GCC 8.0 vs. Clang 6.0 AMD EPYC Tuning Comparison

Tests for a future article on Phoronix.

HTML result view exported from: https://openbenchmarking.org/result/1801022-AL-CLANGCC8680&rdt&grw.

ProcessorMotherboardChipsetMemoryDiskGraphicsMonitorNetworkOSKernelDesktopDisplay DriverOpenCLCompilerFile-SystemScreen ResolutionSystem LayerClang 6.0GCC 8.0 -march=znver1 -march=x86-64 -march=znver1 -march=x86-64AMD EPYC 7601 32-Core @ 2.20GHz (64 Cores)TYAN B8026T70AE24HRAMD Device 1450126976MB280GB INTEL SSDPE21D280GAASPEED ASPEED FamilyVE228Broadcom Limited NetXtreme BCM5720 Gigabit PCIeUbuntu 17.104.13.0-21-generic (x86_64)GNOME Shell 3.26.1modesetting 1.19.5OpenCL 1.2 pocl 1.0 LLVM 5.0.0Clang 6.0.0 (SVN 321623) + LLVM 6.0.0svnext41920x1080vm-other Xen 4.9.0 HypervisorGCC 8.0.0 20171231 + clang (GCC) 8.0.0 20171231 (experimental) + LLVM 5.0.0OpenBenchmarking.orgCompiler Details- Clang 6.0: -march=znver1: Optimized build; Default target: x86_64-unknown-linux-gnu; Host CPU: znver1- Clang 6.0: -march=x86-64: Optimized build; Default target: x86_64-unknown-linux-gnu; Host CPU: znver1- GCC 8.0: -march=znver1: --disable-multilib --enable-checking=release- GCC 8.0: -march=x86-64: --disable-multilib --enable-checking=releaseDisk Details- NONE / data=ordered,errors=remount-ro,relatime,rw

bullet: Raytestsbullet: 3000 Fallbullet: 1000 Stackbullet: 1000 Convexbullet: 136 Ragdollsbullet: Prim Trimeshbullet: Convex Trimeshtscp: AI Chess Performancescimark2: Compositescimark2: Monte Carloscimark2: Fast Fourier Transformscimark2: Sparse Matrix Multiplyscimark2: Dense LU Matrix Factorizationscimark2: Jacobi Successive Over-Relaxationencode-flac: WAV To FLACencode-mp3: WAV To MP3fftw: Stock - 2D FFT Size 4096fftw: Float + SSE - 2D FFT Size 4096hmmer: Pfam Database Searchhimeno: Poisson Pressure Solvergraphics-magick: Blurgraphics-magick: Sharpengraphics-magick: HWB Color Spacegraphics-magick: Local Adaptive Thresholdingc-ray: Total Timeapache: Static Web Page Servingsqlite: Default Test Directoryebizzy: polybench-c: 3 Matrix MultiplicationsClang 6.0GCC 8.0 -march=znver1 -march=x86-64 -march=znver1 -march=x86-643.185.346.085.313.231.091.329182691699.32552.19226.682258.644034.891424.216.6312.815031.601248111.091052.47104136155984.489663.937.48114540562.753.225.486.305.433.281.101.339176581479.53531.38179.292190.103190.431110.657.9411.334660.831364912.851032.71101131150974.539531.437.53107664862.983.065.275.935.283.191.101.308750851680.45555.76231.092259.953678.861676.627.4510.815627.831363012.40935.64123165186953.379791.237.16110117665.453.125.346.185.443.261.101.348742511579.48561.03233.892263.873513.111423.147.1211.104959.7313.65949.19116157177923.939841.307.61112603260.68OpenBenchmarking.org

Bullet Physics Engine

Test: Raytests

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: Raytests-march=znver1-march=x86-640.72451.4492.17352.8983.6225SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.01, N = 3SE +/- 0.00, N = 33.183.223.063.121. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

Bullet Physics Engine

Test: 3000 Fall

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: 3000 Fall-march=znver1-march=x86-641.2332.4663.6994.9326.165SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 35.345.485.275.341. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

Bullet Physics Engine

Test: 1000 Stack

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: 1000 Stack-march=znver1-march=x86-64246810SE +/- 0.01, N = 3SE +/- 0.03, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 36.086.305.936.181. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

Bullet Physics Engine

Test: 1000 Convex

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: 1000 Convex-march=znver1-march=x86-641.2242.4483.6724.8966.12SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.00, N = 35.315.435.285.441. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

Bullet Physics Engine

Test: 136 Ragdolls

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: 136 Ragdolls-march=znver1-march=x86-640.7381.4762.2142.9523.69SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.01, N = 3SE +/- 0.00, N = 33.233.283.193.261. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

Bullet Physics Engine

Test: Prim Trimesh

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: Prim Trimesh-march=znver1-march=x86-640.24750.4950.74250.991.2375SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 31.091.101.101.101. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

Bullet Physics Engine

Test: Convex Trimesh

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterBullet Physics Engine 2.81Test: Convex Trimesh-march=znver1-march=x86-640.30150.6030.90451.2061.5075SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 31.321.331.301.341. (CXX) g++ options: -O3 -rdynamic -lglut -lGL -lGLU

TSCP

AI Chess Performance

Clang 6.0GCC 8.0OpenBenchmarking.orgNodes Per Second, More Is BetterTSCP 1.81AI Chess Performance-march=znver1-march=x86-64200K400K600K800K1000KSE +/- 306.40, N = 5SE +/- 572.37, N = 5SE +/- 556.88, N = 5SE +/- 438.77, N = 59182699176588750858742511. (CC) gcc options: -O3 -march=native

SciMark

Computational Test: Composite

Clang 6.0GCC 8.0OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Composite-march=znver1-march=x86-64400800120016002000SE +/- 25.95, N = 4SE +/- 36.18, N = 8SE +/- 20.36, N = 4SE +/- 20.51, N = 71699.321479.531680.451579.481. (CC) gcc options: -O3 -lm

SciMark

Computational Test: Monte Carlo

Clang 6.0GCC 8.0OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Monte Carlo-march=znver1-march=x86-64120240360480600SE +/- 0.05, N = 4SE +/- 0.05, N = 4SE +/- 0.01, N = 4SE +/- 0.04, N = 4552.19531.38555.76561.031. (CC) gcc options: -O3 -lm

SciMark

Computational Test: Fast Fourier Transform

Clang 6.0GCC 8.0OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Fast Fourier Transform-march=znver1-march=x86-6450100150200250SE +/- 0.38, N = 4SE +/- 45.12, N = 4SE +/- 0.14, N = 4SE +/- 0.35, N = 4226.68179.29231.09233.891. (CC) gcc options: -O3 -lm

SciMark

Computational Test: Sparse Matrix Multiply

Clang 6.0GCC 8.0OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Sparse Matrix Multiply-march=znver1-march=x86-645001000150020002500SE +/- 29.07, N = 4SE +/- 11.89, N = 4SE +/- 9.55, N = 4SE +/- 6.72, N = 42258.642190.102259.952263.871. (CC) gcc options: -O3 -lm

SciMark

Computational Test: Dense LU Matrix Factorization

Clang 6.0GCC 8.0OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Dense LU Matrix Factorization-march=znver1-march=x86-649001800270036004500SE +/- 135.63, N = 4SE +/- 17.92, N = 4SE +/- 110.38, N = 4SE +/- 177.13, N = 44034.893190.433678.863513.111. (CC) gcc options: -O3 -lm

SciMark

Computational Test: Jacobi Successive Over-Relaxation

Clang 6.0GCC 8.0OpenBenchmarking.orgMflops, More Is BetterSciMark 2.0Computational Test: Jacobi Successive Over-Relaxation-march=znver1-march=x86-64400800120016002000SE +/- 0.49, N = 4SE +/- 310.16, N = 4SE +/- 0.77, N = 4SE +/- 0.28, N = 41424.211110.651676.621423.141. (CC) gcc options: -O3 -lm

FLAC Audio Encoding

WAV To FLAC

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterFLAC Audio Encoding 1.3.1WAV To FLAC-march=znver1-march=x86-64246810SE +/- 0.00, N = 5SE +/- 0.01, N = 5SE +/- 0.01, N = 5SE +/- 0.01, N = 56.637.947.457.121. (CXX) g++ options: -O3 -fvisibility=hidden -logg -lm

LAME MP3 Encoding

WAV To MP3

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterLAME MP3 Encoding 3.99.5WAV To MP3-march=znver1-march=x86-643691215SE +/- 0.01, N = 5SE +/- 0.00, N = 5SE +/- 0.01, N = 5SE +/- 0.02, N = 512.8111.3310.8111.101. (CC) gcc options: -O3 -ffast-math -funroll-loops -fschedule-insns2 -fbranch-count-reg -fforce-addr -pipe -lm

FFTW

Build: Stock - Size: 2D FFT Size 4096

Clang 6.0GCC 8.0OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Stock - Size: 2D FFT Size 4096-march=znver1-march=x86-6412002400360048006000SE +/- 17.93, N = 3SE +/- 57.25, N = 3SE +/- 10.37, N = 3SE +/- 1.16, N = 35031.604660.835627.834959.731. (CC) gcc options: -pthread -O3 -lm

FFTW

Build: Float + SSE - Size: 2D FFT Size 4096

Clang 6.0GCC 8.0OpenBenchmarking.orgMflops, More Is BetterFFTW 3.3.6Build: Float + SSE - Size: 2D FFT Size 4096-march=znver1-march=x86-643K6K9K12K15KSE +/- 76.70, N = 3SE +/- 100.95, N = 3SE +/- 15.38, N = 31248113649136301. (CC) gcc options: -pthread -O3 -lm

Timed HMMer Search

Pfam Database Search

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterTimed HMMer Search 2.3.2Pfam Database Search-march=znver1-march=x86-6448121620SE +/- 0.12, N = 3SE +/- 1.28, N = 6SE +/- 0.04, N = 3SE +/- 1.74, N = 611.0912.8512.4013.651. (CC) gcc options: -O3 -pthread -lhmmer -lsquid -lm

Himeno Benchmark

Poisson Pressure Solver

Clang 6.0GCC 8.0OpenBenchmarking.orgMFLOPS, More Is BetterHimeno Benchmark 3.0Poisson Pressure Solver-march=znver1-march=x86-642004006008001000SE +/- 1.50, N = 3SE +/- 1.46, N = 3SE +/- 0.79, N = 3SE +/- 0.43, N = 31052.471032.71935.64949.191. (CC) gcc options: -O3 -mavx2

GraphicsMagick

Operation: Blur

Clang 6.0GCC 8.0OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.19Operation: Blur-march=znver1-march=x86-64306090120150SE +/- 0.33, N = 3SE +/- 0.33, N = 31041011231161. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

GraphicsMagick

Operation: Sharpen

Clang 6.0GCC 8.0OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.19Operation: Sharpen-march=znver1-march=x86-644080120160200SE +/- 0.67, N = 3SE +/- 0.33, N = 31361311651571. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

GraphicsMagick

Operation: HWB Color Space

Clang 6.0GCC 8.0OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.19Operation: HWB Color Space-march=znver1-march=x86-644080120160200SE +/- 0.88, N = 3SE +/- 1.20, N = 31551501861771. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

GraphicsMagick

Operation: Local Adaptive Thresholding

Clang 6.0GCC 8.0OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.19Operation: Local Adaptive Thresholding-march=znver1-march=x86-6420406080100SE +/- 0.33, N = 3SE +/- 0.33, N = 3989795921. (CC) gcc options: -fopenmp -O3 -pthread -ljbig -lwebp -ltiff -ljpeg -lXext -lSM -lICE -lX11 -llzma -lbz2 -lz -lm -ldl -lpthread

C-Ray

Total Time

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterC-Ray 1.1Total Time-march=znver1-march=x86-641.01932.03863.05794.07725.0965SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.02, N = 3SE +/- 0.00, N = 34.484.533.373.931. (CC) gcc options: -lm -lpthread -O3

Apache Benchmark

Static Web Page Serving

Clang 6.0GCC 8.0OpenBenchmarking.orgRequests Per Second, More Is BetterApache Benchmark 2.4.7Static Web Page Serving-march=znver1-march=x86-642K4K6K8K10KSE +/- 121.20, N = 3SE +/- 161.49, N = 3SE +/- 35.54, N = 3SE +/- 22.88, N = 39663.939531.439791.239841.301. (CC) gcc options: -shared -fPIC -pthread -O3

SQLite

Test Target: Default Test Directory

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterSQLite 3.8.10.2Test Target: Default Test Directory-march=znver1-march=x86-64246810SE +/- 0.15, N = 6SE +/- 0.06, N = 3SE +/- 0.12, N = 6SE +/- 0.12, N = 37.487.537.167.611. (CC) gcc options: -O3 -ldl -lpthread

ebizzy

Clang 6.0GCC 8.0OpenBenchmarking.orgRecords/s, More Is Betterebizzy 0.3-march=znver1-march=x86-64200K400K600K800K1000KSE +/- 17141.58, N = 6SE +/- 12350.97, N = 3SE +/- 19461.21, N = 3SE +/- 20747.16, N = 611454051076648110117611260321. (CC) gcc options: -pthread -lpthread -O3 -march=native

PolyBench-C

Test: 3 Matrix Multiplications

Clang 6.0GCC 8.0OpenBenchmarking.orgSeconds, Fewer Is BetterPolyBench-C 3.2Test: 3 Matrix Multiplications-march=znver1-march=x86-641530456075SE +/- 0.26, N = 3SE +/- 0.24, N = 3SE +/- 0.21, N = 3SE +/- 0.03, N = 362.7562.9865.4560.681. (CC) gcc options: -O3 -march=native


Phoronix Test Suite v10.8.4