Ryzen 9 3900X Znver2 Compiler Tuning

AMD Ryzen 9 3900X 12-Core testing of GCC 9 and GCC 10 development with Znver2 tuning following recent cost table updates, etc. Benchmarks by Michael Larabel for a future article..

HTML result view exported from: https://openbenchmarking.org/result/1907290-HV-RYZEN939034&obr_sgm=y&grs.

Ryzen 9 3900X Znver2 Compiler TuningProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerDisplay DriverOpenGLCompilerFile-SystemScreen ResolutionGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2AMD Ryzen 9 3900X 12-Core @ 3.80GHz (12 Cores / 24 Threads)ASUS ROG CROSSHAIR VIII HERO (WI-FI) (0702 BIOS)AMD Device 148016384MB2000GB Force MP600Sapphire AMD Radeon RX 550 640SP / 560/560X 4GB (1300/1750MHz)AMD Device aae0ASUS VP28URealtek Device 8125 + Intel I211 + Intel Device 2723Ubuntu 18.045.3.0-999-generic (x86_64) 20190725GNOME Shell 3.28.4X Server 1.20.4modesetting 1.20.44.5 Mesa 19.0.2 (LLVM 8.0.0)GCC 9.1.0ext43840x2160GCC 10.0.0 20190727OpenBenchmarking.orgEnvironment Details- GCC 9.1.0: CXXFLAGS=-O3 CFLAGS=-O3- GCC 9.1.0 znver2: CXXFLAGS=-O3-march=znver2 CFLAGS=-O3-march=znver2- GCC 10.0.0: CXXFLAGS=-O3 CFLAGS=-O3- GCC 10.0.0 znver2: CXXFLAGS=-O3-march=znver2 CFLAGS=-O3-march=znver2Compiler Details- --disable-multilib --enable-checking=releaseProcessor Details- Scaling Governor: acpi-cpufreq ondemandPython Details- Python 2.7.15+ + Python 3.6.8Security Details- l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: always-on RSB filling

Ryzen 9 3900X Znver2 Compiler Tuningscimark2: Dense LU Matrix Factorizationscimark2: Compositefftw: Stock - 2D FFT Size 32fftw: Stock - 2D FFT Size 512fftw: Stock - 1D FFT Size 32aom-av1: AV1 Video Encodingscimark2: Fast Fourier Transformscimark2: Jacobi Successive Over-Relaxationfftw: Stock - 2D FFT Size 4096c-ray: Total Time - 4K, 16 Rays Per Pixelbullet: 1000 Stackaobench: 2048 x 2048 - Total Timegraphics-magick: Sharpenhpcc: G-Ptranslzbench: XZ 0 - Compressionbullet: Prim Trimeshtscp: AI Chess Performancebullet: 136 Ragdollsscimark2: Sparse Matrix Multiplylzbench: Libdeflate 1 - Compressionbullet: 3000 Fallsvt-vp9: 1080p 8-bit YUV To VP9 Video Encodelzbench: XZ 0 - Decompressionencode-mp3: WAV To MP3graphics-magick: Enhancedbuild-llvm: Time To Compilecpp-perf-bench: Atolbullet: Convex Trimeshcpp-perf-bench: Function Objectsbullet: Raytestscpp-perf-bench: Rand Numbersbullet: 1000 Convexencode-ogg: WAV To Oggredis: SETgraphics-magick: Rotatelzbench: Libdeflate 1 - Decompressionlzbench: Brotli 0 - Compressionencode-flac: WAV To FLACscimark2: Monte Carlomcperf: Getgraphics-magick: HWB Color Spacegraphics-magick: Swirlmkl-dnn: IP Batch All - f32cpp-perf-bench: Stepanov Abstractionhimeno: Poisson Pressure Solvermkl-dnn: Deconvolution Batch deconv_all - f32graphics-magick: Resizingcpp-perf-bench: Stepanov Vectorsmallpt: Global Illumination Renderer; 128 Samplessockperf: Latency Ping Pongmkl-dnn: Deconvolution Batch deconv_1d - f32mkl-dnn: Deconvolution Batch deconv_3d - f32tjbench: Decompression Throughputlzbench: Zstd 1 - Compressionbuild-php: Time To Compilehpcc: Rand Ring Bandwidthfftw: Float + SSE - 2D FFT Size 32lzbench: Zstd 1 - Decompressionsockperf: Throughputmpcbench: Multi-Precision Benchmarkhpcc: G-Fftehpcc: G-Fftecpuminer-opt: lbrycoremark: CoreMark Size 666 - Iterations Per Secondcpp-perf-bench: Math Librarygromacs: Water Benchmarkhpcc: Rand Ring Latencycpuminer-opt: sha256tgraphics-magick: Noise-Gaussianhpcc: Max Ping Pong Bandwidthmkl-dnn: Convolution Batch conv_3d - f32ffmpeg: H.264 HD To NTSC DVmkl-dnn: Convolution Batch conv_alexnet - f32hpcc: EP-STREAM Triadapache: Static Web Page Servingsvt-hevc: 1080p 8-bit YUV To HEVC Video Encodex265: H.265 1080p Video Encodinghpcc: G-HPLcpuminer-opt: skeinx264: H.264 Video Encodingopenssl: RSA 4096-bit Performancenginx: Static Web Page Servingmkl-dnn: Convolution Batch conv_all - f32pgbench: Buffer Test - Normal Load - Read Onlyhpcg: stockfish: Total Timejohn-the-ripper: Blowfishcpuminer-opt: myr-grhpcc: EP-DGEMMpgbench: Buffer Test - Normal Load - Read Writemkl-dnn: Convolution Batch conv_googlenet_v3 - f32cpp-perf-bench: Ctypecompress-xz: Compressing ubuntu-16.04.3-server-i386.img, Compression Level 9svt-av1: 1080p 8-bit YUV To AV1 Video Encodecpuminer-opt: m7mhpcc: G-Rand Accessm-queens: Time To Solveapache-siege: 250apache-siege: 200mcperf: Setredis: GETcpuminer-opt: deepmkl-dnn: IP Batch 1D - f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver26891.372768.16119099028.10129580.27295.182125.267063.0343.093.8834.601812.73255390.7513057812.053767.632393.2089.991167.25209280.2759.310.8914.401.98751.153.515.132122162.9426211194947.70761.3892376.402872511523.5227.601322.9051813.0527576.457.783.12221.2359.00218.0946852.714.8916145253126951455195978.598038.5980334583567987.34309.360.980.325968795117024227.253117.606.862527.501.7082038392.29246.0152.9470.9690039397139.593516.2739734.8519613.70300353.091.0939278964203351412732.8308329178.231147.6231.5225.2346.45593.890.0975747.1298050.9160835.7952914.103297713.3311190152.5111370.273686.601431410814118280.31273.492408.267920.1739.423.7733.201952.95225400.7713371882.043580.732573.2296.541166.94221284.1060.380.9014.152.04750.663.575.052169531.0026311835157.99800.2393850.592932591599.6828.931378.4652238.8028074.087.673.15217.0257.97225.6446853.914.9883244951126851709595778.605148.6051434420555154.60302.820.990.326988723817123832.608118.476.832520.011.7166838022.79247.3352.5371.7766339797138.413481.5039602.4919696.57297539.891.0839561655202531402332.6026329149.201153.4631.4325.2546.39590.660.0979847.2796842.1399824.4959232.073066070.2810230.34155.348526.663127.49129029583.73127480.27301.172175.857071.3042.634.1135.981812.72974400.8113660172.203856.632503.4489.841137.28208292.5359.970.9515.102.11799.883.735.052051361.3326211475077.72777.1795710.602882541582.7828.191385.2350039.1327474.267.843.03212.8358.16220.3346754.434.9494745361128751474895808.817488.8174835288568329.00307.230.970.331868641717023885.438116.626.882507.161.7220538490.98248.8553.0071.0701039720138.743492.5339525.7019803.57300244.811.0939631993204261413032.8634329372.391145.0131.5125.2646.22591.320.0977847.1462725.2482293.1457193.253042507.4711137154.0910777.883553.671411910531141130.32261.102293.467823.2739.363.8533.051962.94730370.7714087522.053675.942483.2792.351087.45223300.3163.340.9114.902.06787.773.605.362084989.8827711594998.11759.9797228.273022641556.9128.301385.8850679.5328677.227.533.04218.6656.87225.4445353.765.0460346305125052965793578.637948.6379434630567096.65306.020.980.325218644017323993.043118.026.782543.931.7305538009.25247.9952.4071.0493039843139.823487.1039346.9119694.33298969.751.0839540328204261413732.8436329148.601145.9531.3025.3946.49590.800.0977147.21102423.0783275.0652910.873031706.2211123157.72OpenBenchmarking.org

SciMark

Computational Test: Dense LU Matrix Factorization

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Dense LU Matrix FactorizationGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver22K4K6K8K10KSE +/- 60.72, N = 3SE +/- 15.98, N = 3SE +/- 19.91, N = 3SE +/- 12.24, N = 36891.3711370.278526.6610777.88-march=znver2-march=znver21. (CC) gcc options: -O3 -lm

SciMark

Computational Test: Composite

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: CompositeGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver28001600240032004000SE +/- 25.64, N = 3SE +/- 5.91, N = 3SE +/- 5.96, N = 3SE +/- 13.97, N = 32768.163686.603127.493553.67-march=znver2-march=znver21. (CC) gcc options: -O3 -lm

FFTW

Build: Stock - Size: 2D FFT Size 32

OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Stock - Size: 2D FFT Size 32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver23K6K9K12K15KSE +/- 155.95, N = 3SE +/- 141.66, N = 3SE +/- 2.19, N = 311909143141290214119-march=znver2-march=znver21. (CC) gcc options: -pthread -O3 -lm

FFTW

Build: Stock - Size: 2D FFT Size 512

OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Stock - Size: 2D FFT Size 512GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver22K4K6K8K10KSE +/- 30.08, N = 3SE +/- 148.34, N = 4SE +/- 19.17, N = 3SE +/- 10.67, N = 39028.1010814.009583.7310531.00-march=znver2-march=znver21. (CC) gcc options: -pthread -O3 -lm

FFTW

Build: Stock - Size: 1D FFT Size 32

OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Stock - Size: 1D FFT Size 32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver23K6K9K12K15KSE +/- 1.76, N = 3SE +/- 15.90, N = 3SE +/- 110.06, N = 3SE +/- 5.51, N = 312958118281274814113-march=znver2-march=znver21. (CC) gcc options: -pthread -O3 -lm

AOM AV1

AV1 Video Encoding

OpenBenchmarking.orgFrames Per Second, More Is BetterAOM AV1 2019-02-11AV1 Video EncodingGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.0720.1440.2160.2880.36SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 30.270.310.270.32-march=znver2-march=znver21. (CXX) g++ options: -O3 -std=c++11 -U_FORTIFY_SOURCE -lm -lpthread

SciMark

Computational Test: Fast Fourier Transform

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Fast Fourier TransformGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver270140210280350SE +/- 2.85, N = 3SE +/- 0.21, N = 3SE +/- 0.54, N = 3SE +/- 0.24, N = 3295.18273.49301.17261.10-march=znver2-march=znver21. (CC) gcc options: -O3 -lm

SciMark

Computational Test: Jacobi Successive Over-Relaxation

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Jacobi Successive Over-RelaxationGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver25001000150020002500SE +/- 20.16, N = 3SE +/- 0.53, N = 3SE +/- 0.16, N = 3SE +/- 0.89, N = 32125.262408.262175.852293.46-march=znver2-march=znver21. (CC) gcc options: -O3 -lm

FFTW

Build: Stock - Size: 2D FFT Size 4096

OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Stock - Size: 2D FFT Size 4096GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver22K4K6K8K10KSE +/- 73.85, N = 3SE +/- 67.62, N = 3SE +/- 40.30, N = 3SE +/- 95.02, N = 37063.037920.177071.307823.27-march=znver2-march=znver21. (CC) gcc options: -pthread -O3 -lm

C-Ray

Total Time - 4K, 16 Rays Per Pixel

OpenBenchmarking.orgSeconds, Fewer Is BetterC-Ray 1.1Total Time - 4K, 16 Rays Per PixelGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21020304050SE +/- 0.02, N = 3SE +/- 0.04, N = 3SE +/- 0.06, N = 3SE +/- 0.05, N = 343.0939.4242.6339.36-march=znver2-march=znver21. (CC) gcc options: -lm -lpthread -O3

Bullet Physics Engine

Test: 1000 Stack

OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: 1000 StackGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.92481.84962.77443.69924.624SE +/- 0.02, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.04, N = 33.883.774.113.85-march=znver2-march=znver21. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

AOBench

Size: 2048 x 2048 - Total Time

OpenBenchmarking.orgSeconds, Fewer Is BetterAOBenchSize: 2048 x 2048 - Total TimeGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2816243240SE +/- 0.05, N = 3SE +/- 0.10, N = 3SE +/- 0.06, N = 3SE +/- 0.02, N = 334.6033.2035.9833.05-march=znver2-march=znver21. (CC) gcc options: -lm -O3

GraphicsMagick

Operation: Sharpen

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.30Operation: SharpenGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver24080120160200SE +/- 0.33, N = 3181195181196-march=znver2-march=znver21. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -lwebpmux -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

HPC Challenge

Test / Class: G-Ptrans

OpenBenchmarking.orgGB/s, More Is BetterHPC Challenge 1.5.0Test / Class: G-PtransGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.66431.32861.99292.65723.3215SE +/- 0.00047, N = 3SE +/- 0.00095, N = 3SE +/- 0.00151, N = 3SE +/- 0.00082, N = 32.732552.952252.729742.94730-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

lzbench

Test: XZ 0 - Process: Compression

OpenBenchmarking.orgMB/s, More Is Betterlzbench 2017-08-08Test: XZ 0 - Process: CompressionGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2918273645SE +/- 0.33, N = 3394040371. (CXX) g++ options: -lrt -static -lpthread -fomit-frame-pointer -fstrict-aliasing -ffast-math -O3

Bullet Physics Engine

Test: Prim Trimesh

OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: Prim TrimeshGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.18230.36460.54690.72920.9115SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.01, N = 30.750.770.810.77-march=znver2-march=znver21. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

TSCP

AI Chess Performance

OpenBenchmarking.orgNodes Per Second, More Is BetterTSCP 1.81AI Chess PerformanceGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2300K600K900K1200K1500KSE +/- 620.00, N = 5SE +/- 10688.23, N = 5SE +/- 676.60, N = 5SE +/- 6261.48, N = 51305781133718813660171408752-march=znver2-march=znver21. (CC) gcc options: -O3 -march=native

Bullet Physics Engine

Test: 136 Ragdolls

OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: 136 RagdollsGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.4950.991.4851.982.475SE +/- 0.01, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.02, N = 32.052.042.202.05-march=znver2-march=znver21. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

SciMark

Computational Test: Sparse Matrix Multiply

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Sparse Matrix MultiplyGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver28001600240032004000SE +/- 37.65, N = 3SE +/- 13.73, N = 3SE +/- 15.78, N = 3SE +/- 58.43, N = 33767.633580.733856.633675.94-march=znver2-march=znver21. (CC) gcc options: -O3 -lm

lzbench

Test: Libdeflate 1 - Process: Compression

OpenBenchmarking.orgMB/s, More Is Betterlzbench 2017-08-08Test: Libdeflate 1 - Process: CompressionGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver260120180240300SE +/- 1.86, N = 3SE +/- 0.67, N = 32392572502481. (CXX) g++ options: -lrt -static -lpthread -fomit-frame-pointer -fstrict-aliasing -ffast-math -O3

Bullet Physics Engine

Test: 3000 Fall

OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: 3000 FallGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.7741.5482.3223.0963.87SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.00, N = 3SE +/- 0.03, N = 33.203.223.443.27-march=znver2-march=znver21. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

SVT-VP9

1080p 8-bit YUV To VP9 Video Encode

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-VP9 2019-02-171080p 8-bit YUV To VP9 Video EncodeGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver220406080100SE +/- 0.28, N = 3SE +/- 0.08, N = 3SE +/- 0.15, N = 3SE +/- 0.19, N = 389.9996.5489.8492.35-march=znver2-march=znver21. (CC) gcc options: -O3 -fPIE -fPIC -O2 -flto -fvisibility=hidden -mavx -pie -rdynamic -lpthread -lrt -lm

lzbench

Test: XZ 0 - Process: Decompression

OpenBenchmarking.orgMB/s, More Is Betterlzbench 2017-08-08Test: XZ 0 - Process: DecompressionGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2306090120150SE +/- 0.33, N = 31161161131081. (CXX) g++ options: -lrt -static -lpthread -fomit-frame-pointer -fstrict-aliasing -ffast-math -O3

LAME MP3 Encoding

WAV To MP3

OpenBenchmarking.orgSeconds, Fewer Is BetterLAME MP3 Encoding 3.100WAV To MP3GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2246810SE +/- 0.04, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.01, N = 37.256.947.287.45-march=znver2-march=znver21. (CC) gcc options: -O3 -lncurses -lm

GraphicsMagick

Operation: Enhanced

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.30Operation: EnhancedGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver250100150200250SE +/- 1.20, N = 3209221208223-march=znver2-march=znver21. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -lwebpmux -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

Timed LLVM Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed LLVM Compilation 6.0.1Time To CompileGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver270140210280350280.27284.10292.53300.31

CppPerformanceBenchmarks

Test: Atol

OpenBenchmarking.orgSeconds, Fewer Is BetterCppPerformanceBenchmarks 9Test: AtolGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21428425670SE +/- 0.30, N = 3SE +/- 0.53, N = 11SE +/- 0.17, N = 3SE +/- 0.06, N = 359.3160.3859.9763.34-march=znver2-march=znver21. (CXX) g++ options: -O3 -std=c++11

Bullet Physics Engine

Test: Convex Trimesh

OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: Convex TrimeshGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.21380.42760.64140.85521.069SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.01, N = 30.890.900.950.91-march=znver2-march=znver21. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

CppPerformanceBenchmarks

Test: Function Objects

OpenBenchmarking.orgSeconds, Fewer Is BetterCppPerformanceBenchmarks 9Test: Function ObjectsGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver248121620SE +/- 0.08, N = 3SE +/- 0.03, N = 3SE +/- 0.17, N = 3SE +/- 0.20, N = 414.4014.1515.1014.90-march=znver2-march=znver21. (CXX) g++ options: -O3 -std=c++11

Bullet Physics Engine

Test: Raytests

OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: RaytestsGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.47480.94961.42441.89922.374SE +/- 0.01, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 6SE +/- 0.02, N = 31.982.042.112.06-march=znver2-march=znver21. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

CppPerformanceBenchmarks

Test: Random Numbers

OpenBenchmarking.orgSeconds, Fewer Is BetterCppPerformanceBenchmarks 9Test: Random NumbersGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver22004006008001000SE +/- 2.69, N = 3SE +/- 0.27, N = 3SE +/- 4.15, N = 3SE +/- 10.35, N = 5751.15750.66799.88787.77-march=znver2-march=znver21. (CXX) g++ options: -O3 -std=c++11

Bullet Physics Engine

Test: 1000 Convex

OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: 1000 ConvexGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.83931.67862.51793.35724.1965SE +/- 0.02, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.04, N = 33.513.573.733.60-march=znver2-march=znver21. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

Ogg Encoding

WAV To Ogg

OpenBenchmarking.orgSeconds, Fewer Is BetterOgg Encoding 1.3.3WAV To OggGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21.2062.4123.6184.8246.03SE +/- 0.00, N = 3SE +/- 0.01, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 45.135.055.055.36-march=znver2-march=znver21. (CC) gcc options: -O2 -ffast-math -fsigned-char -O3 -logg

Redis

Test: SET

OpenBenchmarking.orgRequests Per Second, More Is BetterRedis 4.0.8Test: SETGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2500K1000K1500K2000K2500KSE +/- 30290.32, N = 4SE +/- 19021.82, N = 3SE +/- 28123.08, N = 3SE +/- 14796.01, N = 32122162.942169531.002051361.332084989.881. (CC) gcc options: -ggdb -rdynamic -lm -ldl -pthread

GraphicsMagick

Operation: Rotate

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.30Operation: RotateGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver260120180240300SE +/- 1.20, N = 3SE +/- 4.33, N = 3SE +/- 1.86, N = 3262263262277-march=znver2-march=znver21. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -lwebpmux -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

lzbench

Test: Libdeflate 1 - Process: Decompression

OpenBenchmarking.orgMB/s, More Is Betterlzbench 2017-08-08Test: Libdeflate 1 - Process: DecompressionGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver230060090012001500SE +/- 0.33, N = 3SE +/- 10.00, N = 311191183114711591. (CXX) g++ options: -lrt -static -lpthread -fomit-frame-pointer -fstrict-aliasing -ffast-math -O3

lzbench

Test: Brotli 0 - Process: Compression

OpenBenchmarking.orgMB/s, More Is Betterlzbench 2017-08-08Test: Brotli 0 - Process: CompressionGCC 10.0.0 znver2GCC 10.0.0GCC 9.1.0GCC 9.1.0 znver2110220330440550SE +/- 0.88, N = 3SE +/- 0.88, N = 3SE +/- 0.67, N = 3SE +/- 4.10, N = 35055224945151. (CXX) g++ options: -lrt -static -lpthread -fomit-frame-pointer -fstrict-aliasing -ffast-math -O3

FLAC Audio Encoding

WAV To FLAC

OpenBenchmarking.orgSeconds, Fewer Is BetterFLAC Audio Encoding 1.3.2WAV To FLACGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2246810SE +/- 0.02, N = 5SE +/- 0.01, N = 5SE +/- 0.03, N = 5SE +/- 0.04, N = 57.707.997.728.11-march=znver2-march=znver21. (CXX) g++ options: -O3 -fvisibility=hidden -logg -lm

SciMark

Computational Test: Monte Carlo

OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Monte CarloGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver22004006008001000SE +/- 7.16, N = 3SE +/- 0.74, N = 3SE +/- 0.24, N = 3SE +/- 0.29, N = 3761.38800.23777.17759.97-march=znver2-march=znver21. (CC) gcc options: -O3 -lm

Memcached mcperf

Method: Get

OpenBenchmarking.orgOperations Per Second, More Is BetterMemcached mcperf 1.5.10Method: GetGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver220K40K60K80K100KSE +/- 1551.16, N = 3SE +/- 937.65, N = 15SE +/- 1025.20, N = 15SE +/- 1267.59, N = 392376.4093850.5995710.6097228.27-march=znver2-march=znver21. (CC) gcc options: -O3 -lm -rdynamic

GraphicsMagick

Operation: HWB Color Space

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.30Operation: HWB Color SpaceGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver270140210280350SE +/- 2.60, N = 3SE +/- 2.19, N = 3SE +/- 2.19, N = 3SE +/- 0.33, N = 3287293288302-march=znver2-march=znver21. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -lwebpmux -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

GraphicsMagick

Operation: Swirl

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.30Operation: SwirlGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver260120180240300SE +/- 1.86, N = 3SE +/- 1.20, N = 3SE +/- 0.88, N = 3251259254264-march=znver2-march=znver21. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -lwebpmux -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

MKL-DNN

Harness: IP Batch All - Data Type: f32

OpenBenchmarking.orgms, Fewer Is BetterMKL-DNN 2019-04-16Harness: IP Batch All - Data Type: f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver230060090012001500SE +/- 7.48, N = 3SE +/- 17.21, N = 3SE +/- 5.99, N = 3SE +/- 25.50, N = 31523.521599.681582.781556.91MIN: 1357.02-march=znver2 - MIN: 1393.73MIN: 1385.56-march=znver2 - MIN: 1368.21. (CXX) g++ options: -O3 -std=c++11 -march=native -mtune=native -fPIC -fopenmp -pie -lmklml_intel -ldl

CppPerformanceBenchmarks

Test: Stepanov Abstraction

OpenBenchmarking.orgSeconds, Fewer Is BetterCppPerformanceBenchmarks 9Test: Stepanov AbstractionGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2714212835SE +/- 0.03, N = 3SE +/- 0.00, N = 3SE +/- 0.08, N = 3SE +/- 0.45, N = 327.6028.9328.1928.30-march=znver2-march=znver21. (CXX) g++ options: -O3 -std=c++11

Himeno Benchmark

Poisson Pressure Solver

OpenBenchmarking.orgMFLOPS, More Is BetterHimeno Benchmark 3.0Poisson Pressure SolverGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver230060090012001500SE +/- 11.21, N = 3SE +/- 6.19, N = 3SE +/- 2.93, N = 3SE +/- 0.48, N = 31322.901378.461385.231385.88-march=znver2-march=znver21. (CC) gcc options: -O3 -mavx2

MKL-DNN

Harness: Deconvolution Batch deconv_all - Data Type: f32

OpenBenchmarking.orgms, Fewer Is BetterMKL-DNN 2019-04-16Harness: Deconvolution Batch deconv_all - Data Type: f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver211K22K33K44K55KSE +/- 589.20, N = 6SE +/- 668.40, N = 3SE +/- 224.88, N = 3SE +/- 390.75, N = 351813.0552238.8050039.1350679.53MIN: 48543.1-march=znver2 - MIN: 49224.9MIN: 46883.1-march=znver2 - MIN: 48056.61. (CXX) g++ options: -O3 -std=c++11 -march=native -mtune=native -fPIC -fopenmp -pie -lmklml_intel -ldl

GraphicsMagick

Operation: Resizing

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.30Operation: ResizingGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver260120180240300SE +/- 2.19, N = 3SE +/- 1.15, N = 3SE +/- 2.65, N = 3SE +/- 1.53, N = 3275280274286-march=znver2-march=znver21. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -lwebpmux -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

CppPerformanceBenchmarks

Test: Stepanov Vector

OpenBenchmarking.orgSeconds, Fewer Is BetterCppPerformanceBenchmarks 9Test: Stepanov VectorGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver220406080100SE +/- 0.35, N = 3SE +/- 0.12, N = 3SE +/- 0.88, N = 3SE +/- 0.04, N = 376.4574.0874.2677.22-march=znver2-march=znver21. (CXX) g++ options: -O3 -std=c++11

Smallpt

Global Illumination Renderer; 128 Samples

OpenBenchmarking.orgSeconds, Fewer Is BetterSmallpt 1.0Global Illumination Renderer; 128 SamplesGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2246810SE +/- 0.00, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 37.787.677.847.53-march=znver2-march=znver21. (CXX) g++ options: -fopenmp -O3

Sockperf

Test: Latency Ping Pong

OpenBenchmarking.orgusec, Fewer Is BetterSockperf 3.4Test: Latency Ping PongGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.70881.41762.12642.83523.544SE +/- 0.04, N = 5SE +/- 0.04, N = 6SE +/- 0.02, N = 25SE +/- 0.02, N = 253.123.153.033.04-march=znver2-march=znver21. (CXX) g++ options: --param -O3 -rdynamic -ldl -lpthread

MKL-DNN

Harness: Deconvolution Batch deconv_1d - Data Type: f32

OpenBenchmarking.orgms, Fewer Is BetterMKL-DNN 2019-04-16Harness: Deconvolution Batch deconv_1d - Data Type: f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver250100150200250SE +/- 2.00, N = 15SE +/- 1.79, N = 13SE +/- 0.29, N = 3SE +/- 1.85, N = 15221.23217.02212.83218.66MIN: 202.07-march=znver2 - MIN: 203.65MIN: 201.7-march=znver2 - MIN: 203.421. (CXX) g++ options: -O3 -std=c++11 -march=native -mtune=native -fPIC -fopenmp -pie -lmklml_intel -ldl

MKL-DNN

Harness: Deconvolution Batch deconv_3d - Data Type: f32

OpenBenchmarking.orgms, Fewer Is BetterMKL-DNN 2019-04-16Harness: Deconvolution Batch deconv_3d - Data Type: f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21326395265SE +/- 0.66, N = 7SE +/- 0.49, N = 15SE +/- 0.69, N = 15SE +/- 0.58, N = 859.0057.9758.1656.87MIN: 50.8-march=znver2 - MIN: 51.57MIN: 50.91-march=znver2 - MIN: 50.961. (CXX) g++ options: -O3 -std=c++11 -march=native -mtune=native -fPIC -fopenmp -pie -lmklml_intel -ldl

libjpeg-turbo tjbench

Test: Decompression Throughput

OpenBenchmarking.orgMegapixels/sec, More Is Betterlibjpeg-turbo tjbench 2.0.2Test: Decompression ThroughputGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver250100150200250SE +/- 0.44, N = 3SE +/- 0.31, N = 3SE +/- 2.32, N = 3SE +/- 0.30, N = 3218.09225.64220.33225.44-march=znver2-march=znver21. (CC) gcc options: -O3 -rdynamic

lzbench

Test: Zstd 1 - Process: Compression

OpenBenchmarking.orgMB/s, More Is Betterlzbench 2017-08-08Test: Zstd 1 - Process: CompressionGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2100200300400500SE +/- 3.18, N = 3SE +/- 4.91, N = 8SE +/- 0.33, N = 34684684674531. (CXX) g++ options: -lrt -static -lpthread -fomit-frame-pointer -fstrict-aliasing -ffast-math -O3

Timed PHP Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed PHP Compilation 7.1.9Time To CompileGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21224364860SE +/- 0.15, N = 3SE +/- 0.36, N = 3SE +/- 0.09, N = 3SE +/- 0.51, N = 352.7153.9154.4353.76-march=znver2-march=znver21. (CC) gcc options: -O3 -pedantic -ldl -lz -lm

HPC Challenge

Test / Class: Random Ring Bandwidth

OpenBenchmarking.orgGB/s, More Is BetterHPC Challenge 1.5.0Test / Class: Random Ring BandwidthGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21.13542.27083.40624.54165.677SE +/- 0.02698, N = 3SE +/- 0.07571, N = 3SE +/- 0.05697, N = 3SE +/- 0.04322, N = 34.891614.988324.949475.04603-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

FFTW

Build: Float + SSE - Size: 2D FFT Size 32

OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Float + SSE - Size: 2D FFT Size 32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver210K20K30K40K50KSE +/- 663.38, N = 4SE +/- 105.51, N = 3SE +/- 54.85, N = 3SE +/- 28.47, N = 345253449514536146305-march=znver2-march=znver21. (CC) gcc options: -pthread -O3 -lm

lzbench

Test: Zstd 1 - Process: Decompression

OpenBenchmarking.orgMB/s, More Is Betterlzbench 2017-08-08Test: Zstd 1 - Process: DecompressionGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver230060090012001500SE +/- 9.50, N = 3SE +/- 12.79, N = 8SE +/- 0.58, N = 3SE +/- 0.33, N = 312691268128712501. (CXX) g++ options: -lrt -static -lpthread -fomit-frame-pointer -fstrict-aliasing -ffast-math -O3

Sockperf

Test: Throughput

OpenBenchmarking.orgMessages Per Second, More Is BetterSockperf 3.4Test: ThroughputGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2110K220K330K440K550KSE +/- 4767.11, N = 5SE +/- 4175.03, N = 5SE +/- 5409.10, N = 5SE +/- 3715.76, N = 18514551517095514748529657-march=znver2-march=znver21. (CXX) g++ options: --param -O3 -rdynamic -ldl -lpthread

GNU MPC

Multi-Precision Benchmark

OpenBenchmarking.orgGlobal Score, More Is BetterGNU MPC 1.1.0Multi-Precision BenchmarkGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver22K4K6K8K10KSE +/- 31.80, N = 3SE +/- 50.44, N = 3SE +/- 26.46, N = 3SE +/- 102.03, N = 39597957795809357-march=znver2-march=znver21. (CC) gcc options: -lm -O3 -MT -MD -MP -MF

HPC Challenge

Test / Class: G-Ffte

OpenBenchmarking.orgGFLOP/s, More Is BetterHPC Challenge 1.5.0Test / Class: G-FfteGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2246810SE +/- 0.02013, N = 3SE +/- 0.06300, N = 3SE +/- 0.18198, N = 3SE +/- 0.02559, N = 38.598038.605148.817488.63794-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

HPC Challenge

Test / Class: G-Ffte

OpenBenchmarking.orgGFLOPS, More Is BetterHPC Challenge 1.5.0Test / Class: G-FfteGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2246810SE +/- 0.02013, N = 3SE +/- 0.06300, N = 3SE +/- 0.18198, N = 3SE +/- 0.02559, N = 38.598038.605148.817488.63794-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

Cpuminer-Opt

Algorithm: lbry

OpenBenchmarking.orgkH/s - Hash Speed, More Is BetterCpuminer-Opt 3.8.8.1Algorithm: lbryGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver28K16K24K32K40KSE +/- 550.28, N = 3SE +/- 5.77, N = 3SE +/- 460.86, N = 5SE +/- 20.82, N = 334583344203528834630-march=znver2-march=znver21. (CXX) g++ options: -O3 -lcurl -lz -lpthread -lssl -lcrypto -lgmp

Coremark

CoreMark Size 666 - Iterations Per Second

OpenBenchmarking.orgIterations/Sec, More Is BetterCoremark 1.0CoreMark Size 666 - Iterations Per SecondGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2120K240K360K480K600KSE +/- 1430.19, N = 3SE +/- 2761.64, N = 3SE +/- 1210.22, N = 3SE +/- 1036.74, N = 3567987.34555154.60568329.00567096.65-march=znver2-march=znver21. (CC) gcc options: -O2 -O3 -lrt" -lrt

CppPerformanceBenchmarks

Test: Math Library

OpenBenchmarking.orgSeconds, Fewer Is BetterCppPerformanceBenchmarks 9Test: Math LibraryGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver270140210280350SE +/- 0.26, N = 3SE +/- 3.91, N = 3SE +/- 4.29, N = 3SE +/- 2.37, N = 3309.36302.82307.23306.02-march=znver2-march=znver21. (CXX) g++ options: -O3 -std=c++11

GROMACS

Water Benchmark

OpenBenchmarking.orgNs Per Day, More Is BetterGROMACS 2018.3Water BenchmarkGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.22280.44560.66840.89121.114SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 30.980.990.970.98-march=znver2-march=znver21. (CXX) g++ options: -march=core-avx2 -O3 -std=c++11 -funroll-all-loops -fopenmp -lrt -lpthread -lm

HPC Challenge

Test / Class: Random Ring Latency

OpenBenchmarking.orgusecs, Fewer Is BetterHPC Challenge 1.5.0Test / Class: Random Ring LatencyGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.07470.14940.22410.29880.3735SE +/- 0.00047, N = 3SE +/- 0.00042, N = 3SE +/- 0.00125, N = 3SE +/- 0.00071, N = 30.325960.326980.331860.32521-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

Cpuminer-Opt

Algorithm: sha256t

OpenBenchmarking.orgkH/s - Hash Speed, More Is BetterCpuminer-Opt 3.8.8.1Algorithm: sha256tGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver220K40K60K80K100KSE +/- 990.16, N = 7SE +/- 1027.26, N = 6SE +/- 116.81, N = 3SE +/- 180.83, N = 387951872388641786440-march=znver2-march=znver21. (CXX) g++ options: -O3 -lcurl -lz -lpthread -lssl -lcrypto -lgmp

GraphicsMagick

Operation: Noise-Gaussian

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.30Operation: Noise-GaussianGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver24080120160200SE +/- 0.67, N = 3SE +/- 0.58, N = 3SE +/- 0.33, N = 3170171170173-march=znver2-march=znver21. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -lwebpmux -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

HPC Challenge

Test / Class: Max Ping Pong Bandwidth

OpenBenchmarking.orgMB/s, More Is BetterHPC Challenge 1.5.0Test / Class: Max Ping Pong BandwidthGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver25K10K15K20K25KSE +/- 159.70, N = 3SE +/- 119.42, N = 3SE +/- 62.37, N = 3SE +/- 195.64, N = 324227.2523832.6123885.4423993.04-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

MKL-DNN

Harness: Convolution Batch conv_3d - Data Type: f32

OpenBenchmarking.orgms, Fewer Is BetterMKL-DNN 2019-04-16Harness: Convolution Batch conv_3d - Data Type: f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2306090120150SE +/- 0.16, N = 3SE +/- 0.45, N = 3SE +/- 0.79, N = 3SE +/- 1.48, N = 4117.60118.47116.62118.02MIN: 103.13-march=znver2 - MIN: 103.47MIN: 102.39-march=znver2 - MIN: 102.111. (CXX) g++ options: -O3 -std=c++11 -march=native -mtune=native -fPIC -fopenmp -pie -lmklml_intel -ldl

FFmpeg

H.264 HD To NTSC DV

OpenBenchmarking.orgSeconds, Fewer Is BetterFFmpeg 4.0.2H.264 HD To NTSC DVGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2246810SE +/- 0.01, N = 3SE +/- 0.03, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 36.866.836.886.78-march=znver2-march=znver21. (CC) gcc options: -lavdevice -lavfilter -lavformat -lavcodec -lswresample -lswscale -lavutil -lXv -lX11 -lXext -lm -lxcb -lxcb-shape -lxcb-xfixes -lasound -lSDL2 -lsndio -pthread -lbz2 -llzma -O3 -std=c11 -fomit-frame-pointer -fno-math-errno -fno-signed-zeros -fno-tree-vectorize -MMD -MF -MT

MKL-DNN

Harness: Convolution Batch conv_alexnet - Data Type: f32

OpenBenchmarking.orgms, Fewer Is BetterMKL-DNN 2019-04-16Harness: Convolution Batch conv_alexnet - Data Type: f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver25001000150020002500SE +/- 9.57, N = 3SE +/- 9.61, N = 3SE +/- 6.13, N = 3SE +/- 17.20, N = 32527.502520.012507.162543.93MIN: 2462.11-march=znver2 - MIN: 2467.07MIN: 2461.57-march=znver2 - MIN: 2467.761. (CXX) g++ options: -O3 -std=c++11 -march=native -mtune=native -fPIC -fopenmp -pie -lmklml_intel -ldl

HPC Challenge

Test / Class: EP-STREAM Triad

OpenBenchmarking.orgGB/s, More Is BetterHPC Challenge 1.5.0Test / Class: EP-STREAM TriadGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.38940.77881.16821.55761.947SE +/- 0.00015, N = 3SE +/- 0.00081, N = 3SE +/- 0.00098, N = 3SE +/- 0.00091, N = 31.708201.716681.722051.73055-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

Apache Benchmark

Static Web Page Serving

OpenBenchmarking.orgRequests Per Second, More Is BetterApache Benchmark 2.4.29Static Web Page ServingGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver28K16K24K32K40KSE +/- 57.64, N = 3SE +/- 139.15, N = 3SE +/- 65.39, N = 3SE +/- 79.10, N = 338392.2938022.7938490.9838009.25-march=znver2-march=znver21. (CC) gcc options: -shared -fPIC -pthread -O3

SVT-HEVC

1080p 8-bit YUV To HEVC Video Encode

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-HEVC 2019-02-031080p 8-bit YUV To HEVC Video EncodeGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver250100150200250SE +/- 3.72, N = 3SE +/- 0.72, N = 3SE +/- 1.73, N = 3SE +/- 1.78, N = 3246.01247.33248.85247.99-march=znver2-march=znver21. (CC) gcc options: -O3 -fPIE -fPIC -O2 -flto -fvisibility=hidden -march=native -pie -rdynamic -lpthread -lrt

x265

H.265 1080p Video Encoding

OpenBenchmarking.orgFrames Per Second, More Is Betterx265 3.0H.265 1080p Video EncodingGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21224364860SE +/- 0.28, N = 3SE +/- 0.06, N = 3SE +/- 0.20, N = 3SE +/- 0.19, N = 352.9452.5353.0052.40-march=znver2-march=znver21. (CXX) g++ options: -O3 -rdynamic -lpthread -lrt -ldl -lnuma

HPC Challenge

Test / Class: G-HPL

OpenBenchmarking.orgGFLOPS, More Is BetterHPC Challenge 1.5.0Test / Class: G-HPLGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21632486480SE +/- 0.23, N = 3SE +/- 0.37, N = 3SE +/- 0.08, N = 3SE +/- 0.22, N = 370.9771.7871.0771.05-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

Cpuminer-Opt

Algorithm: skein

OpenBenchmarking.orgkH/s - Hash Speed, More Is BetterCpuminer-Opt 3.8.8.1Algorithm: skeinGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver29K18K27K36K45KSE +/- 602.50, N = 3SE +/- 21.86, N = 3SE +/- 5.77, N = 3SE +/- 133.46, N = 339397397973972039843-march=znver2-march=znver21. (CXX) g++ options: -O3 -lcurl -lz -lpthread -lssl -lcrypto -lgmp

x264

H.264 Video Encoding

OpenBenchmarking.orgFrames Per Second, More Is Betterx264 2018-09-25H.264 Video EncodingGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2306090120150SE +/- 1.55, N = 7SE +/- 2.03, N = 3SE +/- 2.27, N = 3SE +/- 2.09, N = 4139.59138.41138.74139.82-march=znver2-march=znver21. (CC) gcc options: -ldl -m64 -lm -lpthread -O3 -ffast-math -std=gnu99 -fPIC -fomit-frame-pointer -fno-tree-vectorize

OpenSSL

RSA 4096-bit Performance

OpenBenchmarking.orgSigns Per Second, More Is BetterOpenSSL 1.1.1RSA 4096-bit PerformanceGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver28001600240032004000SE +/- 7.07, N = 3SE +/- 1.42, N = 3SE +/- 1.89, N = 3SE +/- 0.70, N = 33516.273481.503492.533487.10-march=znver2-march=znver21. (CC) gcc options: -pthread -m64 -O3 -lssl -lcrypto -ldl

NGINX Benchmark

Static Web Page Serving

OpenBenchmarking.orgRequests Per Second, More Is BetterNGINX Benchmark 1.9.9Static Web Page ServingGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver29K18K27K36K45KSE +/- 102.83, N = 3SE +/- 112.05, N = 3SE +/- 158.42, N = 3SE +/- 23.74, N = 339734.8539602.4939525.7039346.91-march=znver2-march=znver21. (CC) gcc options: -lpthread -lcrypt -lcrypto -lz -O3 -march=native

MKL-DNN

Harness: Convolution Batch conv_all - Data Type: f32

OpenBenchmarking.orgms, Fewer Is BetterMKL-DNN 2019-04-16Harness: Convolution Batch conv_all - Data Type: f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver24K8K12K16K20KSE +/- 42.61, N = 3SE +/- 22.35, N = 3SE +/- 87.03, N = 3SE +/- 41.11, N = 319613.7019696.5719803.5719694.33MIN: 18961.5-march=znver2 - MIN: 19033.5MIN: 19014.9-march=znver2 - MIN: 18995.61. (CXX) g++ options: -O3 -std=c++11 -march=native -mtune=native -fPIC -fopenmp -pie -lmklml_intel -ldl

PostgreSQL pgbench

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

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 10.3Scaling: Buffer Test - Test: Normal Load - Mode: Read OnlyGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver260K120K180K240K300KSE +/- 513.53, N = 3SE +/- 235.79, N = 3SE +/- 102.78, N = 3SE +/- 237.85, N = 3300353.09297539.89300244.81298969.75-march=znver2-march=znver21. (CC) gcc options: -fno-strict-aliasing -fwrapv -O3 -lpgcommon -lpgport -lpq -lpthread -lrt -lcrypt -ldl -lm

High Performance Conjugate Gradient

OpenBenchmarking.orgGFLOP/s, More Is BetterHigh Performance Conjugate Gradient 3.0GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.24530.49060.73590.98121.2265SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 4SE +/- 0.00, N = 31.091.081.091.08

Stockfish

Total Time

OpenBenchmarking.orgNodes Per Second, More Is BetterStockfish 9Total TimeGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver28M16M24M32M40MSE +/- 210046.69, N = 3SE +/- 164232.11, N = 3SE +/- 237875.03, N = 3SE +/- 131167.27, N = 339278964395616553963199339540328-march=znver2-march=znver21. (CXX) g++ options: -m64 -lpthread -O3 -fno-exceptions -std=c++11 -pedantic -msse -msse3 -mpopcnt -flto

John The Ripper

Test: Blowfish

OpenBenchmarking.orgReal C/S, More Is BetterJohn The Ripper 1.9.0-jumbo-1Test: BlowfishGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver24K8K12K16K20KSE +/- 64.93, N = 3SE +/- 63.01, N = 3SE +/- 63.74, N = 3SE +/- 64.22, N = 3203352025320426204261. (CC) gcc options: -m64 -lssl -lcrypto -fopenmp -lgmp -pthread -lm -lz -ldl -lcrypt -lbz2

Cpuminer-Opt

Algorithm: myr-gr

OpenBenchmarking.orgkH/s - Hash Speed, More Is BetterCpuminer-Opt 3.8.8.1Algorithm: myr-grGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver23K6K9K12K15KSE +/- 49.78, N = 3SE +/- 26.03, N = 3SE +/- 40.00, N = 3SE +/- 6.67, N = 314127140231413014137-march=znver2-march=znver21. (CXX) g++ options: -O3 -lcurl -lz -lpthread -lssl -lcrypto -lgmp

HPC Challenge

Test / Class: EP-DGEMM

OpenBenchmarking.orgGFLOPS, More Is BetterHPC Challenge 1.5.0Test / Class: EP-DGEMMGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2816243240SE +/- 0.19, N = 3SE +/- 0.42, N = 3SE +/- 0.11, N = 3SE +/- 0.22, N = 332.8332.6032.8632.84-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

PostgreSQL pgbench

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

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 10.3Scaling: Buffer Test - Test: Normal Load - Mode: Read WriteGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver26K12K18K24K30KSE +/- 55.36, N = 3SE +/- 40.41, N = 3SE +/- 31.16, N = 3SE +/- 124.84, N = 329178.2329149.2029372.3929148.60-march=znver2-march=znver21. (CC) gcc options: -fno-strict-aliasing -fwrapv -O3 -lpgcommon -lpgport -lpq -lpthread -lrt -lcrypt -ldl -lm

MKL-DNN

Harness: Convolution Batch conv_googlenet_v3 - Data Type: f32

OpenBenchmarking.orgms, Fewer Is BetterMKL-DNN 2019-04-16Harness: Convolution Batch conv_googlenet_v3 - Data Type: f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver22004006008001000SE +/- 6.51, N = 3SE +/- 6.23, N = 3SE +/- 6.39, N = 3SE +/- 5.61, N = 31147.621153.461145.011145.95MIN: 1052.13-march=znver2 - MIN: 1057.54MIN: 1050.58-march=znver2 - MIN: 1052.711. (CXX) g++ options: -O3 -std=c++11 -march=native -mtune=native -fPIC -fopenmp -pie -lmklml_intel -ldl

CppPerformanceBenchmarks

Test: Ctype

OpenBenchmarking.orgSeconds, Fewer Is BetterCppPerformanceBenchmarks 9Test: CtypeGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2714212835SE +/- 0.28, N = 3SE +/- 0.14, N = 3SE +/- 0.38, N = 5SE +/- 0.03, N = 331.5231.4331.5131.30-march=znver2-march=znver21. (CXX) g++ options: -O3 -std=c++11

XZ Compression

Compressing ubuntu-16.04.3-server-i386.img, Compression Level 9

OpenBenchmarking.orgSeconds, Fewer Is BetterXZ Compression 5.2.4Compressing ubuntu-16.04.3-server-i386.img, Compression Level 9GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2612182430SE +/- 0.10, N = 3SE +/- 0.13, N = 3SE +/- 0.12, N = 3SE +/- 0.11, N = 325.2325.2525.2625.39-march=znver2-march=znver21. (CC) gcc options: -pthread -fvisibility=hidden -O3

SVT-AV1

1080p 8-bit YUV To AV1 Video Encode

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 0.51080p 8-bit YUV To AV1 Video EncodeGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21122334455SE +/- 0.15, N = 3SE +/- 0.19, N = 3SE +/- 0.27, N = 3SE +/- 0.13, N = 346.4546.3946.2246.49-march=znver2-march=znver21. (CXX) g++ options: -O3 -pie -lpthread -lm

Cpuminer-Opt

Algorithm: m7m

OpenBenchmarking.orgkH/s - Hash Speed, More Is BetterCpuminer-Opt 3.8.8.1Algorithm: m7mGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2130260390520650SE +/- 0.29, N = 3SE +/- 0.15, N = 3SE +/- 0.27, N = 3SE +/- 0.35, N = 3593.89590.66591.32590.80-march=znver2-march=znver21. (CXX) g++ options: -O3 -lcurl -lz -lpthread -lssl -lcrypto -lgmp

HPC Challenge

Test / Class: G-Random Access

OpenBenchmarking.orgGUP/s, More Is BetterHPC Challenge 1.5.0Test / Class: G-Random AccessGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver20.0220.0440.0660.0880.11SE +/- 0.00036, N = 3SE +/- 0.00042, N = 3SE +/- 0.00041, N = 3SE +/- 0.00044, N = 30.097570.097980.097780.09771-march=znver2-march=znver21. (CC) gcc options: -lblas -lm -pthread -lmpi -fomit-frame-pointer -O3 -funroll-loops2. OpenBLAS + Open MPI 2.1.1

m-queens

Time To Solve

OpenBenchmarking.orgSeconds, Fewer Is Betterm-queens 1.2Time To SolveGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver21122334455SE +/- 0.12, N = 3SE +/- 0.08, N = 3SE +/- 0.11, N = 3SE +/- 0.10, N = 347.1247.2747.1447.21-march=znver2-march=znver21. (CXX) g++ options: -fopenmp -O3 -O2 -march=native

Apache Siege

Concurrent Users: 250

OpenBenchmarking.orgTransactions Per Second, More Is BetterApache Siege 2.4.29Concurrent Users: 250GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver220K40K60K80K100KSE +/- 3755.13, N = 15SE +/- 4063.46, N = 12SE +/- 122.71, N = 3SE +/- 1636.75, N = 1298050.9196842.1362725.24102423.07-march=znver2-march=znver21. (CC) gcc options: -O3 -lpthread -ldl -lssl -lcrypto

Apache Siege

Concurrent Users: 200

OpenBenchmarking.orgTransactions Per Second, More Is BetterApache Siege 2.4.29Concurrent Users: 200GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver220K40K60K80K100KSE +/- 798.56, N = 3SE +/- 3575.15, N = 15SE +/- 3302.37, N = 12SE +/- 1288.23, N = 1560835.7999824.4982293.1483275.06-march=znver2-march=znver21. (CC) gcc options: -O3 -lpthread -ldl -lssl -lcrypto

Memcached mcperf

Method: Set

OpenBenchmarking.orgOperations Per Second, More Is BetterMemcached mcperf 1.5.10Method: SetGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver213K26K39K52K65KSE +/- 293.82, N = 3SE +/- 3850.96, N = 15SE +/- 2058.38, N = 15SE +/- 393.33, N = 352914.1059232.0757193.2552910.87-march=znver2-march=znver21. (CC) gcc options: -O3 -lm -rdynamic

Redis

Test: GET

OpenBenchmarking.orgRequests Per Second, More Is BetterRedis 4.0.8Test: GETGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2700K1400K2100K2800K3500KSE +/- 40781.06, N = 3SE +/- 61029.58, N = 15SE +/- 47460.73, N = 15SE +/- 51486.64, N = 153297713.333066070.283042507.473031706.221. (CC) gcc options: -ggdb -rdynamic -lm -ldl -pthread

Cpuminer-Opt

Algorithm: deep

OpenBenchmarking.orgkH/s - Hash Speed, More Is BetterCpuminer-Opt 3.8.8.1Algorithm: deepGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver22K4K6K8K10KSE +/- 926.03, N = 12SE +/- 3.33, N = 3SE +/- 8.82, N = 311190.0010230.3411137.0011123.00-march=znver2-march=znver21. (CXX) g++ options: -O3 -lcurl -lz -lpthread -lssl -lcrypto -lgmp

MKL-DNN

Harness: IP Batch 1D - Data Type: f32

OpenBenchmarking.orgms, Fewer Is BetterMKL-DNN 2019-04-16Harness: IP Batch 1D - Data Type: f32GCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver2306090120150SE +/- 3.35, N = 15SE +/- 2.27, N = 15SE +/- 3.18, N = 12SE +/- 3.21, N = 14152.51155.34154.09157.72MIN: 111.42-march=znver2 - MIN: 127.99MIN: 129-march=znver2 - MIN: 1271. (CXX) g++ options: -O3 -std=c++11 -march=native -mtune=native -fPIC -fopenmp -pie -lmklml_intel -ldl

Geometric Mean Of All Test Results

Result Composite - Ryzen 9 3900X Znver2 Compiler Tuning

OpenBenchmarking.orgGeometric Mean, More Is BetterGeometric Mean Of All Test ResultsResult Composite - Ryzen 9 3900X Znver2 Compiler TuningGCC 9.1.0GCC 9.1.0 znver2GCC 10.0.0GCC 10.0.0 znver24080120160200162.44166.33162.54165.41


Phoronix Test Suite v10.8.5