CPU2017 Flag Description
Fujitsu PRIMERGY GX2460 M1, AMD EPYC 7402, 2.80 GHz

Base Optimization Flags

C benchmarks

• • -flto
  • 
  • COPTIMIZE, LDFLAGS

  • Generate output files in LLVM formats suitable for link time optimization. When used with -S this generates LLVM intermediate language assembly files, otherwise this generates LLVM bitcode format object files (which may be passed to the linker depending on the stage selection options).

• • -Wl,-mllvm -Wl,-function-specialize
  • 
  • LDFLAGS

  • This option enables an optimization that generates and calls specialized function versions when they are called with constant arguments. This optimization helps in function inlining.

• • -Wl,-mllvm -Wl,-region-vectorize
  • 
  • LDFLAGS

  • This flag enables vectorization of loops with complex control flow that can not be vectorized by loop and slp vectorizers.

• • -Wl,-mllvm -Wl,-vector-library=LIBMVEC
  • 
  • LDFLAGS

  • This option instructs the compiler to use vector math library.

• • -Wl,-mllvm -Wl,-reduce-array-computations=3
  • 
  • LDFLAGS

  • This option eliminates the array computations based on their usage. The computations on unused array elements and computations on zero valued array elements are eliminated with this optimization. -flto as whole program analysis is required to perform this optimization.

    Possible values:

    • 1: Eliminates the computations on unused array elements
    • 2: Eliminates the computations on zero valued array elements
    • 3: Eliminates the computations on unused and zero valued array elements
• • -O3
  • 
  • COPTIMIZE

  • Like -O2, except that it enables optimizations that take longer to perform or that may generate larger code (in an attempt to make the program run faster).

    If multiple "O" options are used, with or without level numbers, the last such option is the one that is effective.

  • Includes:
    • -O2
      • -O1
• • -ffast-math
  • 
  • COPTIMIZE

  • Enables a range of optimizations that provide faster, though sometimes less precise, mathematical operations that may not conform to the IEEE-754 specifications. When this option is specified, the __STDC_IEC_559__ macro is ignored even if set by the system headers.

• • -march=znver2
  • 
  • COPTIMIZE

  • Specify that Clang should generate code for a specific processor family member and later. For example, if you specify -march=znver1, the compiler is allowed to generate instructions that are valid on AMD Zen processors, but which may not exist on earlier products.

• • -fstruct-layout=3
  • 
  • COPTIMIZE

  • This option enables transformation of the layout of arrays of structure types and their fields to improve the cache locality. Aggressive analysis and transformations are performed at higher levels. This option is effective only with -flto as whole program analysis is required to perform this optimization.

    Possible values:

    • fstruct-layout=1: enables structure peeling.
    • fstruct-layout=2: enables structure peeling and selectively compresses self-referential pointers in these structures to 32-bit pointers.
    • fstruct-layout=3: enables structure peeling and selectively compresses self-referential pointers in these structures to 16-bit pointers if the values of the addresses are such that 16-bit pointers are sufficient to hold these addresses. The compression to 16-bit pointers is done only if it is safe to do so.
    • fstruct-layout=4: enables structure peeling, pointer compression as in level 2 and further enables compression of structure fields which are of type 64-bit "signed int" or "unsigned int". The user needs to ensure that the values assigned to 64-bit "signed int" fields are in range –(2^31 – 1) to (2^31 -1) and 64-bit "unsigned int" fields are in range 0 to (2^31 -1); otherwise, incorrect results may be obtained. This is done without performing safety analysis and the user needs to enforce this based on the program compiled.
    • fstruct-layout=5: enables structure peeling, pointer compression as in level 3 and further enables compression of structure fields which are of type 64-bit "signed int" or "unsigned int". The user needs to ensure that the values assigned to 64-bit "signed int" fields are in range –(2^31 – 1) to (2^31 -1) and 64-bit "unsigned int" fields are in range 0 to (2^31 -1); otherwise, incorrect results may be obtained. This is done without performing safety analysis and the user needs to enforce this based on the program compiled.
    • Note: The distinction between -fstruct-layout=4 and 5 is one which is inherited from the difference between -fstruct-layout=2 and 3. "-fstruct-layout=4" is same as "-fstruct-layout=2" with the added feature of compressing structure data fields from 64 bits to 32 bits. "-fstruct-layout=5" is same as "-fstruct-layout=3" with the added feature of compressing structure data field from 64 bits to 32 bits. Going from -fstruct-layout=4 to -fstruct-layout=5 may result in higher performance if the pointer values are such that the pointers can be compressed to 16 bits.
• • -mllvm -unroll-threshold=50
  • 
  • COPTIMIZE

  • Sets the limit at which loops will be unrolled. For example, if unroll threshold is set to 100 then only loops with 100 or fewer instructions will be unrolled.

• • -fremap-arrays
  • 
  • COPTIMIZE

  • This option enables an optimization that transforms the data layout of a single dimensional array to provide better cache locality by analysing the access patterns.

• • -mllvm -function-specialize
  • 
  • COPTIMIZE

  • This option enables an optimization that generates and calls specialized function versions when they are called with constant arguments. This optimization helps in function inlining.

• • -mllvm -enable-gvn-hoist
  • 
  • COPTIMIZE
  • >

    This option enables the GVN hoist pass, which is used to hoist computations from branches.

• • -mllvm -reduce-array-computations=3
  • 
  • COPTIMIZE

  • This option eliminates the array computations based on their usage. The computations on unused array elements and computations on zero valued array elements are eliminated with this optimization. -flto as whole program analysis is required to perform this optimization.

    Possible values:

    • 1: Eliminates the computations on unused array elements
    • 2: Eliminates the computations on zero valued array elements
    • 3: Eliminates the computations on unused and zero valued array elements
• • -mllvm -global-vectorize-slp
  • 
  • COPTIMIZE

  • This option enables an optimization that does the slp vectorization across basic blocks. The SLP vectorizer vectorizes instructions within basic blocks. The global slp vectorizer analyzes instructions across basic blocks and vectorizes them.

• • -mllvm -vector-library=LIBMVEC
  • 
  • COPTIMIZE

  • This option instructs the compiler to use vector math library.

• • -mllvm -inline-threshold=1000
  • 
  • COPTIMIZE

  • Sets the compiler's inlining threshold level to the value passed. The inline threshold is used in the inliner heuristics to decide which functions should be inlined.

• • -flv-function-specialization
  • 
  • COPTIMIZE

  • This option enables an optimization that generates and calls specialized function versions when the loops inside function are vectorizable and the arguments are not aliased with each other. This optimization helps in function inlining and vectorization.

• • -z muldefs
  • 
  • LDOPTIMIZE

  • Instructs the linker to use the first definition encountered for a symbol, and ignore all others.

• • -lmvec
  • 
  • EXTRA_LIBS

  • Instructs the compiler to link with system vector math libraries.

• • -lamdlibm
  • 
  • EXTRA_LIBS

  • Instructs the compiler to link with AMD-supported optimized math library.

• • -ljemalloc
  • 
  • EXTRA_LIBS

  • Use the jemalloc library, which is a general purpose malloc(3) implementation that emphasizes fragmentation avoidance and scalable concurrency support.

• • -lflang
  • 
  • EXTRA_LIBS

  • Instructs the compiler to link with flang Fortran runtime libraries.

C++ benchmarks

• • -flto
  • 
  • CXXOPTIMIZE, LDFLAGS

  • Generate output files in LLVM formats suitable for link time optimization. When used with -S this generates LLVM intermediate language assembly files, otherwise this generates LLVM bitcode format object files (which may be passed to the linker depending on the stage selection options).

• • -Wl,-mllvm -Wl,-function-specialize
  • 
  • LDFLAGS

  • This option enables an optimization that generates and calls specialized function versions when they are called with constant arguments. This optimization helps in function inlining.

• • -Wl,-mllvm -Wl,-region-vectorize
  • 
  • LDFLAGS

  • This flag enables vectorization of loops with complex control flow that can not be vectorized by loop and slp vectorizers.

• • -Wl,-mllvm -Wl,-vector-library=LIBMVEC
  • 
  • LDFLAGS

  • This option instructs the compiler to use vector math library.

• • -Wl,-mllvm -Wl,-reduce-array-computations=3
  • 
  • LDFLAGS

  • This option eliminates the array computations based on their usage. The computations on unused array elements and computations on zero valued array elements are eliminated with this optimization. -flto as whole program analysis is required to perform this optimization.

    Possible values:

    • 1: Eliminates the computations on unused array elements
    • 2: Eliminates the computations on zero valued array elements
    • 3: Eliminates the computations on unused and zero valued array elements
• • -Wl,-mllvm -Wl,-suppress-fmas
  • 
  • LDCXXFLAGS

  • Disable generation of fma instructions when there is a chain of fma instructions and output of one fma instruction is used as input to other fma instruction.

• • -O3
  • 
  • CXXOPTIMIZE

  • Like -O2, except that it enables optimizations that take longer to perform or that may generate larger code (in an attempt to make the program run faster).

    If multiple "O" options are used, with or without level numbers, the last such option is the one that is effective.

  • Includes:
    • -O2
      • -O1
• • -ffast-math
  • 
  • CXXOPTIMIZE

  • Enables a range of optimizations that provide faster, though sometimes less precise, mathematical operations that may not conform to the IEEE-754 specifications. When this option is specified, the __STDC_IEC_559__ macro is ignored even if set by the system headers.

• • -march=znver2
  • 
  • CXXOPTIMIZE

  • Specify that Clang should generate code for a specific processor family member and later. For example, if you specify -march=znver1, the compiler is allowed to generate instructions that are valid on AMD Zen processors, but which may not exist on earlier products.

• • -mllvm -loop-unswitch-threshold=200000
  • 
  • CXXOPTIMIZE

  • Sets the limit at which loops will be unswitched. For example, if unswitch threshold is set to 100 then only loops with 100 or fewer instructions will be unswtched.

• • -mllvm -vector-library=LIBMVEC
  • 
  • CXXOPTIMIZE

  • This option instructs the compiler to use vector math library.

• • -mllvm -unroll-threshold=100
  • 
  • CXXOPTIMIZE

  • Sets the limit at which loops will be unrolled. For example, if unroll threshold is set to 100 then only loops with 100 or fewer instructions will be unrolled.

• • -flv-function-specialization
  • 
  • CXXOPTIMIZE

  • This option enables an optimization that generates and calls specialized function versions when the loops inside function are vectorizable and the arguments are not aliased with each other. This optimization helps in function inlining and vectorization.

• • -mllvm -enable-partial-unswitch
  • 
  • CXXOPTIMIZE
  • >

    This optimization does partial unswitching of loops where some part of the unswitched control flow remains in the loop.

• • -z muldefs
  • 
  • LDOPTIMIZE

  • Instructs the linker to use the first definition encountered for a symbol, and ignore all others.

• • -lmvec
  • 
  • EXTRA_LIBS

  • Instructs the compiler to link with system vector math libraries.

• • -lamdlibm
  • 
  • EXTRA_LIBS

  • Instructs the compiler to link with AMD-supported optimized math library.

• • -ljemalloc
  • 
  • EXTRA_LIBS

  • Use the jemalloc library, which is a general purpose malloc(3) implementation that emphasizes fragmentation avoidance and scalable concurrency support.

• • -lflang
  • 
  • EXTRA_LIBS

  • Instructs the compiler to link with flang Fortran runtime libraries.

Fortran benchmarks

• • -flto
  • 
  • FOPTIMIZE, LDFLAGS

  • Generate output files in LLVM formats suitable for link time optimization. When used with -S this generates LLVM intermediate language assembly files, otherwise this generates LLVM bitcode format object files (which may be passed to the linker depending on the stage selection options).

• • -Wl,-mllvm -Wl,-function-specialize
  • 
  • LDFLAGS

  • This option enables an optimization that generates and calls specialized function versions when they are called with constant arguments. This optimization helps in function inlining.

• • -Wl,-mllvm -Wl,-region-vectorize
  • 
  • LDFLAGS

  • This flag enables vectorization of loops with complex control flow that can not be vectorized by loop and slp vectorizers.

• • -Wl,-mllvm -Wl,-vector-library=LIBMVEC
  • 
  • LDFLAGS

  • This option instructs the compiler to use vector math library.

• • -Wl,-mllvm -Wl,-reduce-array-computations=3
  • 
  • LDFLAGS

  • This option eliminates the array computations based on their usage. The computations on unused array elements and computations on zero valued array elements are eliminated with this optimization. -flto as whole program analysis is required to perform this optimization.

    Possible values:

    • 1: Eliminates the computations on unused array elements
    • 2: Eliminates the computations on zero valued array elements
    • 3: Eliminates the computations on unused and zero valued array elements
• • -ffast-math
  • 
  • EXTRA_FFLAGS, LDFFLAGS

  • Enables a range of optimizations that provide faster, though sometimes less precise, mathematical operations that may not conform to the IEEE-754 specifications. When this option is specified, the __STDC_IEC_559__ macro is ignored even if set by the system headers.

• • -Wl,-mllvm -Wl,-inline-recursion=4
  • 
  • LDFFLAGS

  • Enables inlining for recursive functions based on heuristics, with level 4 being most aggressive. Higher levels may lead to code bloat due to expansion of recursive functions at call sites.

    Levels:

    • 0 [DEFAULT]: Disables inlining for recursive functions.
    • 1: Enables inlining for recursive functions using heuristics with inline depth 1.
    • 2: Same as level 1 but with more aggressive heuristics.
    • 3: Enables inlining for all recursive functions with inline depth 1.
    • 4: Enables inlining for all recursive function with inline depth 10.
• • -Wl,-mllvm -Wl,-lsr-in-nested-loop
  • 
  • LDFFLAGS

  • Enables loop strength reduction for nested loop structures. By default, the compiler will do loop strength reduction only for the innermost loop.

• • -Wl,-mllvm -Wl,-enable-iv-split
  • 
  • LDFFLAGS

  • Enables splitting of long live ranges of loop induction variables which span loop boundaries. This helps reduce register pressure and can help avoid needless spills to memory and reloads from memory.

• • -O3
  • 
  • FOPTIMIZE

  • Like -O2, except that it enables optimizations that take longer to perform or that may generate larger code (in an attempt to make the program run faster).

    If multiple "O" options are used, with or without level numbers, the last such option is the one that is effective.

  • Includes:
    • -O2
      • -O1
• • -march=znver2
  • 
  • FOPTIMIZE

  • Specify that Clang should generate code for a specific processor family member and later. For example, if you specify -march=znver1, the compiler is allowed to generate instructions that are valid on AMD Zen processors, but which may not exist on earlier products.

• • -funroll-loops
  • 
  • FOPTIMIZE

  • This option instructs the compiler to unroll loops wherever possible.

• • -Mrecursive
  • 
  • FOPTIMIZE

  • Allocate local variables on the stack, thus allowing recursion. SAVEd, data-initialized, or namelist members are always allocated statically, regardless of the setting of this switch.

• • -mllvm -vector-library=LIBMVEC
  • 
  • FOPTIMIZE

  • This option instructs the compiler to use vector math library.

• • -z muldefs
  • 
  • LDOPTIMIZE

  • Instructs the linker to use the first definition encountered for a symbol, and ignore all others.

• • -mllvm -disable-indvar-simplify
  • 
  • EXTRA_FFLAGS
  • >

    The induction variable simplification optimization transforms induction variables to simpler forms. The option disables this optimization.

• • -mllvm -unroll-aggressive
  • 
  • EXTRA_FFLAGS

  • Enables aggressive heuristics to get loop unrolling.

• • -mllvm -unroll-threshold=150
  • 
  • EXTRA_FFLAGS

  • Sets the limit at which loops will be unrolled. For example, if unroll threshold is set to 100 then only loops with 100 or fewer instructions will be unrolled.

• • -lmvec
  • 
  • EXTRA_FLIBS, EXTRA_LIBS

  • Instructs the compiler to link with system vector math libraries.

• • -lamdlibm
  • 
  • EXTRA_FLIBS, EXTRA_LIBS

  • Instructs the compiler to link with AMD-supported optimized math library.

• • -ljemalloc
  • 
  • EXTRA_LIBS

  • Use the jemalloc library, which is a general purpose malloc(3) implementation that emphasizes fragmentation avoidance and scalable concurrency support.

• • -lflang
  • 
  • EXTRA_LIBS

  • Instructs the compiler to link with flang Fortran runtime libraries.