SPEC® CPU2017 Floating Point Rate Result

Copyright 2017-2018 Standard Performance Evaluation Corporation

Hewlett Packard Enterprise (Test Sponsor: HPE)

ProLiant DL325 Gen10
(2.00 GHz, AMD EPYC 7401P)

SPECrate2017_fp_base = 11500

SPECrate2017_fp_peak = 11400

CPU2017 License: 3 Test Date: Jul-2017
Test Sponsor: HPE Hardware Availability: Jul-2018
Tested by: HPE Software Availability: Feb-2018

Benchmark result graphs are available in the PDF report.

Hardware
CPU Name: AMD EPYC 7401P
  Max MHz.: 3000
  Nominal: 2000
Enabled: 24 cores, 1 chip, 2 threads/core
Orderable: 1 chip
Cache L1: 64 KB I + 32 KB D on chip per core
  L2: 512 KB I+D on chip per core
  L3: 64 MB I+D on chip per chip, 8 MB shared / 3 cores
  Other: None
Memory: 512 GB (8 x 64 GB 4Rx4 PC4-2666V-L,
running at 2400)
Storage: 1 x 400 GB SATA SSD, RAID 0
Other: None
Software
OS: SUSE Linux Enterprise Server 12 (x86_64) SP3
Kernel 4.4.131-94.25-default
Compiler: C/C++: Version 1.0.0 of AOCC
Fortran: Version 4.8.2 of GCC
Parallel: No
Firmware: HPE BIOS Version A41 04/06/2018 released Apr-2018
File System: xfs
System State: Run level 3 (multi-user)
Base Pointers: 64-bit
Peak Pointers: 64-bit
Other: jemalloc general purpose malloc implementation
v4.5.0;

Results Table

Benchmark Base Peak
Copies Seconds Ratio Seconds Ratio Seconds Ratio Copies Seconds Ratio Seconds Ratio Seconds Ratio
SPECrate2017_fp_base 11500
SPECrate2017_fp_peak 11400
Results appear in the order in which they were run. Bold underlined text indicates a median measurement.
503.bwaves_r 48 1532 3140 1531 3140 1532 3140 24 756 3180 760 3170 756 3180
507.cactuBSSN_r 48 514 1180 512 1190 513 1180 48 522 1160 519 1170 521 1170
508.namd_r 48 488 93.4 492 92.7 488 93.5 48 446 1020 446 1020 446 1020
510.parest_r 48 1586 79.2 1585 79.2 1584 79.3 24 772 81.4 771 81.5 771 81.4
511.povray_r 48 941 1190 944 1190 943 1190 48 913 1230 908 1230 914 1230
519.lbm_r 48 660 76.6 662 76.4 661 76.6 24 336 75.3 345 73.2 348 72.7
521.wrf_r 48 854 1260 853 1260 859 1250 24 542 99.2 543 99.1 547 98.3
526.blender_r 48 555 1320 554 1320 554 1320 48 556 1310 557 1310 558 1310
527.cam4_r 48 733 1150 733 1140 732 1150 48 734 1140 734 1140 735 1140
538.imagick_r 48 691 1730 692 1730 689 1730 48 672 1780 682 1750 681 1750
544.nab_r 48 539 1500 536 1510 537 1510 48 536 1510 535 1510 537 1510
549.fotonik3d_r 48 1963 95.3 1960 95.5 1976 94.7 24 1043 89.6 1040 89.9 1042 89.7
554.roms_r 48 1408 54.2 1406 54.2 1407 54.2 24 656 58.1 655 58.3 658 58.0

Submit Notes

The config file option 'submit' was used.
'numactl' was used to bind copies to the cores.
See the configuration file for details.

Operating System Notes

'ulimit -s unlimited' was used to set environment stack size
'ulimit -l 2097152' was used to set environment locked pages in memory limit

runspec command invoked through numactl i.e.:
numactl --interleave=all runspec <etc>

Set dirty_ratio=8 to limit dirty cache to 8% of memory
Set swappiness=1 to swap only if necessary
Set zone_reclaim_mode=1 to free local node memory and avoid remote memory
sync then drop_caches=3 to reset caches before invoking runcpu
Linux governor set to performance with cpupower "cpupower frequency-set -r -g performance"
dirty_ratio, swappiness, zone_reclaim_mode and drop_caches were
all set using privileged echo (e.g. echo 1 > /proc/sys/vm/swappiness).
Transparent huge pages were enabled for this run (OS default)
Huge pages were not configured for this run.
IRQ balance service stopped using "systemctl stop irqbalance.service"
Tuned profile set with "tuned-adm profile throughput-performance"

General Notes

Environment variables set by runcpu before the start of the run:
LD_LIBRARY_PATH = "/home/cpu2017/amd1704-rate-libs-revC/64;/home/cpu2017/amd1704-rate-libs-revC/32:"
MALLOC_CONF = "lg_chunk:28"

The AMD64 AOCC Compiler Suite is available at
http://developer.amd.com/amd-aocc/

Binaries were compiled on a system with 2x AMD EPYC 7601 CPU + 512GB Memory using RHEL 7.4


The AOCC Gold Linker plugin was installed and used for the link stage.

The AOCC Fortran Plugin version 1.0 was used to leverage AOCC optimizers
with gfortran. It is available here:
http://developer.amd.com/amd-aocc/

 NA: The test sponsor attests, as of date of publication, that CVE-2017-5754 (Meltdown)
 is mitigated in the system as tested and documented.
 Yes: The test sponsor attests, as of date of publication, that CVE-2017-5753 (Spectre variant 1)
 is mitigated in the system as tested and documented.
 Yes: The test sponsor attests, as of date of publication, that CVE-2017-5715 (Spectre variant 2)
 is mitigated in the system as tested and documented.

jemalloc, a general purpose malloc implementation, was obtained at
https://github.com/jemalloc/jemalloc/releases/download/4.5.0/jemalloc-4.5.0.tar.bz2
jemalloc was built with GCC v4.8.2 in RHEL v7.2 under default conditions.
jemalloc uses environment variable MALLOC_CONF with values narenas and lg_chunk:
  narenas: sets the maximum number of arenas to use for automatic multiplexing
           of threads and arenas.
  lg_chunk: set the virtual memory chunk size (log base 2). For example,
            lg_chunk:21 sets the default chunk size to 2^21 = 2MiB.

Platform Notes

 BIOS Configuration:
  Thermal Configuration set to Maximum Cooling
  Memory Patrol Scrubbing set to Disabled
  Performance Determinism set to Power Deterministic
  Processor Power and Utilization Monitoring set to Disabled
  Workload Profile set to General Throughput Compute
    Minimum Processor Idle Power Core C-State set to C6 State
 Sysinfo program /cpu2017/bin/sysinfo
 Rev: r5974 of 2018-05-19 9bcde8f2999c33d61f64985e45859ea9
 running on linux-6qav Sun Jul 16 18:31:12 2017

 SUT (System Under Test) info as seen by some common utilities.
 For more information on this section, see
    https://www.spec.org/cpu2017/Docs/config.html#sysinfo

 From /proc/cpuinfo
    model name : AMD EPYC 7401P 24-Core Processor
       1  "physical id"s (chips)
       48 "processors"
    cores, siblings (Caution: counting these is hw and system dependent. The following
    excerpts from /proc/cpuinfo might not be reliable.  Use with caution.)
       cpu cores : 24
       siblings  : 48
       physical 0: cores 0 1 2 4 5 6 8 9 10 12 13 14 16 17 18 20 21 22 24 25 26 28 29 30

 From lscpu:
      Architecture:          x86_64
      CPU op-mode(s):        32-bit, 64-bit
      Byte Order:            Little Endian
      CPU(s):                48
      On-line CPU(s) list:   0-47
      Thread(s) per core:    2
      Core(s) per socket:    24
      Socket(s):             1
      NUMA node(s):          4
      Vendor ID:             AuthenticAMD
      CPU family:            23
      Model:                 1
      Model name:            AMD EPYC 7401P 24-Core Processor
      Stepping:              2
      CPU MHz:               2000.000
      CPU max MHz:           2000.0000
      CPU min MHz:           1200.0000
      BogoMIPS:              3992.46
      Virtualization:        AMD-V
      L1d cache:             32K
      L1i cache:             64K
      L2 cache:              512K
      L3 cache:              8192K
      NUMA node0 CPU(s):     0-5,24-29
      NUMA node1 CPU(s):     6-11,30-35
      NUMA node2 CPU(s):     12-17,36-41
      NUMA node3 CPU(s):     18-23,42-47
      Flags:                 fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov
      pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm
      constant_tsc rep_good nopl nonstop_tsc extd_apicid amd_dcm aperfmperf eagerfpu pni
      pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c
      rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch
      osvw skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_l2 mwaitx arat cpb
      hw_pstate rds retpoline retpoline_amd npt lbrv svm_lock nrip_save tsc_scale
      vmcb_clean flushbyasid decodeassists pausefilter pfthreshold vmmcall avic fsgsbase
      bmi1 avx2 smep bmi2 rdseed adx smap clflushopt sha_ni xsaveopt xsavec xgetbv1 clzero
      irperf ibpb overflow_recov succor smca

 /proc/cpuinfo cache data
    cache size : 512 KB

 From numactl --hardware  WARNING: a numactl 'node' might or might not correspond to a
 physical chip.
   available: 4 nodes (0-3)
   node 0 cpus: 0 1 2 3 4 5 24 25 26 27 28 29
   node 0 size: 128839 MB
   node 0 free: 128625 MB
   node 1 cpus: 6 7 8 9 10 11 30 31 32 33 34 35
   node 1 size: 129021 MB
   node 1 free: 128827 MB
   node 2 cpus: 12 13 14 15 16 17 36 37 38 39 40 41
   node 2 size: 129021 MB
   node 2 free: 128824 MB
   node 3 cpus: 18 19 20 21 22 23 42 43 44 45 46 47
   node 3 size: 129020 MB
   node 3 free: 128826 MB
   node distances:
   node   0   1   2   3
     0:  10  16  16  16
     1:  16  10  16  16
     2:  16  16  10  16
     3:  16  16  16  10

 From /proc/meminfo
    MemTotal:       528285304 kB
    HugePages_Total:       0
    Hugepagesize:       2048 kB

 /usr/bin/lsb_release -d
    SUSE Linux Enterprise Server 12 SP3

 From /etc/*release* /etc/*version*
    SuSE-release:
       SUSE Linux Enterprise Server 12 (x86_64)
       VERSION = 12
       PATCHLEVEL = 3
       # This file is deprecated and will be removed in a future service pack or release.
       # Please check /etc/os-release for details about this release.
    os-release:
       NAME="SLES"
       VERSION="12-SP3"
       VERSION_ID="12.3"
       PRETTY_NAME="SUSE Linux Enterprise Server 12 SP3"
       ID="sles"
       ANSI_COLOR="0;32"
       CPE_NAME="cpe:/o:suse:sles:12:sp3"

 uname -a:
    Linux linux-6qav 4.4.131-94.25-default #1 SMP Mon May 7 11:22:19 UTC 2018 (9700bac)
    x86_64 x86_64 x86_64 GNU/Linux

 Kernel self-reported vulnerability status:

 CVE-2017-5754 (Meltdown):          Not affected
 CVE-2017-5753 (Spectre variant 1): Mitigation: __user pointer sanitization
 CVE-2017-5715 (Spectre variant 2): Mitigation: Full AMD retpoline + IBPB

 run-level 3 Jul 16 07:59

 SPEC is set to: /cpu2017
    Filesystem     Type   Size  Used Avail Use% Mounted on
    /dev/sda3      btrfs  371G   21G  349G   6% /

 Additional information from dmidecode follows.  WARNING: Use caution when you interpret
 this section. The 'dmidecode' program reads system data which is "intended to allow
 hardware to be accurately determined", but the intent may not be met, as there are
 frequent changes to hardware, firmware, and the "DMTF SMBIOS" standard.
   BIOS HPE A41 04/06/2018
   Memory:
    8x UNKNOWN NOT AVAILABLE
    8x UNKNOWN NOT AVAILABLE 64 GB 4 rank 2666

 (End of data from sysinfo program)

Compiler Version Notes

==============================================================================
 CC  519.lbm_r(base, peak) 538.imagick_r(base, peak) 544.nab_r(base, peak)
------------------------------------------------------------------------------
AOCC.LLVM.4.0.0.B35.2017_04_26 clang version 4.0.0 (CLANG:) (based on LLVM
  AOCC.LLVM.4.0.0.B35.2017_04_26)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /root/work/compilers/AOCC-1.0-Compiler/bin
------------------------------------------------------------------------------

==============================================================================
 CXXC 508.namd_r(base, peak) 510.parest_r(base, peak)
------------------------------------------------------------------------------
AOCC.LLVM.4.0.0.B35.2017_04_26 clang version 4.0.0 (CLANG:) (based on LLVM
  AOCC.LLVM.4.0.0.B35.2017_04_26)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /root/work/compilers/AOCC-1.0-Compiler/bin
------------------------------------------------------------------------------

==============================================================================
 CC  511.povray_r(base, peak) 526.blender_r(base, peak)
------------------------------------------------------------------------------
AOCC.LLVM.4.0.0.B35.2017_04_26 clang version 4.0.0 (CLANG:) (based on LLVM
  AOCC.LLVM.4.0.0.B35.2017_04_26)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /root/work/compilers/AOCC-1.0-Compiler/bin
AOCC.LLVM.4.0.0.B35.2017_04_26 clang version 4.0.0 (CLANG:) (based on LLVM
  AOCC.LLVM.4.0.0.B35.2017_04_26)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /root/work/compilers/AOCC-1.0-Compiler/bin
------------------------------------------------------------------------------

==============================================================================
 FC  507.cactuBSSN_r(base, peak)
------------------------------------------------------------------------------
AOCC.LLVM.4.0.0.B35.2017_04_26 clang version 4.0.0 (CLANG:) (based on LLVM
  AOCC.LLVM.4.0.0.B35.2017_04_26)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /root/work/compilers/AOCC-1.0-Compiler/bin
AOCC.LLVM.4.0.0.B35.2017_04_26 clang version 4.0.0 (CLANG:) (based on LLVM
  AOCC.LLVM.4.0.0.B35.2017_04_26)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /root/work/compilers/AOCC-1.0-Compiler/bin
GNU Fortran (GCC) 4.8.2
Copyright (C) 2013 Free Software Foundation, Inc.
GNU Fortran comes with NO WARRANTY, to the extent permitted by law.
You may redistribute copies of GNU Fortran
under the terms of the GNU General Public License.
For more information about these matters, see the file named COPYING
------------------------------------------------------------------------------

==============================================================================
 FC  503.bwaves_r(base, peak) 549.fotonik3d_r(base, peak) 554.roms_r(base,
      peak)
------------------------------------------------------------------------------
GNU Fortran (GCC) 4.8.2
Copyright (C) 2013 Free Software Foundation, Inc.
GNU Fortran comes with NO WARRANTY, to the extent permitted by law.
You may redistribute copies of GNU Fortran
under the terms of the GNU General Public License.
For more information about these matters, see the file named COPYING
------------------------------------------------------------------------------

==============================================================================
 CC  521.wrf_r(base, peak) 527.cam4_r(base, peak)
------------------------------------------------------------------------------
GNU Fortran (GCC) 4.8.2
Copyright (C) 2013 Free Software Foundation, Inc.
GNU Fortran comes with NO WARRANTY, to the extent permitted by law.
You may redistribute copies of GNU Fortran
under the terms of the GNU General Public License.
For more information about these matters, see the file named COPYING
AOCC.LLVM.4.0.0.B35.2017_04_26 clang version 4.0.0 (CLANG:) (based on LLVM
  AOCC.LLVM.4.0.0.B35.2017_04_26)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /root/work/compilers/AOCC-1.0-Compiler/bin
------------------------------------------------------------------------------

Base Compiler Invocation

C benchmarks:

 clang 

C++ benchmarks:

 clang++ 

Fortran benchmarks:

 clang   gfortran 

Benchmarks using both Fortran and C:

 clang   gfortran 

Benchmarks using both C and C++:

 clang++   clang 

Benchmarks using Fortran, C, and C++:

 clang++   clang   gfortran 

Base Portability Flags

503.bwaves_r:  -DSPEC_LP64 
507.cactuBSSN_r:  -DSPEC_LP64 
508.namd_r:  -DSPEC_LP64 
510.parest_r:  -DSPEC_LP64 
511.povray_r:  -DSPEC_LP64 
519.lbm_r:  -DSPEC_LP64 
521.wrf_r:  -DSPEC_CASE_FLAG   -fconvert=big-endian   -DSPEC_LP64 
526.blender_r:  -funsigned-char   -D__BOOL_DEFINED   -DSPEC_LP64 
527.cam4_r:  -DSPEC_CASE_FLAG   -DSPEC_LP64 
538.imagick_r:  -DSPEC_LP64 
544.nab_r:  -DSPEC_LP64 
549.fotonik3d_r:  -DSPEC_LP64 
554.roms_r:  -DSPEC_LP64 

Base Optimization Flags

C benchmarks:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -disable-vect-cmp   -O3   -ffast-math   -march=znver1   -fstruct-layout=2   -mllvm   -unroll-threshold=100   -fremap-arrays   -mno-avx2   -inline-threshold=1000   -z muldefs   -ljemalloc 

C++ benchmarks:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -disable-vect-cmp   -O3   -march=znver1   -mllvm   -unroll-threshold=100   -finline-aggressive   -fremap-arrays   -inline-threshold=1000   -z muldefs   -ljemalloc 

Fortran benchmarks:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -disable-vect-cmp   -O3(gfortran)   -O3(clang)   -mavx   -madx   -funroll-loops   -ffast-math   -z muldefs   -fplugin=dragonegg.so   -fplugin-arg-dragonegg-llvm-option="-merge-constant   -disable-vect-cmp"   -ljemalloc   -lgfortran    -lamdlibm  

Benchmarks using both Fortran and C:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -disable-vect-cmp   -O3(clang)   -ffast-math   -march=znver1   -fstruct-layout=2   -mllvm   -unroll-threshold=100   -fremap-arrays   -mno-avx2   -inline-threshold=1000   -O3(gfortran)   -mavx   -madx   -funroll-loops   -z muldefs   -fplugin=dragonegg.so   -fplugin-arg-dragonegg-llvm-option="-merge-constant   -disable-vect-cmp"   -ljemalloc   -lgfortran    -lamdlibm  

Benchmarks using both C and C++:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -disable-vect-cmp   -O3   -ffast-math   -march=znver1   -fstruct-layout=2   -mllvm   -unroll-threshold=100   -fremap-arrays   -mno-avx2   -inline-threshold=1000   -finline-aggressive   -z muldefs   -ljemalloc 

Benchmarks using Fortran, C, and C++:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -disable-vect-cmp   -O3(clang)   -ffast-math   -march=znver1   -fstruct-layout=2   -mllvm   -unroll-threshold=100   -fremap-arrays   -mno-avx2   -inline-threshold=1000   -finline-aggressive   -O3(gfortran)   -mavx   -madx   -funroll-loops   -z muldefs   -fplugin=dragonegg.so   -fplugin-arg-dragonegg-llvm-option="-merge-constant   -disable-vect-cmp"   -ljemalloc 

Peak Compiler Invocation

C benchmarks:

 clang 

C++ benchmarks:

 clang++ 

Fortran benchmarks:

 clang   gfortran 

Benchmarks using both Fortran and C:

 clang   gfortran 

Benchmarks using both C and C++:

 clang++   clang 

Benchmarks using Fortran, C, and C++:

 clang++   clang   gfortran 

Peak Portability Flags

Same as Base Portability Flags

Peak Optimization Flags

C benchmarks:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -Ofast   -march=znver1   -fstruct-layout=3   -mllvm   -vectorize-memory-aggressively   -mno-avx2   -unroll-threshold=100   -fremap-arrays   -inline-threshold=1000   -ljemalloc 

C++ benchmarks:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -Ofast   -march=znver1   -finline-aggressive   -mllvm   -unroll-threshold=100   -fremap-arrays   -inline-threshold=1000   -ljemalloc 

Fortran benchmarks:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -O3(gfortran)   -O3(clang)   -mavx2   -madx   -funroll-loops   -ffast-math   -fplugin=dragonegg.so   -fplugin-arg-dragonegg-llvm-option="-merge-constant   -inline-threshold:1000"   -ljemalloc   -lgfortran    -lamdlibm  

Benchmarks using both Fortran and C:

521.wrf_r:  -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -O3(clang)   -mavx   -ffast-math   -O3(gfortran)   -funroll-loops   -fplugin=dragonegg.so   -fplugin-arg-dragonegg-llvm-option="-merge-constant   -inline-threshold:1000"   -ljemalloc   -lgfortran    -lamdlibm  
527.cam4_r:  -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -Ofast   -march=znver1   -fstruct-layout=3   -mllvm   -vectorize-memory-aggressively   -mno-avx2   -unroll-threshold=100   -fremap-arrays   -inline-threshold=1000   -O3(gfortran)   -O3(clang)   -mavx2   -madx   -funroll-loops   -ffast-math   -fplugin=dragonegg.so   -fplugin-arg-dragonegg-llvm-option="-merge-constant   -inline-threshold:1000"   -ljemalloc   -lgfortran    -lamdlibm  

Benchmarks using both C and C++:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -Ofast   -march=znver1   -fstruct-layout=3   -mllvm   -vectorize-memory-aggressively   -mno-avx2   -unroll-threshold=100   -fremap-arrays   -inline-threshold=1000   -finline-aggressive   -ljemalloc 

Benchmarks using Fortran, C, and C++:

 -flto   -Wl,-plugin-opt=   -merge-constant   -Wl,-plugin-opt=-lsr-in-nested-loop   -Ofast   -march=znver1   -fstruct-layout=3   -mllvm   -vectorize-memory-aggressively   -mno-avx2   -unroll-threshold=100   -fremap-arrays   -inline-threshold=1000   -finline-aggressive   -O3   -mavx2   -madx   -funroll-loops   -ffast-math   -fplugin=dragonegg.so   -fplugin-arg-dragonegg-llvm-option="-merge-constant   -inline-threshold:1000"   -ljemalloc 

The flags files that were used to format this result can be browsed at
http://www.spec.org/cpu2017/flags/gcc.2017-11-20.html,
http://www.spec.org/cpu2017/flags/aocc100-flags-revC-I.html,
http://www.spec.org/cpu2017/flags/HPE-Platform-Flags-AMD-V1.2-EPYC-revD.html.

You can also download the XML flags sources by saving the following links:
http://www.spec.org/cpu2017/flags/gcc.2017-11-20.xml,
http://www.spec.org/cpu2017/flags/aocc100-flags-revC-I.xml,
http://www.spec.org/cpu2017/flags/HPE-Platform-Flags-AMD-V1.2-EPYC-revD.xml.