Cavium ThunderX 96-Core vs. Ampere eMAG

ARM64 Workstation Comparison Benchmark

HTML result view exported from: https://openbenchmarking.org/result/1905310-JONA-181009031&sro.

Cavium ThunderX 96-Core vs. Ampere eMAGProcessorMotherboardMemoryDiskGraphicsNetworkChipsetOSKernelDisplay DriverCompilerFile-SystemScreen ResolutionCavium ThunderX 96-CoreAmpere eMAG OSPREYClearFog-ITXCavium ThunderX (96 Cores)FOXCONN C2U4N_MB (G31FB18A BIOS)4 x 32 GB DDR4-2133MHz 36ASF4G72PZ-2G3B1250GB Samsung SSD 850ASPEED ASPEED FamilyCavium THUNDERX Interface + Cavium THUNDERX BGXUbuntu 16.044.10.0-38-generic (aarch64)modesetting 1.18.4GCC 5.4.0 20160609ext4800x600Ampere eMAG ARMv8 @ 3.00GHz (32 Cores)AmpereComputing OSPREY (4.8.19 BIOS)Applied Micro Circuits X-Gene8 x 16384 MB DDR4-2400MHz Samsung M393A2K43CB2-CTD256GB Samsung SSD 860Intel I210 Gigabit ConnectionCentOS 7.5.18044.14.0-49.el7.centos.180925+amp.ilp32.aarch64 (aarch64) 20180926GCC 6.0.0 20150826xfs1024x768ARMv8 Cortex-A72 @ 2.00GHz (16 Cores)SolidRun LX2160A COM type 7 module14336MB250GB Samsung SSD 860 + 8GB SL08G + 63GB DF4064Ubuntu 19.044.14.83-dirty (aarch64)GCC 8.3.0ext4OpenBenchmarking.orgCompiler Details- Cavium ThunderX 96-Core: --build=aarch64-linux-gnu --disable-browser-plugin --disable-libquadmath --disable-werror --enable-checking=release --enable-clocale=gnu --enable-fix-cortex-a53-843419 --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-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-nls --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --target=aarch64-linux-gnu --with-arch-directory=aarch64 --with-default-libstdcxx-abi=new -v - Ampere eMAG OSPREY: --build=x86_64-unknown-linux-gnu --disable-libmudflap --disable-libquadmath --disable-libsanitizer --disable-libssp --enable-languages=c,c++,fortran,objc --enable-multilib --enable-shared --host=aarch64-apm-linux-gnu --target=aarch64-apm-linux-gnu --with-abi=lp64 --with-build-sysroot=/home/bhuber/sandbox/gcc/toolchain-build-scripts/buildroot/9.0.7-le/install/opt/apm-aarch64/9.0.7-le/bin/../aarch64-apm-linux-gnu/libc --with-cpu=xgene1 --with-multilib-list=lp64,ilp32 --with-plugin-ld=ld - ClearFog-ITX: --build=aarch64-linux-gnu --disable-libphobos --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --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-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-default-libstdcxx-abi=new --with-gcc-major-version-only -v Python Details- Cavium ThunderX 96-Core: Python 2.7.12 + Python 3.5.2- Ampere eMAG OSPREY: Python 2.7.5- ClearFog-ITX: Python 2.7.16 + Python 3.7.3Disk Details- Ampere eMAG OSPREY: DEADLINE / attr2,inode64,noquota,relatime,rwProcessor Details- Ampere eMAG OSPREY: Scaling Governor: cppc_cpufreq performance- ClearFog-ITX: Scaling Governor: qoriq_cpufreq performance

Cavium ThunderX 96-Core vs. Ampere eMAGrodinia: OpenMP LavaMDmafft: Multiple Sequence Alignmentbuild-linux-kernel: Time To Compilec-ray: Total Timeencode-mp3: WAV To MP3pybench: Total For Average Test Timesaobench: 2048 x 2048 - Total Timepgbench: Buffer Test - Normal Load - Read Onlypgbench: Buffer Test - Normal Load - Read Writepgbench: Buffer Test - Single Thread - Read Onlypgbench: Buffer Test - Single Thread - Read WriteCavium ThunderX 96-CoreAmpere eMAG OSPREYClearFog-ITX66.5517.652198.37212.6011425235.3117707348253028245121.773.702768.97110.354775151.241633944680586643151.225.4129412.8290.68408292.89OpenBenchmarking.org

Rodinia

Test: OpenMP LavaMD

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 2.4Test: OpenMP LavaMDAmpere eMAG OSPREYCavium ThunderX 96-CoreClearFog-ITX306090120150SE +/- 2.16, N = 9SE +/- 0.71, N = 3SE +/- 0.03, N = 3121.7766.5551.22-O3 -fopenmp-O2 -lOpenCL-O2 -lOpenCL1. (CXX) g++ options:

Timed MAFFT Alignment

Multiple Sequence Alignment

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed MAFFT Alignment 6.864Multiple Sequence AlignmentAmpere eMAG OSPREYCavium ThunderX 96-CoreClearFog-ITX48121620SE +/- 0.06, N = 3SE +/- 0.76, N = 6SE +/- 0.20, N = 33.7017.655.411. (CC) gcc options: -O3 -lm -lpthread

Timed Linux Kernel Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Linux Kernel Compilation 4.13Time To CompileAmpere eMAG OSPREYCavium ThunderX 96-CoreClearFog-ITX60120180240300SE +/- 2.72, N = 3SE +/- 3.36, N = 3SE +/- 1.94, N = 3276219294

C-Ray

Total Time

OpenBenchmarking.orgSeconds, Fewer Is BetterC-Ray 1.1Total TimeAmpere eMAG OSPREYCavium ThunderX 96-CoreClearFog-ITX3691215SE +/- 0.02, N = 3SE +/- 0.08, N = 3SE +/- 0.25, N = 38.978.3712.821. (CC) gcc options: -lm -lpthread -O3

LAME MP3 Encoding

WAV To MP3

OpenBenchmarking.orgSeconds, Fewer Is BetterLAME MP3 Encoding 3.100WAV To MP3Ampere eMAG OSPREYCavium ThunderX 96-CoreClearFog-ITX50100150200250SE +/- 0.03, N = 3SE +/- 0.19, N = 3SE +/- 0.14, N = 3110.35212.6090.681. (CC) gcc options: -lm

PyBench

Total For Average Test Times

OpenBenchmarking.orgMilliseconds, Fewer Is BetterPyBench 2008-08-14Total For Average Test TimesAmpere eMAG OSPREYCavium ThunderX 96-CoreClearFog-ITX2K4K6K8K10KSE +/- 6.81, N = 3SE +/- 21.88, N = 3SE +/- 12.78, N = 34775114254082

AOBench

Size: 2048 x 2048 - Total Time

OpenBenchmarking.orgSeconds, Fewer Is BetterAOBenchSize: 2048 x 2048 - Total TimeAmpere eMAG OSPREYCavium ThunderX 96-CoreClearFog-ITX50100150200250SE +/- 0.01, N = 3SE +/- 0.03, N = 3SE +/- 0.20, N = 3151.24235.3192.891. (CC) gcc options: -lm -O3

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 OnlyAmpere eMAG OSPREYCavium ThunderX 96-Core40K80K120K160K200KSE +/- 147.63, N = 3SE +/- 1351.32, N = 31633941770731. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -fPIC -lpgcommon -lpgport -lpthread -lrt -lcrypt -ldl -lm

PostgreSQL pgbench

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

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 10.0Scaling: Buffer Test - Test: Normal Load - Mode: Read WriteAmpere eMAG OSPREYCavium ThunderX 96-Core10002000300040005000SE +/- 2.90, N = 3SE +/- 13.13, N = 3468048251. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -fPIC -lpgcommon -lpgport -lpthread -lrt -lcrypt -ldl -lm

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 OnlyAmpere eMAG OSPREYCavium ThunderX 96-Core13002600390052006500SE +/- 79.07, N = 3SE +/- 40.28, N = 3586630281. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -fPIC -lpgcommon -lpgport -lpthread -lrt -lcrypt -ldl -lm

PostgreSQL pgbench

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

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 10.0Scaling: Buffer Test - Test: Single Thread - Mode: Read WriteAmpere eMAG OSPREYCavium ThunderX 96-Core90180270360450SE +/- 0.24, N = 3SE +/- 1.10, N = 34312451. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -fPIC -lpgcommon -lpgport -lpthread -lrt -lcrypt -ldl -lm

Phoronix Test Suite v10.8.5