OpenCL ROCm vs. AMDGPU-PRO 17.40 + 2MB FS

AMDGPU-PRO 17.40 vs. ROCm. Linux OpenCL compute performance. Tests for a future article.

HTML result view exported from: https://openbenchmarking.org/result/1710282-AL-OPENCLNEW89&grs&sro.

ProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay DriverOpenGLOpenCLCompilerFile-SystemScreen ResolutionAMDGPU-PRO 17.40ROCm RX 560 RX 580 R9 Fury RX Vega 56 RX Vega 64 RX 560 RX 580 R9 Fury RX Vega 56 RX Vega 64Intel Core i7-7700K @ 4.50GHz (4 Cores / 8 Threads)MSI Z270-A PRO (MS-7A71) v1.0Intel Device 591f16384MB128GB THNSN5128GPU7 TOSHIBAAMD Radeon RX 560 4096MBRealtek ALC892DELL P2415QRealtek RTL8111/8168/8411Ubuntu 16.044.10.0-37-generic (x86_64)Unity 7.4.0amdgpu 1.3.994.5.13496OpenCL 2.0 AMD-APP (2482.3)GCC 5.4.0 20160609ext43840x2160MSI AMD Radeon RX 580 8192MBmodesetting 1.19.3Sapphire AMD Radeon 4096MBAMD Radeon RX Vega 8176MBamdgpu 1.3.99AMD POLARIS11 4096MB4.11.0-kfd-compute-rocm-rel-1.6-180 (x86_64)amdgpu 1.3.04.5 Mesa 17.0.7 Gallium 0.4 (LLVM 4.0.0)OpenCL 2.0 AMD-APP (2508.0)MSI AMD POLARIS10 8192MBmodesetting 1.19.3Sapphire AMD Radeon R9 FURY / NANO 4096MBLLVMpipe3.3 Mesa 17.0.7 Gallium 0.4 (LLVM 4.0 256 bits)OpenBenchmarking.orgCompiler Details- --build=x86_64-linux-gnu --disable-browser-plugin --disable-vtable-verify --disable-werror --enable-checking=release --enable-clocale=gnu --enable-gnu-unique-object --enable-gtk-cairo --enable-java-awt=gtk --enable-java-home --enable-languages=c,ada,c++,java,go,d,fortran,objc,obj-c++ --enable-libmpx --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-arch-directory=amd64 --with-default-libstdcxx-abi=new --with-multilib-list=m32,m64,mx32 --with-tune=generic -v Processor Details- Scaling Governor: intel_pstate performanceGraphics Details- AMDGPU-PRO 17.40: RX 560, AMDGPU-PRO 17.40: RX 580, AMDGPU-PRO 17.40: R9 Fury, AMDGPU-PRO 17.40: RX Vega 56, AMDGPU-PRO 17.40: RX Vega 64, ROCm: RX 560, ROCm: RX 580, ROCm: R9 Fury: GLAMOREnvironment Details- ROCm: RX Vega 56, ROCm: RX Vega 64: LIBGL_ALWAYS_SOFTWARE=1

luxmark: GPU - Hotelcl-mem: Readcl-mem: Writecl-mem: Copyethminer: GPU OpenCLluxmark: GPU - Luxball HDRAMDGPU-PRO 17.40ROCm RX 560 RX 580 R9 Fury RX Vega 56 RX Vega 64 RX 560 RX 580 R9 Fury RX Vega 56 RX Vega 64109090.2080.3077.071091247263142529208.33177.97173.2021963298146222468447.90374.07330.1025445444207643708337.00321.60306.6329370321215434244397.50378.23363.77323092472217858493.5080.8380.97849783363091384166.43180.33184.2716895180144811561123.50390.97206.3024804647196471561178.47317.50204.1729552366230581835189.13389.13222.203299082223626OpenBenchmarking.org

LuxMark

OpenCL Device: GPU - Scene: Hotel

AMDGPU-PRO 17.40ROCmOpenBenchmarking.orgScore, More Is BetterLuxMark 3.0OpenCL Device: GPU - Scene: HotelR9 FuryRX 560RX 580RX Vega 56RX Vega 649001800270036004500SE +/- 8.17, N = 3SE +/- 2.33, N = 3SE +/- 4.81, N = 3SE +/- 20.37, N = 3SE +/- 15.59, N = 3SE +/- 3.00, N = 3SE +/- 1.20, N = 3SE +/- 2.85, N = 3SE +/- 3.18, N = 3SE +/- 7.22, N = 3246810902529370842441561584138415611835

cl-mem

Benchmark: Read

AMDGPU-PRO 17.40ROCmOpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: ReadR9 FuryRX 560RX 580RX Vega 56RX Vega 64100200300400500SE +/- 0.12, N = 3SE +/- 0.66, N = 3SE +/- 0.20, N = 3SE +/- 1.71, N = 3SE +/- 0.00, N = 3SE +/- 0.06, N = 3SE +/- 0.00, N = 3SE +/- 1.95, N = 3SE +/- 0.15, N = 3SE +/- 0.19, N = 3447.9090.20208.33337.00397.50123.5093.50166.43178.47189.131. (CC) gcc options: -O2 -flto -lOpenCL

cl-mem

Benchmark: Write

AMDGPU-PRO 17.40ROCmOpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: WriteR9 FuryRX 560RX 580RX Vega 56RX Vega 6480160240320400SE +/- 2.49, N = 3SE +/- 0.06, N = 3SE +/- 0.33, N = 3SE +/- 0.00, N = 3SE +/- 0.03, N = 3SE +/- 3.08, N = 3SE +/- 0.03, N = 3SE +/- 0.26, N = 3SE +/- 3.10, N = 3SE +/- 0.09, N = 3374.0780.30177.97321.60378.23390.9780.83180.33317.50389.131. (CC) gcc options: -O2 -flto -lOpenCL

cl-mem

Benchmark: Copy

AMDGPU-PRO 17.40ROCmOpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: CopyR9 FuryRX 560RX 580RX Vega 56RX Vega 6480160240320400SE +/- 1.55, N = 3SE +/- 0.52, N = 3SE +/- 0.75, N = 3SE +/- 1.56, N = 3SE +/- 0.03, N = 3SE +/- 0.15, N = 3SE +/- 0.09, N = 3SE +/- 0.03, N = 3SE +/- 0.07, N = 3SE +/- 0.10, N = 3330.1077.07173.20306.63363.77206.3080.97184.27204.17222.201. (CC) gcc options: -O2 -flto -lOpenCL

Ethereum Ethminer

Device: GPU OpenCL

AMDGPU-PRO 17.40ROCmOpenBenchmarking.orgH/s, More Is BetterEthereum Ethminer 1.2.9Device: GPU OpenCLR9 FuryRX 560RX 580RX Vega 56RX Vega 647M14M21M28M35MSE +/- 100519.99, N = 3SE +/- 39105.19, N = 3SE +/- 51510.63, N = 3SE +/- 131902.80, N = 3SE +/- 446721.48, N = 3SE +/- 121114.30, N = 3SE +/- 39402.49, N = 3SE +/- 65584.53, N = 3SE +/- 126928.78, N = 3SE +/- 578374.82, N = 32544544410912472219632982937032132309247248046478497833168951802955236632990822

LuxMark

OpenCL Device: GPU - Scene: Luxball HDR

AMDGPU-PRO 17.40ROCmOpenBenchmarking.orgScore, More Is BetterLuxMark 3.0OpenCL Device: GPU - Scene: Luxball HDRR9 FuryRX 560RX 580RX Vega 56RX Vega 645K10K15K20K25KSE +/- 112.25, N = 3SE +/- 8.76, N = 3SE +/- 62.00, N = 3SE +/- 3.84, N = 3SE +/- 194.29, N = 3SE +/- 99.50, N = 3SE +/- 12.14, N = 3SE +/- 17.85, N = 3SE +/- 15.19, N = 3SE +/- 157.33, N = 3207646314146222154322178196476309144812305823626

Phoronix Test Suite v10.8.4