#------------------------------------------------------------------------------
# AMD AOCC 400 SPEC CPU 2017 Rate Configuration File for 64-bit Linux
# based off of AMD's amd_rate_aocc400_genoa_B1.cfg
#------------------------------------------------------------------------------
#--------- Label --------------------------------------------------------------
# Arbitrary string to tag binaries (no spaces allowed)
# Two Suggestions: # (1) EDIT this label as you try new ideas.
% define label "aocc400-ofast-zen4" # (2) Use a label meaningful to *you*.
#--------- Preprocessor -------------------------------------------------------
%ifndef %{bits} # EDIT to control 32 or 64 bit compilation. Or,
% define bits 64 # you can set it on the command line using:
%endif # 'runcpu --define bits=nn'
%ifndef %{build_ncpus} # EDIT to adjust number of simultaneous compiles.
% define build_ncpus 64 # Or, you can set it on the command line:
%endif # 'runcpu --define build_ncpus=nn'
# Don't change this part.
%if %{bits} == 64
% define model -m64
%elif %{bits} == 32
% define model -m32
%else
% error Please define number of bits - see instructions in config file
%endif
#--------- Global Settings ----------------------------------------------------
# For info, see:
# https://www.spec.org/cpu2017/Docs/config.html#fieldname
# Example: https://www.spec.org/cpu2017/Docs/config.html#tune
backup_config = 0
command_add_redirect = 1
flagsurl = $[top]/config/flags/gcc.xml
ignore_errors = 1
iterations = 1
label = % {label}-m%{bits}. # fix this line because the @#$@$ upload process will not accept it
line_width = 1020
log_line_width = 1020
makeflags = --jobs=%{build_ncpus}
mean_anyway = 1
output_format = txt,csv
preenv = 1
reportable = 0
nobuild = 0
verify_binaries = no # Somewhat dangerous!
tune = base,peak # EDIT if needed: set to "base" for old GCC.
# See note "Older GCC" above.
#--------- How Many CPUs? -----------------------------------------------------
# Both SPECrate and SPECspeed can test multiple chips / cores / hw threads
# - For SPECrate, you set the number of copies.
# - For SPECspeed, you set the number of threads.
# See: https://www.spec.org/cpu2017/Docs/system-requirements.html#MultipleCPUs
#
# q. How many should I set?
# a. Unknown, you will have to try it and see!
#
# To get you started, some suggestions:
#
# copies - This config file defaults to testing only 1 copy. You might
# try changing it to match the number of cores on your system,
# or perhaps the number of virtual CPUs as reported by:
# grep -c processor /proc/cpuinfo
# Be sure you have enough memory. See:
# https://www.spec.org/cpu2017/Docs/system-requirements.html#memory
#
# threads - This config file sets a starting point. You could try raising
# it. A higher thread count is much more likely to be useful for
# fpspeed than for intspeed.
#
intrate,fprate:
copies = 128 # EDIT to change number of copies (see above)
intspeed,fpspeed:
threads = 4 # EDIT to change number of OpenMP threads (see above)
# default copy counts:
default:
copies = 128
# Bind commands for assigning affinity:
submit = numactl --localalloc --physcpubind=$SPECCOPYNUM -- $command
################################################################################
#------- Compilers ------------------------------------------------------------
default:
#
%define aocc_dir "/opt/AMD/aocc-compiler-4.0.0" # EDIT
################################################################################
# Paths and Environment Variables
# # ** MODIFY AS NEEDED (modification should not be necessary for runs) **
# ################################################################################
# # Allow environment variables to be set before runs:
preenv = 1
#
# # Necessary to avoid out-of-memory exceptions on certain SUTs:
preENV_MALLOC_CONF = retain:true
#
# # Define the name of the directory that holds AMD library files:
%define binary_package_name
%define lib_dir %{binary_package_name}_lib
# Set the shared object library path for runs and builds:
#preENV_LD_LIBRARY_PATH = $[top]/%{lib_dir}/lib:$[top]/%{lib_dir}/lib32:%{ENV_LD_LIBRARY_PATH}
#
preENV_LD_LIBRARY_PATH = %{aocc_dir}/lib
#preENV_LD_LIBRARY_PATH = %{gcc_dir}/lib64/:%{gcc_dir}/lib/:/lib64:%{ENV_LD_LIBRARY_PATH}
SPECLANG = %{aocc_dir}/bin/
CC = $(SPECLANG)clang %{model}
CXX = $(SPECLANG)clang++ %{model}
FC = $(SPECLANG)flang %{model}
CLD = $(SPECLANG)clang %{model}
CXXLD = $(SPECLANG)clang++ %{model}
FLD = $(SPECLANG)flang %{model}
# How to say "Show me your version, please"
CC_VERSION_OPTION = --version
CXX_VERSION_OPTION = --version
FC_VERSION_OPTION = --version
default:
%if %{bits} == 64
sw_base_ptrsize = 64-bit
sw_peak_ptrsize = 64-bit
%else
sw_base_ptrsize = 32-bit
sw_peak_ptrsize = 32-bit
%endif
#--------- Portability --------------------------------------------------------
default: # data model applies to all benchmarks
EXTRA_PORTABILITY = -DSPEC_LP64
# *** Benchmark-specific portability ***
# # Anything other than the data model is only allowed where a need is proven.
# # (ordered by last 2 digits of benchmark number)
#
500.perlbench_r: #lang='C'
PORTABILITY = -DSPEC_LINUX_X64
521.wrf_r: #lang='F,C'
CPORTABILITY = -DSPEC_CASE_FLAG
FPORTABILITY = -Mbyteswapio
523.xalancbmk_r: #lang='CXX'
PORTABILITY = -DSPEC_LINUX
526.blender_r: #lang='CXX,C'
CPORTABILITY = -funsigned-char
527.cam4_r: #lang='F,C'
PORTABILITY = -DSPEC_CASE_FLAG
################################################################################
# Default libraries and variables
################################################################################
###############################################################################
# AOCC 4.0.0 workarounds that do not count as PORTABILITY
################################################################################
# The workarounds in this section would not qualify under the SPEC CPU
# PORTABILITY rule.
# - In peak, they can be set as needed for individual benchmarks.
# - In base, individual settings are not allowed; set for whole suite.
# Use EXTRA_CFLAGS, EXTRA_CXXFLAGS, and EXTRA_FFLAGS for them.
#
# See:
# https://www.spec.org/cpu2017/Docs/runrules.html#portability
# https://www.spec.org/cpu2017/Docs/runrules.html#BaseFlags
#######################
# Default workarounds #
#######################
default:
# Allow unused compile/link arguments without triggering warnings during build:
EXTRA_CFLAGS = -Wno-unused-command-line-argument
EXTRA_CXXFLAGS = -Wno-unused-command-line-argument
EXTRA_FFLAGS = -Wno-unused-command-line-argument
LDOPTIONS = -Wno-unused-command-line-argument
####################
# Base workarounds #
####################
#
# *** NONE ***
#
##############################
# Integer workarounds - base #
##############################
#
intrate=base:
# The following is necessary for 502/602 gcc:
EXTRA_LDFLAGS = -z muldefs
#########################
# FP workarounds - base #
#########################
#
# *** NONE ***
#
####################
# Peak workarounds #
####################
#
# *** NONE ***
#
##############################
# Integer workarounds - peak #
##############################
502.gcc_r=peak: #lang='C'
EXTRA_CFLAGS = -Wno-unused-command-line-argument \
-fgnu89-inline
EXTRA_LDFLAGS = -z muldefs
#####################################
# Floating Point workarounds - peak #
#####################################
#
# *** NONE ***
#
################################################################################
# Tuning Flags
################################################################################
# Libraries:
# EXTRA_LIBS = -lamdalloc -lamdlibm -lm
# MATHLIBOPT = #clearing this variable or else SPEC will set it to -lm
# VECMATHLIB = -fveclib=AMDLIBM
# # Variables:
# OPT_ROOT = -march=znver3 $(VECMATHLIB) -ffast-math
# OPT_ROOT_BASE = -O3 $(OPT_ROOT)
# OPT_ROOT_PEAK = -Ofast $(OPT_ROOT) -flto #Ofast enables -ffast-math
#####################
# Base tuning flags #
#####################
default=base:
OPTIMIZE = -Ofast -march=znver4 -flto
# Libraries:
EXTRA_LIBS = -lflang
#EXTRA_LIBS = -lamdlibm -lm -lamdalloc -lflang
# Don't put the AMD and mvec math libraries in MATH_LIBS because it will trigger a reporting issue
# because GCC won't use them. Forcefeed all benchmarks the math libraries in EXTRA_LIBS and clear
# out MATH_LIBS.
#MATH_LIBS =
default:
basepeak = yes
#--------- EDIT to match your version -----------------------------------------
default:
sw_compiler001 = AMD Optimizing Compiler (AOCC) 4.0.0
sw_compiler002 = AOCC_4.0.0-Build 434 2022_10_28 based on LLVM Mirror.Version.14.0.6
#--------- EDIT info about you ------------------------------------------------
# To understand the difference between hw_vendor/sponsor/tester, see:
# https://www.spec.org/cpu2017/Docs/config.html#test_sponsor
intrate,intspeed,fprate,fpspeed: # Important: keep this line
hw_vendor = Dell
tester = Alex Mericas
test_sponsor = Rivos Inc
license_num = 6386
# prepared_by = # Ima Pseudonym # Whatever you like: is never output
hw_model000 = PowerEdge R7615 (AMD EPYC 9654 96-Core Processor)
#--------- EDIT system information --------------------------------------------
intrate,intspeed,fprate,fpspeed: # Important: keep this line
# Example # Brief info about field
################################################################################
# Hardware, firmware and software information
################################################################################
hw_avail =Feb-2023
sw_avail =Nov-2022
hw_cpu_name =AMD EPYC 9654
hw_cpu_nominal_mhz =2450
hw_cpu_max_mhz =3500
hw_ncores =96
hw_nthreadspercore =1
hw_ncpuorder =1 chips
hw_other =None # Other perf-relevant hw, or "None"
fw_bios =Dell
sw_base_ptrsize =64-bit
hw_pcache =32 KB I + 32 KB D on chip per core
hw_scache =1 MB I+D on chip per core
hw_tcache000 =384 MB I+D on chip per chip, 32 MB shared / 8
hw_tcache001 = cores
hw_ocache =None
# sw_file = # ext99 # File system
# sw_os001 = # Linux Sailboat # Operating system
# sw_os002 = # Distribution 7.2 SP1 # and version
sw_other = jemalloc # Other perf-relevant sw, or "None"
# sw_state = # Run level 99 # Software state.
power_management = # briefly summarize power settings
# Note: Some commented-out fields above are automatically set to preliminary
# values by sysinfo
# https://www.spec.org/cpu2017/Docs/config.html#sysinfo
# Uncomment lines for which you already know a better answer than sysinfo
__HASH__
