Talos II Dual 22-Core POWER9 Spectre Benchmarks

POWER9 Spectre mitigation benchmarks by Michael Larabel.

HTML result view exported from: https://openbenchmarking.org/result/1811098-SK-TALOSIIDU15&grr.

Talos II Dual 22-Core POWER9 Spectre BenchmarksProcessorMotherboardMemoryDiskGraphicsMonitorNetworkOSKernelCompilerFile-SystemScreen ResolutionKernel ProtectionDefault Kernel + User ProtectionNo ProtectionPOWER9 altivec supported @ 3.80GHz (44 Cores / 176 Threads)PowerNV T2P9D01 REV 1.0165536MBSamsung SSD 960 EVO 500GBASPEED ASPEED FamilyVE228Broadcom NetXtreme BCM5719 Gigabit PCIeUbuntu 18.104.18.0-10-generic (ppc64le)GCC 8.2.0ext41920x1080OpenBenchmarking.orgCompiler Details- --build=powerpc64le-linux-gnu --disable-libphobos --disable-multilib --disable-werror --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++ --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-plugin --enable-secureplt --enable-shared --enable-targets=powerpcle-linux --enable-threads=posix --host=powerpc64le-linux-gnu --program-prefix=powerpc64le-linux-gnu- --target=powerpc64le-linux-gnu --with-cpu=power8 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-long-double-128 -v Processor Details- Scaling Governor: powernv-cpufreq ondemandPython Details- Python 2.7.15+ + Python 3.6.7Security Details- Kernel Protection: __user pointer sanitization ori31 speculation barrier enabled + Indirect branch serialisation (kernel only) + Kernel entry/exit barrier (eieio) - Default Kernel + User Protection: __user pointer sanitization ori31 speculation barrier enabled + Indirect branch cache disabled + Kernel entry/exit barrier (eieio) Disk Details- Default Kernel + User Protection, No Protection: NONE / errors=remount-ro,relatime,rw

Talos II Dual 22-Core POWER9 Spectre Benchmarksstockfish: Total Timebyte: Dhrystone 2build-linux-kernel: Time To Compilestress-ng: Hsearchaobench: 2048 x 2048 - Total Timepybench: Total For Average Test Timesbuild-llvm: Time To Compilego-benchmark: buildmpcbench: Multi-Precision Benchmarkcython-bench: go-benchmark: httpphpbench: PHP Benchmark Suitebuild-imagemagick: Time To Compilecompilebench: Compilestress-ng: Memory Copyingstress-ng: CPU Stressstress-ng: Semaphoresstress-ng: Tsearchstress-ng: Forkingstress-ng: Socket Activitystress-ng: Cryptostress-ng: Vector Mathstress-ng: Glibc Qsort Data Sortingstress-ng: Lsearchstress-ng: Glibc C String Functionsstress-ng: System V Message Passingrodinia: OpenMP Streamclusterpovray: Trace Timehackbench: 32 - Processgo-benchmark: garbagebork: File Encryption Timego-benchmark: jsonopenssl: RSA 4096-bit Performancerodinia: OpenMP LavaMDc-ray: Total Time - 4K, 16 Rays Per Pixelprimesieve: 1e12 Prime Number Generationrodinia: OpenMP CFD Solverrust-prime: Prime Number Test To 200,000,000ffte: N=256, 1D Complex FFT Routinecompilebench: Read Compiled Treecompilebench: Initial CreateKernel ProtectionDefault Kernel + User ProtectionNo Protection797992272625851350.6618693159.51193917428049160231537040.66192851033004427.77253383401232134716688958592270948149884163683861370824649220429.6024.9227.17121124522.224238819741718.3719.0517.4913.253.7267701823346791351032675582155.4616413864.61409018529899607203400766.62134171715973130.03232483171227934751785959360262728075881542532021290587629361729.7725.9928.14123370022.514898644739418.7618.9317.4714.793.511538296830118462661871850.0120166059.52181617227993756394543338.88156838132978327.28253983681240245663889860585287178298886993693891379226660701228.5721.8523.89110682522.163813471739016.7517.8316.3812.933.321887348OpenBenchmarking.org

Stockfish

Total Time

OpenBenchmarking.orgNodes Per Second, More Is BetterStockfish 9Total TimeKernel ProtectionDefault Kernel + User ProtectionNo Protection20M40M60M80M100MSE +/- 1474158.28, N = 3SE +/- 662670.61, N = 3SE +/- 743288.42, N = 37979922779135103830118461. (CXX) g++ options: -m64 -lpthread -fno-exceptions -std=c++11 -pedantic -O3 -flto

BYTE Unix Benchmark

Computational Test: Dhrystone 2

OpenBenchmarking.orgLPS, More Is BetterBYTE Unix Benchmark 3.6Computational Test: Dhrystone 2Kernel ProtectionDefault Kernel + User ProtectionNo Protection6M12M18M24M30MSE +/- 494133.76, N = 3SE +/- 13055.64, N = 3SE +/- 109377.13, N = 3262585132675582126618718

Timed Linux Kernel Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Linux Kernel Compilation 4.18Time To CompileKernel ProtectionDefault Kernel + User ProtectionNo Protection1224364860SE +/- 0.56, N = 9SE +/- 0.92, N = 4SE +/- 0.71, N = 550.6655.4650.01

Stress-NG

Test: Hsearch

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: HsearchKernel ProtectionDefault Kernel + User ProtectionNo Protection40K80K120K160K200KSE +/- 4467.03, N = 12SE +/- 3274.73, N = 12SE +/- 1283.68, N = 31869311641382016601. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

AOBench

Size: 2048 x 2048 - Total Time

OpenBenchmarking.orgSeconds, Fewer Is BetterAOBenchSize: 2048 x 2048 - Total TimeKernel ProtectionDefault Kernel + User ProtectionNo Protection1428425670SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 359.5164.6159.521. (CC) gcc options: -lm -O3

PyBench

Total For Average Test Times

OpenBenchmarking.orgMilliseconds, Fewer Is BetterPyBench 2018-02-16Total For Average Test TimesKernel ProtectionDefault Kernel + User ProtectionNo Protection9001800270036004500SE +/- 5.24, N = 3SE +/- 3.06, N = 3193940901816

Timed LLVM Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed LLVM Compilation 6.0.1Time To CompileKernel ProtectionDefault Kernel + User ProtectionNo Protection4080120160200174185172

Go Benchmarks

Test: build

OpenBenchmarking.orgNanoseconds/Operation, Fewer Is BetterGo BenchmarksTest: buildKernel ProtectionDefault Kernel + User ProtectionNo Protection6000M12000M18000M24000M30000MSE +/- 94097418.28, N = 3SE +/- 573457278.59, N = 3SE +/- 318891587.38, N = 12280491602312989960720327993756394

GNU MPC

Multi-Precision Benchmark

OpenBenchmarking.orgGlobal Score, More Is BetterGNU MPC 1.1.0Multi-Precision BenchmarkKernel ProtectionDefault Kernel + User ProtectionNo Protection12002400360048006000SE +/- 3.33, N = 3SE +/- 3.33, N = 35370400754331. (CC) gcc options: -m64 -mtune=power8 -O3 -MT -MD -MP -MF

Cython benchmark

OpenBenchmarking.orgSeconds, Fewer Is BetterCython benchmark 0.27Kernel ProtectionDefault Kernel + User ProtectionNo Protection1530456075SE +/- 0.65, N = 3SE +/- 0.04, N = 3SE +/- 0.03, N = 340.6666.6238.88

Go Benchmarks

Test: http

OpenBenchmarking.orgNanoseconds/Operation, Fewer Is BetterGo BenchmarksTest: httpKernel ProtectionDefault Kernel + User ProtectionNo Protection400K800K1200K1600K2000KSE +/- 220954.01, N = 10SE +/- 107637.64, N = 12SE +/- 209033.95, N = 9192851013417171568381

PHPBench

PHP Benchmark Suite

OpenBenchmarking.orgScore, More Is BetterPHPBench 0.8.1PHP Benchmark SuiteKernel ProtectionDefault Kernel + User ProtectionNo Protection70K140K210K280K350KSE +/- 554.58, N = 3SE +/- 1702.94, N = 3SE +/- 3520.86, N = 3330044159731329783

Timed ImageMagick Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed ImageMagick Compilation 6.9.0Time To CompileKernel ProtectionDefault Kernel + User ProtectionNo Protection714212835SE +/- 0.40, N = 5SE +/- 0.44, N = 4SE +/- 0.12, N = 327.7730.0327.28

Compile Bench

Test: Compile

OpenBenchmarking.orgMB/s, More Is BetterCompile Bench 0.6Test: CompileKernel ProtectionDefault Kernel + User ProtectionNo Protection5001000150020002500SE +/- 9.03, N = 3SE +/- 8.20, N = 3SE +/- 11.40, N = 3253323242539

Stress-NG

Test: Memory Copying

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: Memory CopyingKernel ProtectionDefault Kernel + User ProtectionNo Protection2K4K6K8K10KSE +/- 154.58, N = 3SE +/- 122.58, N = 3SE +/- 154.42, N = 38340831783681. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: CPU Stress

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: CPU StressKernel ProtectionDefault Kernel + User ProtectionNo Protection3K6K9K12K15KSE +/- 18.47, N = 3SE +/- 7.41, N = 3SE +/- 9.31, N = 31232112279124021. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: Semaphores

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: SemaphoresKernel ProtectionDefault Kernel + User ProtectionNo Protection100K200K300K400K500KSE +/- 5258.60, N = 3SE +/- 4183.35, N = 3SE +/- 8664.98, N = 33471663475174566381. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: Tsearch

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: TsearchKernel ProtectionDefault Kernel + User ProtectionNo Protection2004006008001000SE +/- 2.77, N = 3SE +/- 5.85, N = 3SE +/- 2.73, N = 38898598981. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: Forking

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: ForkingKernel ProtectionDefault Kernel + User ProtectionNo Protection13K26K39K52K65KSE +/- 643.64, N = 3SE +/- 106.66, N = 3SE +/- 278.00, N = 35859259360605851. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: Socket Activity

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: Socket ActivityKernel ProtectionDefault Kernel + User ProtectionNo Protection6K12K18K24K30KSE +/- 103.85, N = 3SE +/- 143.39, N = 3SE +/- 211.09, N = 32709426272287171. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: Crypto

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: CryptoKernel ProtectionDefault Kernel + User ProtectionNo Protection2K4K6K8K10KSE +/- 66.57, N = 3SE +/- 41.59, N = 3SE +/- 11.73, N = 38149807582981. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: Vector Math

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: Vector MathKernel ProtectionDefault Kernel + User ProtectionNo Protection20K40K60K80K100KSE +/- 212.84, N = 3SE +/- 114.43, N = 3SE +/- 225.56, N = 38841688154886991. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: Glibc Qsort Data Sorting

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: Glibc Qsort Data SortingKernel ProtectionDefault Kernel + User ProtectionNo Protection80160240320400SE +/- 0.71, N = 3SE +/- 0.49, N = 3SE +/- 1.48, N = 33682533691. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: Lsearch

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: LsearchKernel ProtectionDefault Kernel + User ProtectionNo Protection80160240320400SE +/- 1.09, N = 3SE +/- 1.34, N = 3SE +/- 0.70, N = 33862023891. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: Glibc C String Functions

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: Glibc C String FunctionsKernel ProtectionDefault Kernel + User ProtectionNo Protection300K600K900K1200K1500KSE +/- 8505.49, N = 3SE +/- 6341.64, N = 3SE +/- 14521.20, N = 31370824129058713792261. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Stress-NG

Test: System V Message Passing

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.07.26Test: System V Message PassingKernel ProtectionDefault Kernel + User ProtectionNo Protection1.4M2.8M4.2M5.6M7MSE +/- 32592.10, N = 3SE +/- 63949.45, N = 3SE +/- 92751.45, N = 36492204629361766070121. (CC) gcc options: -O2 -std=gnu99 -lm -lz -lcrypt -lrt -lpthread -laio -lc

Rodinia

Test: OpenMP Streamcluster

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 2.4Test: OpenMP StreamclusterKernel ProtectionDefault Kernel + User ProtectionNo Protection714212835SE +/- 0.05, N = 3SE +/- 0.02, N = 3SE +/- 0.05, N = 329.6029.7728.571. (CXX) g++ options: -O2 -lOpenCL

POV-Ray

Trace Time

OpenBenchmarking.orgSeconds, Fewer Is BetterPOV-Ray 3.7.0.7Trace TimeKernel ProtectionDefault Kernel + User ProtectionNo Protection612182430SE +/- 0.13, N = 3SE +/- 0.09, N = 3SE +/- 0.15, N = 324.9225.9921.851. (CXX) g++ options: -pipe -O3 -ffast-math -pthread -R/usr/lib -lSDL -lX11 -lIlmImf -lImath -lHalf -lIex -lIexMath -lIlmThread -lpthread -ltiff -ljpeg -lpng -lz -lrt -lm -lboost_thread -lboost_system

Hackbench

Count: 32 - Type: Process

OpenBenchmarking.orgSeconds, Fewer Is BetterHackbenchCount: 32 - Type: ProcessKernel ProtectionDefault Kernel + User ProtectionNo Protection714212835SE +/- 0.21, N = 3SE +/- 0.02, N = 3SE +/- 0.04, N = 327.1728.1423.891. (CC) gcc options: -lpthread

Go Benchmarks

Test: garbage

OpenBenchmarking.orgNanoseconds/Operation, Fewer Is BetterGo BenchmarksTest: garbageKernel ProtectionDefault Kernel + User ProtectionNo Protection300K600K900K1200K1500KSE +/- 12454.68, N = 12SE +/- 12597.46, N = 12SE +/- 3498.26, N = 3121124512337001106825

Bork File Encrypter

File Encryption Time

OpenBenchmarking.orgSeconds, Fewer Is BetterBork File Encrypter 1.4File Encryption TimeKernel ProtectionDefault Kernel + User ProtectionNo Protection510152025SE +/- 0.04, N = 3SE +/- 0.02, N = 3SE +/- 0.03, N = 322.2222.5122.16

Go Benchmarks

Test: json

OpenBenchmarking.orgNanoseconds/Operation, Fewer Is BetterGo BenchmarksTest: jsonKernel ProtectionDefault Kernel + User ProtectionNo Protection1000K2000K3000K4000K5000KSE +/- 3340.75, N = 3SE +/- 87285.98, N = 12SE +/- 54441.22, N = 3423881948986443813471

OpenSSL

RSA 4096-bit Performance

OpenBenchmarking.orgSigns Per Second, More Is BetterOpenSSL 1.1.1RSA 4096-bit PerformanceKernel ProtectionDefault Kernel + User ProtectionNo Protection16003200480064008000SE +/- 23.66, N = 3SE +/- 51.34, N = 3SE +/- 26.62, N = 37417739473901. (CC) gcc options: -pthread -m64 -O3 -lssl -lcrypto -ldl

Rodinia

Test: OpenMP LavaMD

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 2.4Test: OpenMP LavaMDKernel ProtectionDefault Kernel + User ProtectionNo Protection510152025SE +/- 0.31, N = 4SE +/- 0.26, N = 3SE +/- 0.20, N = 318.3718.7616.751. (CXX) g++ options: -O2 -lOpenCL

C-Ray

Total Time - 4K, 16 Rays Per Pixel

OpenBenchmarking.orgSeconds, Fewer Is BetterC-Ray 1.1Total Time - 4K, 16 Rays Per PixelKernel ProtectionDefault Kernel + User ProtectionNo Protection510152025SE +/- 0.05, N = 3SE +/- 0.01, N = 3SE +/- 0.03, N = 319.0518.9317.831. (CC) gcc options: -lm -lpthread -O3

Primesieve

1e12 Prime Number Generation

OpenBenchmarking.orgSeconds, Fewer Is BetterPrimesieve 7.11e12 Prime Number GenerationKernel ProtectionDefault Kernel + User ProtectionNo Protection48121620SE +/- 0.03, N = 3SE +/- 0.05, N = 3SE +/- 0.04, N = 317.4917.4716.381. (CXX) g++ options: -O3 -lpthread

Rodinia

Test: OpenMP CFD Solver

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 2.4Test: OpenMP CFD SolverKernel ProtectionDefault Kernel + User ProtectionNo Protection48121620SE +/- 0.07, N = 3SE +/- 0.08, N = 3SE +/- 0.05, N = 313.2514.7912.931. (CXX) g++ options: -O2 -lOpenCL

Rust Prime Benchmark

Prime Number Test To 200,000,000

OpenBenchmarking.orgSeconds, Fewer Is BetterRust Prime BenchmarkPrime Number Test To 200,000,000Kernel ProtectionDefault Kernel + User ProtectionNo Protection0.8371.6742.5113.3484.185SE +/- 0.07, N = 3SE +/- 0.06, N = 3SE +/- 0.06, N = 33.723.513.321. (CC) gcc options: -m64 -pie -nodefaultlibs

FFTE

Test: N=256, 1D Complex FFT Routine

OpenBenchmarking.orgMFLOPS, More Is BetterFFTE 6.0Test: N=256, 1D Complex FFT RoutineKernel Protection15003000450060007500SE +/- 0.43, N = 367701. (F9X) gfortran options: -O3 -fomit-frame-pointer -fopenmp -pthread -lmpi_usempif08 -lmpi_mpifh -lmpi

Compile Bench

Test: Read Compiled Tree

OpenBenchmarking.orgMB/s, More Is BetterCompile Bench 0.6Test: Read Compiled TreeKernel ProtectionDefault Kernel + User ProtectionNo Protection400800120016002000SE +/- 7.11, N = 3SE +/- 4.49, N = 3SE +/- 0.59, N = 3182315381887

Compile Bench

Test: Initial Create

OpenBenchmarking.orgMB/s, More Is BetterCompile Bench 0.6Test: Initial CreateKernel ProtectionDefault Kernel + User ProtectionNo Protection80160240320400SE +/- 0.18, N = 3SE +/- 1.65, N = 3SE +/- 1.84, N = 3346296348


Phoronix Test Suite v10.8.4