ cision. Some floating-point algorithms are sensi­ tive to the accuracy of the significand, or frac­ tional part of the floating-point value. For exam­ ple, iterative operations like division and finding the square root can run faster if you lower the precision with the -pc option. You can specify one of the following values for : 32 - Rounds the significand to 24 bits (single pre­ cision). Note that a change of the default preci­ sion control or rounding mode (for example, by using the -pc32 option or by user intervention) may affect the results returned by some of the mathe­ matical functions. 64 - Rounds the significand to 53 bits (single pre­ cision). This is the default. 80 - Rounds the significand to 64 bits (double pre­ cision). -pg (i32 and i32em) Compiles and links for function profiling with gprof(1). This is the same as specifying -p or -qp, except that they are also available on Ita­ nium-based systems. -prec-div (i32 and i32em) Improves precision of floating-point divides; it has some speed impact. With some optimizations, such as -xN and -xB, the compiler may change float­ ing-point division computations into multiplication by the reciprocal of the denominator. For example, A/B is computed as A * (1/B) to improve the speed of the computation. However, sometimes the value produced by this transformation is not as accurate as full IEEE division. When it is important to have fully pre­ cise IEEE division, use -prec-div to disable the floating-point division-to-multiplication optimiza­ tion. The result is more accurate, with some loss of performance. If you specify -no-prec-div, it enables optimiza­ tions that give slightly less precise results than full IEEE division. -prec-sqrt (i32 and i32em) Improves precision of square root implementations; it has some speed impact. This option inhibits any optimizations that can adversely affect the preci­ sion of square root computation. The result is fully precise square root implementations, with some loss of performance. This is the default set­ ting if you specify -O0, -mp, or -mp1. -preprocess_only Causes the Fortran preprocessor to send output to a file (same as the -F and -P options). Note that the source file is not compiled. -print-multi-lib Prints information about where system libraries should be found, but no compilation occurs. It is provided for compatibility with gcc. -prof-dir Specifies a directory () for profiling output files (*.dyn and *.dpi). -prof-file Specifies a file name () for the profiling summary file. -prof-format-32 (i32 and i64) Produces profile data with 32-bit counters; this option allows compatibility with earlier compilers. The default is to produce profile data with 64-bit counters to handle large numbers of events. -prof-gen Instruments a program for profiling. -prof-gen-sampling (i32 only) Prepares application executables for hardware pro­ filing (sampling) and causes the compiler to gener­ ate source code mapping information. -prof-genx Instruments a program for profiling and gathers extra information for code coverage tools. -prof-use Enables use of profiling information during opti­ mization. -Qinstall Specifies the root directory ()where the com­ piler installation was performed. This option is useful if you want to use a different compiler or if you did not use the ifortvars shell script to set your environment variables. -Qlocation,, Sets as the location of the tool specified by . -Qoption,, Passes options to the tool specified by . -qp Compiles and links for function profiling with gprof(1). This is the same as specifying -p or -pg, except that -pg is not available on Itanium- based systems. -r8 Makes default real and complex variables 8 bytes long. REAL declarations are treated as DOUBLE PRE­ CISION (REAL(KIND=8)) and COMPLEX declarations are treated as DOUBLE COMPLEX (COMPLEX(KIND=8)). This option is the same as specifying -real_size 64 or -autodouble. -r16 Makes default real and complex variables 16 bytes long. REAL declarations are treated as extended precision REAL (REAL(KIND=16); COMPLEX and DOUBLE COMPLEX declarations are treated as extended preci­ sion COMPLEX (COMPLEX(KIND=16)). This option is the same as specifying -real_size 128. -rcd (i32 only) Enables fast float-to-integer conversions. This option can improve the performance of code that requires floating-point-to-integer conversions. The system default floating-point rounding mode is round-to-nearest. However, the Fortran language requires floating-point values to be truncated when a conversion to an integer is involved. To do this, the compiler must change the rounding mode to trun­ cation before each floating-point-to-integer con­ version and change it back afterwards. The -rcd option disables the change to truncation of the rounding mode for all floating-point calcu­ lations, including floating point-to-integer con­ versions. This option can improve performance, but floating-point conversions to integer will not con­ form to Fortran semantics. -real_size Defines the size of REAL and COMPLEX declarations, constants, functions, and intrinsics. can be 32, 64, or 128. The default is -real_size 32. -recursive Specifies that all routines should be compiled for possible recursive execution. This option sets the -auto option. The default is -norecursive. -reentrancy [keyword] Specifies that the compiler should generate reen­ trant code that supports a multithreaded applica­ tion. The following are -reentrancy options: ⋠-reentrancy async Tells the run-time library (RTL) that the program may contain asynchronous (AST) handlers that could call the RTL. This causes the RTL to guard against AST interrupts inside its own critical regions. ⋠-reentrancy none Tells the run-time library (RTL) that the program does not rely on threaded or asynchronous reen­ trancy. The RTL will not guard against such interrupts inside its own critical regions. This is the default. This option is the same as the -noreentrancy option. ⋠-reentrancy threaded Tells the run-time library (RTL) that the program is multithreaded, such as programs using the POSIX threads library. This causes the RTL to use thread locking to guard its own critical regions. If you do not specify a keyword for -reentrancy, it is the same as specifying -reen­ trancy threaded. -S Causes the compiler to compile to an assembly file (.s) only and not link. -safe-cray-ptr Specifies that Cray pointers do not alias other variables. -save Places variables, except those declared as AUTO­ MATIC, in static memory (same as -noauto or -noau­ tomatic). The default is -auto-scalar. However, if you specify -recursive or -openmp, the default is -automatic. -scalar-rep (i32 only) Enables scalar replacement performed during loop transformation. To use this option, you must also specify -O3. The default is -no-scalar-rep. -shared Tells the compiler to produce a dynamic shared object instead of an executable. On Itanium-based systems and Intel EM64T systems, you must specify -fpic for the compilation of each object file you want to include in the shared library. -shared-libcxa Links the Intel libcxa C++ library dynamically, overriding the default behavior when -static is used. This option is the opposite of -static- libcxa. -sox Tells the compiler to save the compiler options and version number in the executable. The default is -no-sox. -ssp (i32 only) Enables Software-based Speculative Pre-computation (SSP), which is also called Helper-Threading opti­ mization. It provides a way to dynamically prefetch data cache blocks to counterbalance ever-increasing memory latency. It exploits the properties of source code constructs (such as delinquent loads and pointer-chasing loops) in applications. -stand [keyword] Causes the compiler to issue compile-time messages for nonstandard language elements. The following are -stand options: ⋠-stand f90 Causes the compiler to issue messages for lan­ guage elements that are not standard in Fortran 90 (same as the -std90 option). ⋠-stand f95 Causes the compiler to issue messages for lan­ guage elements that are not standard in Fortran 95 (same as the -std95 or -std options). This option is set if you specify -warn stderrors. If you do not specify a keyword for -stand, it is the same as specifying -stand 95. ⋠-stand none Causes the compiler to issue no messages for non­ standard language elements. This is the same as specifying -nostand. This is the default. -static Prevents linking with shared libraries. Causes the executable to link all libraries statically. -static-libcxa Links the Intel libcxa C++ library statically. This option is the opposite of -shared-libcxa. -std90 Causes the compiler to issue messages for language elements that are not standard in Fortran 90 (same as the -stand f90 option). -std95 or -std Causes the compiler to issue messages for language elements that are not standard in Fortran 95 (same as the -stand f95 option). This option is set if you specify -warn stderrors. -syntax_only Specifies that the source file should be checked only for correct syntax (same as the -syntax and -y options). No code is generated, no object file is produced, and some error checking done by the opti­ mizer is bypassed. This option lets you do a quick syntax check of your source file. -T Tells the linker to read link commands from the specified . -tcheck Enables analysis of threaded applications. To use this option, you must have Intel(R) Thread Checker installed, which is one of the Intel(R) Threading Tools. If this threading tool is not installed, this option has no effect. -Tf Specifies that should be compiled as a For­ tran source file. This option is useful when you have a file with a nonstandard filename suffix. -threads Specifies that multithreaded libraries should be linked. This is the default on Intel(R) EM64T sys­ tems. This option sets the -reentrancy threaded option. On IA-32 and Intel(R) Itanium systems, the default is -nothreads. -tpp1 (i64 only) Optimizes for the Intel(R) Itanium(R) processor. -tpp2 (i64 only) Optimizes for the Intel(R) Itanium(R) 2 processor. This is the default on Itanium-based systems. -tpp5 (i32 only) Optimizes for the Intel(R) Pentium(R) processor. -tpp6 (i32 only) Optimizes for the Intel(R) Pentium(R) Pro, Intel(R) Pentium(R) II and Intel(R) Pentium(R) III proces­ sors. -tpp7 (i32 and i32em) Optimizes for the Intel(R) Pentium(R) 4 processors, Intel(R) Xeon(TM) processors, Intel Pentium M pro­ cessors, and Intel Pentium 4 processors with Streaming SIMD Extensions 3 (SSE3) instruction sup­ port. This is the default on IA-32 systems and Intel(R) EM64T systems. -traceback Tells the compiler to generate extra information in the object file to allow the display of source file traceback information at run time when a severe error occurs. The default is -notraceback. -tune (i32 and i32em) Determines the version of the architecture for which the compiler generates instructions. The fol­ lowing are -tune options: ⋠-tune pn1 Optimizes for the Intel(R) Pentium(R) processor. ⋠-tune pn2 Optimizes for the Intel(R) Pentium(R) Pro, Intel(R) Pentium(R) II, and Intel(R) Pentium(R) III processors. ⋠-tune pn3 Optimizes for the Intel(R) Pentium(R) Pro, Intel(R) Pentium(R) II, and Intel(R) Pentium(R) III processors. This is the same as specifying the -tune pn2 option. ⋠-tune pn4 Optimizes for the Intel(R) Pentium(R) 4 proces­ sor. This is the default. The only option available on Intel(R) EM64T systems is -tune pn4. -u Sets the default type of a variable to undefined (IMPLICIT NONE). This is the same as specifying the -implicitnone option. -U Removes the predefined macro named . -unroll[n] Sets the maximum number of times to unroll loops. Use -unroll0 to disable loop unrolling. The default is -unroll, which tells the compiler to use default heuristics. This is the same as specifying -funroll-loops. -uppercase Causes the compiler to ignore case differences in identifiers and to convert external names to upper­ case (same as the -names uppercase option). The default is -lowercase (or -names lowercase). -us Tells the compiler to append an underscore charac­ ter to external user-defined names (opposite of -nus). Specifying -us is the same as specifying the -assume underscore option. -use-asm Tells the compiler to produce objects through the assembler. -v Tells the driver that tool commands should be shown and executed. See also -dryrun. -V Displays the compiler version information (same as the -logo option). -vec-report[n] (i32 and i32em) Specifies the amount of vectorizer diagnostic information to report. You can specify one of the following values for [n]: 0 - Produces no diagnostic information. 1 - Indicates vectorized loops. This is the default. 2 - Indicates vectorized and non-vectorized loops. 3 - Indicates vectorized and non-vectorized loops and prohibiting data dependence information. 4 - Indicates non-vectorized loops. 5 - Indicates non-vectorized loops and prohibiting data dependence information. -vms Causes the run-time system to behave like HP* For­ tran on OpenVMS* Alpha systems and VAX* systems (VAX FORTRAN*) in the following ways: ⋠Modifies certain defaults The -vms option sets the -check format and -check output_conversion options. You can override this by specifying the option on the command line. For example, if you specify -vms -check noformat, you get -check noformat. ⋠Alignment The -vms option does not affect the alignment of fields in records or items in COMMON. Use -align norecords to pack fields of records on the next byte boundary for compatibility with HP Fortran on OpenVMS systems. ⋠INCLUDE qualifiers /LIST and /NOLIST are recognized at the end of the file pathname in an INCLUDE statement at com­ pile time. If the file name in the INCLUDE statement does not specify the complete path, the path used is the current directory. ⋠Quotation mark character (") A quotation mark is recognized as starting an octal constant (such as "177) instead of a char­ acter literal ("..."). ⋠Deleted records in relative files When a record in a relative file is deleted, the first byte of that record is set to a known char­ acter (currently '@'). Attempts to read that record later result in ATTACCNON errors. The rest of the record (the whole record, if -vms is not specified) is set to nulls for unformatted files and spaces for formatted files. ⋠ENDFILE records When an ENDFILE is performed on a sequential unit, an actual one byte record containing a Ctrl/Z is written to the file. If -vms is not specified, an internal ENDFILE flag is set and the file is truncated. The -vms option does not affect ENDFILE on rela­ tive files; such files are truncated. ⋠Reading deleted records and ENDFILE records The run-time direct access READ routine checks the first byte of the retrieved record. If this byte is '@' or NULL ("\0"), then ATTACCNON is returned. The run-time sequential access READ routine checks to see if the record it just read is one byte long and contains a Ctrl/Z. If this is true, it returns EOF. ⋠OPEN statement effects Carriage control defaults to FORTRAN if the file is formatted, and the unit is connected to a ter­ minal (checked by means of isatty(3)). Otherwise, carriage control defaults to LIST. The -vms option affects the record length for direct access and relative organization files. The buffer size is increased by one (to accommo­ date the deleted record character). ⋠Implied logical unit numbers Certain environment variables are recognized at run time for ACCEPT, PRINT, and TYPE statements, and for READ and WRITE statements that do not specify a unit number, such as: READ (*,1000). ⋠Treatment of blanks in input The -vms option causes the defaults for keyword BLANK in OPEN statements to become 'NULL' for an explicit OPEN, and 'ZERO' for an implicit OPEN of an external or internal file. ⋠Carriage control default If -vms -ccdefault default is specified, carriage control defaults to FORTRAN if the file is for­ matted and the unit is connected to a terminal. -w Disables all warning messages (same as the -nowarn and -warn nogeneral options). -W Disables warnings (n=0) or enables warnings (n=1). The default is -W1 (same as the -warn general option). -W0 is the same as specifying -warn nogeneral, -nowarn, or -w. -w90 Suppresses warning messages about Fortran features that are deprecated or obsolescent in Fortran 95 (same as the -stand none option). -w95 Suppresses warning messages about Fortran features that are deprecated or obsolescent in Fortran 95 (same as the -stand none option). -Wa,[,,...] Passes options , , and so forth, to the assembler for processing. If the assembler is not invoked, these options are ignored. -warn [keyword] Specifies the level of diagnostic messages issued by the compiler. The following are -warn options: ⋠-warn all Enables all warning messages. This is the same as specifying -warn with no keyword. ⋠-warn none Disables all warning messages. This is the same as specifying -nowarn. ⋠-warn noalignments Disables warnings for data that is not naturally aligned. The default is -warn alignments. ⋠-warn declarations Enables error messages about any undeclared symbols. The default is -warn nodeclarations. ⋠-warn errors Tells the compiler to change all warning-level messages into error-level messages. The default is -warn noerrors. ⋠-warn nogeneral Disables all information-level and warning-level messages. The default is -warn general. ⋠-warn ignore_loc Enables warnings when %LOC is stripped from an argument. The default is -warn noignore_loc. ⋠-warn interfaces Tells the compiler to check the interfaces of all SUBROUTINEs called and FUNCTIONs invoked in your compilation against a set of interface blocks stored separately from the source being compiled. It causes the compiler to generate a compile-time message if the interface used to invoke a routine does not match the interface defined in a .mod file external to the source (that is, in a .mod generated by -gen-interfaces as opposed to a .mod file USEd in the source). The compiler looks for these .mods in the current directory or in the directory specified by the include or module option. The default is -warn nointerfaces. ⋠-warn stderrors Tells the compiler to change warnings about For­ tran standards violations into error messages. This option sets the -std95 option. If you want Fortran 90 standards violations to become errors, you should specify -warn stderrors and -std90. The default is -warn nostderrors. ⋠-warn truncated_source Enables warnings when source exceeds the maximum columm width in fixed-format source files. The default is -warn notruncated_source. ⋠-warn uncalled Enables warnings when a statement function is never called. The default is -warn nouncalled. ⋠-warn unused Enables warnings about variables that are declared but never used. The default is -warn nounused. ⋠-warn nousage Disables warnings about questionable program practices. The default is -warn usage. -WB Turns a compile-time bounds check into a warning. Normally, compile-time bounds checks are errors. -what Displays the version strings of the Fortran command and the compiler. -Wl,[,,...] Passes options , , and so forth, to the linker for processing. If the linker is not invoked, these options are ignored. -Wp,[,,...] Passes options , , and so forth, to the preprocessor. If the preprocessor is not invoked, these options are ignored. -X Removes standard directories from the include file search path (same as the -nostdinc option). You can use the -X option with the -I option to prevent the compiler from searching the default path for include files and direct it to use an alternate path. -x

(i32 and i32em) Generates the minimum set of processor-specific instructions required for the processor that exe­ cutes your program. The characters K, W, N, B, and P denote the processor types (

). The only options available on Intel(R) EM64T systems are -xW and -xP. The following are -x options: ⋠-xK Generates code for Intel Pentium III processors and compatible Intel processors. ⋠-xW Generates code for Intel Pentium 4 processors and compatible Intel processors. ⋠-xN Generates code for Intel Pentium 4 processors and compatible Intel processors. Also enables new optimizations in addition to Intel processor-spe­ cific optimizations. ⋠-xB Generates code for Intel Pentium M processors and compatible Intel processors. Also enables new optimizations in addition to Intel processor-spe­ cific optimizations. ⋠-xP Generates code for Intel(R) Pentium(R) 4 proces­ sors with Streaming SIMD Extensions 3 (SSE3) instruction support. Also enables new optimiza­ tions in addition to Intel processor-specific optimizations. You can use more than one of the -x options by com­ bining the characters that denote the processor type. For example, you can specify -xNB to generate code for Intel(R) Pentium(R) 4 processors and Intel Pentium M processors. Do not use these options if you are executing a program on a non-Intel(R) processor. If you use these options on a non-compatible processor, the program might fail with an illegal instruction exception or display other unexpected behavior. If you execute a program using option -xN, -xB, or -xP on an unsupported processor, the compiler will display a fatal run-time error. For more informa­ tion, see your user's guide. -Xlinker Passes directly to the linker for process­ ing. -y Specifies that the source file should be checked only for correct syntax (same as the -syntax_only and -syntax options). -zero Initializes to zero all local scalar variables of intrinsic type INTEGER, REAL, COMPLEX, or LOGICAL that are saved but not yet initialized. The default is -zero-. Use -save on the command line to make all local variables specifically marked as SAVE. -Zp[n] Aligns fields of records and components of derived types on the smaller of the size boundary specified or the boundary that will naturally align them (same as the -align recbyte option). The [n] can be 1, 2, 4, 8, or 16. If you do not specify [n], you get -Zp16, which is the default. EXAMPLES The following examples demonstrate optimization and multi­ ple input files: 1) ifort ax.f90 This command compiles ax.f90 producing executable file a.out. Optimizations occur by default. 2) ifort -o abc ax.f90 bx.f90 cx.f90 -ipo This command uses option -o to name the executable file abc and compiles ax.f90, bx.f90, and cx.f90 as one program with interprocedural analysis. 3) ifort -c ax.f90 bx.f90 cx.f90 This command uses option -c to suppress linking and produce individual object files ax.o, bx.o, and cx.o. Interprocedural optimizations are prevented. 4) ifort -c -O1 sub2.f90 ifort -c -O1 sub3.f90 ifort -o main.exe -g -O0 main.f90 sub2.o sub3.o The first two commands show incremental compilation with minimal optimization. The first command generates an object file for sub2; the second generates an object file for sub3. The last command uses option -O0 to disable all com­ piler default optimizations. It uses option -g to gen­ erate symbolic debugging information and line numbers in the object code, which can be used by a source- level debugger. RESTRICTIONS The standard libraries, like libc, are loaded with the -l loader option and not a full pathname. The wrong library can be loaded if there are files with names like libc.a or libc.so in the directories specified with the -L loader option, or in the default directories searched by the loader. For ifort, when multiple source files are compiled together without the -c option, the first source file's suffix determines the default source form for the entire compilation. DIAGNOSTICS The ifort command produces diagnostic messages that are intended to be self-explanatory. The loader can also pro­ duce occasional messages. PREDEFINED SYMBOLS The driver defines symbols (or macros) at the start of compilation and when certain compiler options are speci­ fied during compilation. The following symbols are defined at the start of compila­ tion: __INTEL_COMPILER=900 __INTEL_COMPILER_BUILD_DATE= __ELF__ __unix__ __unix __linux__ __linux __gnu_linux__ unix linux __i386 (i32 only) __i386__ (i32 only) i386 (i32 only) __ia64 (i64 only) __ia64__ (i64 only) ia64 (i64 only) The following symbols are defined during compilation if certain compiler options are specified: _OPENMP=200011 Defined if option -openmp is specified. _PGO_INSTRUMENT Defined if option -prof-gen is specified. ENVIRONMENT VARIABLES You can customize your environment by using run-time and compile-time environment variables, or by using OpenMP or Profile Guided Optimization (PGO) environment variables. Run-Time Environment Variables The following are run-time environment variables. For more information on these variables, see your user's guide. decfort_dump_flag If set to Y or y, a core dump will be taken when any severe Fortran run-time error occurs. FORTn The file specification for the file opened on unit n. FOR_READ FOR_ACCEPT FOR_PRINT FOR_TYPE For programs compiled with the -vms option, the name of the file to use for the specified opera­ tion. FOR_DISABLE_STACK_TRACE If set to true, traceback output is disabled. FORT_BUFFERED If set to true, buffered I/O will be used for sequential output to all I/O units, except those whose output is to the terminal. This provides a run-time mechanism to support the behavior enabled by the -assume buffered_io option. FORT_CONVERTn The -convert option to use on a file opened on unit n. FORT_CONVERT.ext FORT_CONVERT_ext The -convert option to be used when a filename with suffix ext is opened. F_UFMTENDIAN I/O unit numbers to perform -convert on. NLSPATH The path for the Intel Fortran run-time error mes­ sage catalog. TMPDIR The alternate working directory where scratch files are created. Compile-Time Environment Variables The following are compile-time environment variables. For more information on these variables, see your user's guide. FPATH The path for include files. IFORTCFG The configuration file to use instead of the default configuration file. LD_LIBRARY_PATH The path for shared (.so) library files. PATH The path for compiler executable files. TMPDIR The alternate working directory where scratch files are created. Standard OpenMP Environment Variables The following are standard OpenMP environment variables. For more information on these variables, see your user's guide. OMP_DYNAMIC Enables (TRUE) or disables (FALSE) the dynamic adjustment of the number of threads. The default value is FALSE. OMP_NESTED Enables (TRUE) or disables (FALSE) nested paral­ lelism. The default value is FALSE. OMP_NUM_THREADS Sets the number of threads to use during execution. The default is the number of processors currently installed in the system while generating the exe­ cutable. OMP_SCHEDULE Specifies the type of run-time scheduling. The default is static scheduling. Intel(R) Extensions to OpenMP Environment Variables The following are Intel extensions to the OpenMP environ­ ment variables. For more information on these variables, see your user's guide. KMP_LIBRARY Selects the OpenMP run-time library throughput. The options for the variable value are: serial, turnaround, or throughput indicating the execution mode. The default value is throughput. KMP_STACKSIZE Sets the number of bytes to allocate for each par­ allel thread to use as its private stack. Use the optional suffix b, k, m, g, or t, to specify bytes, kilobytes, megabytes, gigabytes, or terabytes. The default on IA-32 systems is 2m; the default on Ita­ nium(R)-based systems is 4m. Profile Guided Optimization Environment Variables The following are Profile Guided Optimization (PGO) envi­ ronment variables. For more information on these vari­ ables, see your user's guide. PROF_DIR Specifies the directory in which dynamic informa­ tion files are created. This variable applies to all three phases of the profiling process. PROF_DUMP_INTERVAL Initiates Interval Profile Dumping in an instru­ mented application. Normally, the _PGOPTI_Set_Interval_Prof_Dump(int interval) func­ tion activates Interval Profile Dumping and sets the approximate frequency at which dumps will occur. The interval parameter is measured in mil­ liseconds and specifies the time interval at which profile dumping will occur. You can use this environment variable as an alter­ native method of initiating Interval Profile Dump­ ing. Set it to the desired interval value before starting the application. PROF_NO_CLOBBER Alters the feedback compilation phase slightly. By default, during the feedback compilation phase, the compiler merges the data from all dynamic informa­ tion files and creates a new pgopti.dpi file if .dyn files are newer than an existing pgopti.dpi file. When this variable is set, the compiler does not overwrite the existing pgopti.dpi file. Instead, the compiler issues a warning and you must remove the pgopti.dpi file if you want to use additional dynamic information files. TECHNICAL SUPPORT The Intel Fortran Compiler product web site offers timely and comprehensive product information, including product features, white papers, and technical articles. For the latest information, visit http://developer.intel.com/soft­ ware/products/. Intel also provides a support web site that contains a rich repository of self-help information, including get­ ting started tips, known product issues, product errata, license information, user forums, and more. Registering your product entitles you to one year of tech­ nical support and product updates through Intel(R) Premier Support. Intel Premier Support is an interactive issue management and communication web site that enables you to submit issues and review their status, and to download product updates anytime of the day. To register your product, to contact Intel, or to seek product support, please visit: http://www.intel.com/soft­ ware/products/support. SEE ALSO icc(1), gprof(1), ld(1) The Intel(R) Fortran Building Applications guide and the Intel Fortran Optimizing Applications guide are the definitive sources for detailed information on using the Intel Fortran Compiler. In addition, see these other documents provided with the Intel Fortran Compiler: ⋠Product Release Notes ⋠The Intel Fortran Compiler Options guide ⋠The Intel Fortran Language Reference ⋠The Intel Fortran Libraries Reference You can access these documents from /doc/doc_index.htm (the default path is /opt/intel/fc/9.0/doc/doc_index.htm). COPYRIGHT INFORMATION Copyright (C) 1985-2005, Intel Corporation. All rights reserved. * Other brands and names are the property of their respec­ tive owners. Copyright(C) 2005 Intel Corporation ifort(1) dplace - a tool for controlling placement of processes onto cpus -c Cpu_number(s). Specified as a list of cpu ranges: Example: "-c1", "-c2-4", "-c1,4-8,3". Cpu numbers are NOT physical cpu numbers. They are logical cpu number that are relative to the cpus that are in the set of allowed cpus as specified by the current cpumemset or "runon" command. Cpu numbers start at 0. If this option is not specified, all cpus of the current cpumemset are available. -e Exact placement. As proceses are created, they are bound to cpus in the exact order that the cpus are specified in the cpu list. Cpu numbers may appear multiple times in the list. A cpu value of "x", in the argument list for -c option, indicates that binding should not be done for that the process. If the end of the list is reached, binding starts over at the beginning of the list. -x Provides the ability to skip placement of processes. is a bitmask. If bit N of is set, then the N+1th process that is forked is not placed. For example, setting the mask to 6 will cause the 2nd & 3rd processes from being placed. The first process (the process named by the ) will be assigned to the first cpu. The second & third processes are not placed. The fourth process is assigned to the second cpu, etc.. This option is useful for certain classes of threaded apps that spawn a few helper processes that are not typically do not use much cpu time. (Hint: Intel OpenMP applications currently should be placed using -x2. This could change in future versions of OpenMP). Example: dplace -e -c0,x,1-3 a.out where a.out is an OpenMP application that has to be run on 4 processors. Pin thread 1 on cpu 0, skip the Intel OpenMP compiler helper thread, and bind the next 3 threads on cpus 1,2,3 respectively. Alternatively, the same functionality could be achieved by issuing 'dplace -e -x2 -c0-3 a.out'. OpenMP Environment variables (settings are case-insensitive): OMP_NUM_THREADS Sets the number of threads to use during execution. Default is the number of processors. OMP_DYNAMIC Enables (true) or disables (false) the dynamic adjustment of the number of threads. Default is false. KMP_LIBRARY Selects the OpenMP runtime library throughput. The options for the variable value are: serial, turnaround, or throughput indicating the execution mode. The default value of throughput is used if this variable is not specified. The turnaround mode is designed to keep active all of the processors involved in the parallel computation in order to minimize the execution time of a single job. In this mode, the worker threads actively wait for more parallel work, without yielding to other threads. KMP_STACKSIZE Sets the number of bytes to allocate for each parallel thread to use as its private stack. Use the optional suffix b, k, m, g, or t, to specify bytes, kilobytes, megabytes, gigabytes, or terabytes. Default on Itanium compiler: 4m KMP_SCHEDULE static,balanced For loops running with OpenMP schedule "static", assign each thread approximately [#iterations / #threads] iterations. This method balances the number of iterations more evenly among all the threads. static,greedy (DEFAULT) For loops running with OpenMP schedule "static", assign each thread approximately [#iterations / #threads] iterations. KMP_SCHEDULE static,greedy (DEFAULT) For loops running with OpenMP schedule "static", this results in (#iterations/#threads) iterations--rounded to the next higher integer--being allocated to most threads, but the final thread(s) may be allocated much fewer, or even zero, iterations. static,balanced For loops running with OpenMP schedule "static", this results in (#iterations/#threads) iterations--rounded to the next lower integer--being allocated to most threads, with at most one additional iteration being allocated to some threads. Although the largest number of iterations assigned to any thread remains the same, this results in a more even sharing of iterations between threads, which may sometimes lead to a performance improvement relative to the default static thread distribution. For example, consider a loop of 9 iterations running on 4 threads: KMP_SCHEDULE Number of iterations Thread 0 Thread 1 Thread 2 Thread 3 "static,greedy" 3 3 3 0 "static,balanced" 3 2 2 2 MPI Environment variables: MPI_COMM_SIZE - determines the size of the group associated with a communicator. MPI_DSM_CPULIST - allows you to manually select processors to use for MPI application. MPI_DSM_MUSTRUN - ensures that each MPI process will get a physical CPU & memory on the node to which it was assigned. Others: mpirun - command to run MPI programs limit [ resource [ max-use ] ] Limit the consumption by the current process or any process it spawns, each not to exceed max-use on the specified resource. If max-use is omitted, print the current limit; if resource is omitted, display all limits. resource may include: stacksize Maximum stack size for the process.