GCC 7.2 vs. GCC 8 vs. LLVM Clang 5 vs. LLVM Clang 6 Znver1 EPYC

AMD EPYC 7601 32-Core testing with a TYAN B8026T70AE24HR, testing various compilers. Tests for a future article on Phoronix.

HTML result view exported from: https://openbenchmarking.org/result/1711053-AL-GCC72VSGC51&sor.

GCC 7.2 vs. GCC 8 vs. LLVM Clang 5 vs. LLVM Clang 6 Znver1 EPYCProcessorMotherboardChipsetMemoryDiskGraphicsMonitorNetworkOSKernelDesktopCompilerFile-SystemScreen ResolutionGCC 7.2.0GCC 8.0.0 20171030LLVM Clang 5.0LLVM Clang 6.0 svn316959AMD EPYC 7601 32-Core @ 2.20GHz (32 Cores / 64 Threads)TYAN B8026T70AE24HRAMD Device 1450129024MB120GB Force MP500ASPEED ASPEED FamilyAcer P243WBroadcom Limited NetXtreme BCM5720 Gigabit PCIeUbuntu 17.104.13.0-16-generic (x86_64)GNOME Shell 3.26.1GCC 7.2.0ext41920x1200GCC 8.0.0 20171030Clang 5.0.0-3Clang 6.0.0-svn316959-1~exp1OpenBenchmarking.orgCompiler Details- GCC 7.2.0, GCC 8.0.0 20171030: --disable-multilib --enable-checking=release --enable-languages=c,c++Processor Details- Scaling Governor: acpi-cpufreq ondemand

GCC 7.2 vs. GCC 8 vs. LLVM Clang 5 vs. LLVM Clang 6 Znver1 EPYChpcg: fftw: Stock - 1D FFT Size 1024fftw: Stock - 2D FFT Size 1024fftw: Float + SSE - 1D FFT Size 1024fftw: Float + SSE - 2D FFT Size 1024gmpbench: Total Timescimark2: Compositescimark2: Monte Carloscimark2: Fast Fourier Transformscimark2: Sparse Matrix Multiplyscimark2: Dense LU Matrix Factorizationscimark2: Jacobi Successive Over-Relaxationtscp: AI Chess Performancettsiod-renderer: Phong Rendering With Soft-Shadow Mappinggraphics-magick: Blurgraphics-magick: Sharpengraphics-magick: Resizinggraphics-magick: HWB Color Spacegraphics-magick: Local Adaptive Thresholdinghimeno: Poisson Pressure Solverc-ray: Total Timeprimesieve: 1e12 Prime Number Generationsmallpt: Global Illumination Renderer; 100 Samplesstockfish: Total Timeencode-mp3: WAV To MP3ffmpeg: H.264 HD To NTSC DVtjbench: Decompression Throughputpgbench: Buffer Test - Normal Load - Read Onlypgbench: Buffer Test - Single Thread - Read Onlyredis: LPOPredis: SADDredis: LPUSHredis: GETredis: SETGCC 7.2.0GCC 8.0.0 20171030LLVM Clang 5.0LLVM Clang 6.0 svn3169590.738539.106420.2326763198343918.801925.72194.45223.692376.615150.441683.40861129407.27148177174198109936.773.0811.914448511.2010.69141.55308134.5710784.92202065.84189478.65193867.82198580.74192063.140.728416.236622.9026876211313926.201933.09554.99223.062444.394757.531685.48853655407.58148179176199110942.442.7411.944450010.7110.66143.79303543.1710709.13195542.12193163.62192219.37198883.33193210.490.747613.476090.3726224187621910.11560.13221.422402.424938.621427.9789528441.26115145161110965.544.2611.824435612.8110.41142.26167561.1910669.81191613.95194776.06203774.66197952.09199028.190.727580.835852.2026395190361923.63551.21221.332534.374883.391427.8693320641.02116146162110978.764.5511.804442912.8910.45144.461823824.961489239.301247939.291584704.001260800.37OpenBenchmarking.org

High Performance Conjugate Gradient

OpenBenchmarking.orgGFLOP/s, More Is BetterHigh Performance Conjugate Gradient 3.0LLVM Clang 5.0GCC 7.2.0LLVM Clang 6.0 svn316959GCC 8.0.0 201710300.16650.3330.49950.6660.8325SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.00, N = 30.740.730.720.72

FFTW

Build: Stock - Size: 1D FFT Size 1024

OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Stock - Size: 1D FFT Size 1024GCC 7.2.0GCC 8.0.0 20171030LLVM Clang 5.0LLVM Clang 6.0 svn3169592K4K6K8K10KSE +/- 4.79, N = 3SE +/- 5.10, N = 3SE +/- 8.50, N = 3SE +/- 14.03, N = 38539.108416.237613.477580.831. (CC) gcc options: -pthread -O3 -march=znver1 -lm

FFTW

Build: Stock - Size: 2D FFT Size 1024

OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Stock - Size: 2D FFT Size 1024GCC 8.0.0 20171030GCC 7.2.0LLVM Clang 5.0LLVM Clang 6.0 svn31695914002800420056007000SE +/- 33.11, N = 3SE +/- 104.31, N = 4SE +/- 10.25, N = 3SE +/- 96.36, N = 46622.906420.236090.375852.201. (CC) gcc options: -pthread -O3 -march=znver1 -lm

FFTW

Build: Float + SSE - Size: 1D FFT Size 1024

OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Float + SSE - Size: 1D FFT Size 1024GCC 8.0.0 20171030GCC 7.2.0LLVM Clang 6.0 svn316959LLVM Clang 5.06K12K18K24K30KSE +/- 120.67, N = 3SE +/- 26.61, N = 3SE +/- 28.81, N = 3SE +/- 35.53, N = 3268762676326395262241. (CC) gcc options: -pthread -O3 -march=znver1 -lm

FFTW

Build: Float + SSE - Size: 2D FFT Size 1024

OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Float + SSE - Size: 2D FFT Size 1024GCC 8.0.0 20171030GCC 7.2.0LLVM Clang 6.0 svn316959LLVM Clang 5.05K10K15K20K25KSE +/- 57.10, N = 3SE +/- 270.69, N = 3SE +/- 25.56, N = 3SE +/- 65.36, N = 3211311983419036187621. (CC) gcc options: -pthread -O3 -march=znver1 -lm

GNU GMP GMPbench

Total Time

OpenBenchmarking.orgGMPbench Score, More Is BetterGNU GMP GMPbench 6.1.2Total TimeGCC 8.0.0 20171030GCC 7.2.080016002400320040003926.203918.801. (CC) gcc options: -O3 -march=znver1 -lm

SciMark

Computational Test: Composite

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: CompositeGCC 8.0.0 20171030GCC 7.2.0LLVM Clang 6.0 svn316959LLVM Clang 5.0400800120016002000SE +/- 7.80, N = 4SE +/- 8.62, N = 4SE +/- 5.14, N = 4SE +/- 9.03, N = 41933.091925.721923.631910.111. (CC) gcc options: -O3 -march=znver1 -lm

SciMark

Computational Test: Monte Carlo

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Monte CarloLLVM Clang 5.0GCC 8.0.0 20171030LLVM Clang 6.0 svn316959GCC 7.2.0120240360480600SE +/- 0.06, N = 4SE +/- 0.07, N = 4SE +/- 0.19, N = 4SE +/- 0.01, N = 4560.13554.99551.21194.451. (CC) gcc options: -O3 -march=znver1 -lm

SciMark

Computational Test: Fast Fourier Transform

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Fast Fourier TransformGCC 7.2.0GCC 8.0.0 20171030LLVM Clang 5.0LLVM Clang 6.0 svn31695950100150200250SE +/- 0.12, N = 4SE +/- 0.07, N = 4SE +/- 0.15, N = 4SE +/- 0.02, N = 4223.69223.06221.42221.331. (CC) gcc options: -O3 -march=znver1 -lm

SciMark

Computational Test: Sparse Matrix Multiply

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Sparse Matrix MultiplyLLVM Clang 6.0 svn316959GCC 8.0.0 20171030LLVM Clang 5.0GCC 7.2.05001000150020002500SE +/- 13.69, N = 4SE +/- 9.19, N = 4SE +/- 11.52, N = 4SE +/- 13.62, N = 42534.372444.392402.422376.611. (CC) gcc options: -O3 -march=znver1 -lm

SciMark

Computational Test: Dense LU Matrix Factorization

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Dense LU Matrix FactorizationGCC 7.2.0LLVM Clang 5.0LLVM Clang 6.0 svn316959GCC 8.0.0 2017103011002200330044005500SE +/- 37.84, N = 4SE +/- 38.03, N = 4SE +/- 14.77, N = 4SE +/- 31.33, N = 45150.444938.624883.394757.531. (CC) gcc options: -O3 -march=znver1 -lm

SciMark

Computational Test: Jacobi Successive Over-Relaxation

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Jacobi Successive Over-RelaxationGCC 8.0.0 20171030GCC 7.2.0LLVM Clang 5.0LLVM Clang 6.0 svn316959400800120016002000SE +/- 0.66, N = 4SE +/- 0.70, N = 4SE +/- 0.12, N = 4SE +/- 0.24, N = 41685.481683.401427.971427.861. (CC) gcc options: -O3 -march=znver1 -lm

TSCP

AI Chess Performance

OpenBenchmarking.orgNodes Per Second, More Is BetterTSCP 1.81AI Chess PerformanceLLVM Clang 6.0 svn316959LLVM Clang 5.0GCC 7.2.0GCC 8.0.0 20171030200K400K600K800K1000KSE +/- 316.00, N = 5SE +/- 544.63, N = 5SE +/- 329.95, N = 5SE +/- 264.40, N = 59332068952848611298536551. (CC) gcc options: -O3 -march=znver1 -march=native

TTSIOD 3D Renderer

Phong Rendering With Soft-Shadow Mapping

OpenBenchmarking.orgFPS, More Is BetterTTSIOD 3D Renderer 2.3aPhong Rendering With Soft-Shadow MappingGCC 8.0.0 20171030GCC 7.2.0LLVM Clang 5.0LLVM Clang 6.0 svn31695990180270360450SE +/- 6.79, N = 4SE +/- 6.30, N = 6SE +/- 0.05, N = 3SE +/- 0.06, N = 3407.58407.2741.2641.021. (CXX) g++ options: -O3 -march=znver1 -fomit-frame-pointer -ffast-math -mtune=native -flto -msse -mrecip -mfpmath=sse -msse2 -mssse3 -lSDL -lstdc++

GraphicsMagick

Operation: Blur

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.19Operation: BlurGCC 8.0.0 20171030GCC 7.2.0LLVM Clang 6.0 svn316959LLVM Clang 5.0306090120150SE +/- 1.00, N = 3SE +/- 0.67, N = 3148148116115-ldl-ldl-lgomp-lgomp1. (CC) gcc options: -fopenmp -O3 -march=znver1 -pthread -ljbig -lwebp -ltiff -ljpeg -lXext -lX11 -llzma -lz -lm -lpthread

GraphicsMagick

Operation: Sharpen

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.19Operation: SharpenGCC 8.0.0 20171030GCC 7.2.0LLVM Clang 6.0 svn316959LLVM Clang 5.04080120160200SE +/- 0.33, N = 3SE +/- 0.67, N = 3179177146145-ldl-ldl-lgomp-lgomp1. (CC) gcc options: -fopenmp -O3 -march=znver1 -pthread -ljbig -lwebp -ltiff -ljpeg -lXext -lX11 -llzma -lz -lm -lpthread

GraphicsMagick

Operation: Resizing

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.19Operation: ResizingGCC 8.0.0 20171030GCC 7.2.04080120160200SE +/- 0.88, N = 3SE +/- 0.67, N = 31761741. (CC) gcc options: -fopenmp -O3 -march=znver1 -pthread -ljbig -lwebp -ltiff -ljpeg -lXext -lX11 -llzma -lz -lm -ldl -lpthread

GraphicsMagick

Operation: HWB Color Space

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.19Operation: HWB Color SpaceGCC 8.0.0 20171030GCC 7.2.0LLVM Clang 6.0 svn316959LLVM Clang 5.04080120160200199198162161-ldl-ldl-lgomp-lgomp1. (CC) gcc options: -fopenmp -O3 -march=znver1 -pthread -ljbig -lwebp -ltiff -ljpeg -lXext -lX11 -llzma -lz -lm -lpthread

GraphicsMagick

Operation: Local Adaptive Thresholding

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.19Operation: Local Adaptive ThresholdingLLVM Clang 6.0 svn316959LLVM Clang 5.0GCC 8.0.0 20171030GCC 7.2.020406080100SE +/- 1.20, N = 3110110110109-lgomp-lgomp-ldl-ldl1. (CC) gcc options: -fopenmp -O3 -march=znver1 -pthread -ljbig -lwebp -ltiff -ljpeg -lXext -lX11 -llzma -lz -lm -lpthread

Himeno Benchmark

Poisson Pressure Solver

OpenBenchmarking.orgMFLOPS, More Is BetterHimeno Benchmark 3.0Poisson Pressure SolverLLVM Clang 6.0 svn316959LLVM Clang 5.0GCC 8.0.0 20171030GCC 7.2.02004006008001000SE +/- 3.60, N = 3SE +/- 2.40, N = 3SE +/- 1.81, N = 3SE +/- 21.77, N = 6978.76965.54942.44936.771. (CC) gcc options: -O3 -march=znver1 -mavx2

C-Ray

Total Time

OpenBenchmarking.orgSeconds, Fewer Is BetterC-Ray 1.1Total TimeGCC 8.0.0 20171030GCC 7.2.0LLVM Clang 5.0LLVM Clang 6.0 svn3169591.02382.04763.07144.09525.119SE +/- 0.07, N = 6SE +/- 0.06, N = 3SE +/- 0.04, N = 3SE +/- 0.02, N = 32.743.084.264.551. (CC) gcc options: -lm -lpthread -O3 -march=znver1

Primesieve

1e12 Prime Number Generation

OpenBenchmarking.orgSeconds, Fewer Is BetterPrimesieve 6.21e12 Prime Number GenerationLLVM Clang 6.0 svn316959LLVM Clang 5.0GCC 7.2.0GCC 8.0.0 201710303691215SE +/- 0.04, N = 3SE +/- 0.02, N = 3SE +/- 0.03, N = 3SE +/- 0.04, N = 311.8011.8211.9111.941. (CXX) g++ options: -O3 -march=znver1 -rdynamic -lpthread

Smallpt

Global Illumination Renderer; 100 Samples

OpenBenchmarking.orgSeconds, Fewer Is BetterSmallpt 1.0Global Illumination Renderer; 100 SamplesGCC 7.2.0GCC 8.0.0 20171030LLVM Clang 5.0LLVM Clang 6.0 svn3169590.91.82.73.64.5SE +/- 0.17, N = 6SE +/- 0.17, N = 644441. (CXX) g++ options: -fopenmp -O3 -march=znver1

Stockfish

Total Time

OpenBenchmarking.orgms, Fewer Is BetterStockfish 2014-11-26Total TimeLLVM Clang 5.0LLVM Clang 6.0 svn316959GCC 7.2.0GCC 8.0.0 2017103010002000300040005000SE +/- 4.04, N = 3SE +/- 44.52, N = 3SE +/- 56.11, N = 3SE +/- 55.17, N = 34356442944854500-flto-flto1. (CXX) g++ options: -lpthread -O3 -march=znver1 -fno-exceptions -fno-rtti -ansi -pedantic -msse -msse3 -mpopcnt

LAME MP3 Encoding

WAV To MP3

OpenBenchmarking.orgSeconds, Fewer Is BetterLAME MP3 Encoding 3.99.5WAV To MP3GCC 8.0.0 20171030GCC 7.2.0LLVM Clang 5.0LLVM Clang 6.0 svn3169593691215SE +/- 0.00, N = 5SE +/- 0.01, N = 5SE +/- 0.00, N = 5SE +/- 0.01, N = 510.7111.2012.8112.891. (CC) gcc options: -O3 -ffast-math -funroll-loops -fschedule-insns2 -fbranch-count-reg -fforce-addr -pipe -march=znver1 -lm

FFmpeg

H.264 HD To NTSC DV

OpenBenchmarking.orgSeconds, Fewer Is BetterFFmpeg 3.3.3H.264 HD To NTSC DVLLVM Clang 5.0LLVM Clang 6.0 svn316959GCC 8.0.0 20171030GCC 7.2.03691215SE +/- 0.11, N = 3SE +/- 0.04, N = 3SE +/- 0.07, N = 3SE +/- 0.04, N = 310.4110.4510.6610.69-Qunused-arguments -mstack-alignment=16-Qunused-arguments -mstack-alignment=16-fno-tree-vectorize-fno-tree-vectorize1. (CC) gcc options: -lavdevice -lavfilter -lavformat -lavcodec -lswresample -lswscale -lavutil -ldl -lxcb -lxcb-xfixes -lxcb-shape -lasound -lm -llzma -pthread -O3 -march=znver1 -std=c11 -fomit-frame-pointer -fno-math-errno -fno-signed-zeros -MMD -MF -MT

libjpeg-turbo tjbench

Test: Decompression Throughput

OpenBenchmarking.orgMegapixels/sec, More Is Betterlibjpeg-turbo tjbench 1.5.1Test: Decompression ThroughputLLVM Clang 6.0 svn316959GCC 8.0.0 20171030LLVM Clang 5.0GCC 7.2.0306090120150SE +/- 0.02, N = 3SE +/- 0.02, N = 3SE +/- 0.09, N = 3SE +/- 0.03, N = 3144.46143.79142.26141.551. (CC) gcc options: -O3 -march=znver1 -lm

PostgreSQL pgbench

Scaling: Buffer Test - Test: Normal Load - Mode: Read Only

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 10.0Scaling: Buffer Test - Test: Normal Load - Mode: Read OnlyGCC 7.2.0GCC 8.0.0 20171030LLVM Clang 5.070K140K210K280K350KSE +/- 1439.36, N = 3SE +/- 3027.80, N = 3SE +/- 15544.05, N = 6308134.57303543.17167561.19-shared-shared-lpgcommon -lpgport -lpthread -lrt -lcrypt -ldl -lm1. (CC) gcc options: -fno-strict-aliasing -fwrapv -O3 -march=znver1 -fPIC

PostgreSQL pgbench

Scaling: Buffer Test - Test: Single Thread - Mode: Read Only

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 10.0Scaling: Buffer Test - Test: Single Thread - Mode: Read OnlyGCC 7.2.0GCC 8.0.0 20171030LLVM Clang 5.02K4K6K8K10KSE +/- 140.64, N = 3SE +/- 47.06, N = 3SE +/- 58.33, N = 310784.9210709.1310669.81-shared-shared-lpgcommon -lpgport -lpthread -lrt -lcrypt -ldl -lm1. (CC) gcc options: -fno-strict-aliasing -fwrapv -O3 -march=znver1 -fPIC

Redis

Test: LPOP

OpenBenchmarking.orgRequests Per Second, More Is BetterRedis 3.0.1Test: LPOPLLVM Clang 6.0 svn316959GCC 7.2.0GCC 8.0.0 20171030LLVM Clang 5.0400K800K1200K1600K2000KSE +/- 97746.43, N = 6SE +/- 3741.75, N = 3SE +/- 3274.37, N = 6SE +/- 1726.15, N = 31823824.96202065.84195542.12191613.951. (CC) gcc options: -ggdb -rdynamic -lm -pthread

Redis

Test: SADD

OpenBenchmarking.orgRequests Per Second, More Is BetterRedis 3.0.1Test: SADDLLVM Clang 6.0 svn316959LLVM Clang 5.0GCC 8.0.0 20171030GCC 7.2.0300K600K900K1200K1500KSE +/- 43532.25, N = 6SE +/- 2862.24, N = 3SE +/- 2485.81, N = 3SE +/- 303.66, N = 31489239.30194776.06193163.62189478.651. (CC) gcc options: -ggdb -rdynamic -lm -pthread

Redis

Test: LPUSH

OpenBenchmarking.orgRequests Per Second, More Is BetterRedis 3.0.1Test: LPUSHLLVM Clang 6.0 svn316959LLVM Clang 5.0GCC 7.2.0GCC 8.0.0 20171030300K600K900K1200K1500KSE +/- 20238.90, N = 6SE +/- 2321.76, N = 3SE +/- 2952.51, N = 6SE +/- 3068.09, N = 41247939.29203774.66193867.82192219.371. (CC) gcc options: -ggdb -rdynamic -lm -pthread

Redis

Test: GET

OpenBenchmarking.orgRequests Per Second, More Is BetterRedis 3.0.1Test: GETLLVM Clang 6.0 svn316959GCC 8.0.0 20171030GCC 7.2.0LLVM Clang 5.0300K600K900K1200K1500KSE +/- 24641.50, N = 3SE +/- 2615.42, N = 6SE +/- 3940.07, N = 6SE +/- 3998.05, N = 61584704.00198883.33198580.74197952.091. (CC) gcc options: -ggdb -rdynamic -lm -pthread

Redis

Test: SET

OpenBenchmarking.orgRequests Per Second, More Is BetterRedis 3.0.1Test: SETLLVM Clang 6.0 svn316959LLVM Clang 5.0GCC 8.0.0 20171030GCC 7.2.0300K600K900K1200K1500KSE +/- 13710.54, N = 3SE +/- 3063.36, N = 6SE +/- 3816.59, N = 3SE +/- 1582.98, N = 31260800.37199028.19193210.49192063.141. (CC) gcc options: -ggdb -rdynamic -lm -pthread

Phoronix Test Suite v10.8.5