Google TauVM AMD EPYC vs. M6g Graviton2 Benchmarks

Initial Google Tau VM AMD EPYC Milan benchmarking compared to Amazon EC2 M6g instances for a future article on Phoronix.com by Michael Larabel.

HTML result view exported from: https://openbenchmarking.org/result/2110179-TJ-TAUVMEC2779&grs.

Google TauVM AMD EPYC vs. M6g Graviton2 BenchmarksProcessorMotherboardMemoryDiskChipsetNetworkOSKernelCompilerFile-SystemSystem Layert2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlargeAMD EPYC 7B13 (8 Cores)Google Compute Engine t2d-standard-832GB11GB PersistentDisk + 107GB PersistentDiskDebian 104.19.0-17-cloud-amd64 (x86_64)GCC 8.3.0ext4KVMAMD EPYC 7B13 (32 Cores)Google Compute Engine t2d-standard-32124GBARMv8 Neoverse-N1 (8 Cores)Amazon EC2 m6g.2xlarge (1.0 BIOS)Amazon Device 020032GB107GB Amazon Elastic Block StoreAmazon Elastic4.19.0-14-arm64 (aarch64)ARMv8 Neoverse-N1 (32 Cores)Amazon EC2 m6g.8xlarge (1.0 BIOS)124GBOpenBenchmarking.orgKernel Details- t2d-standard-8: scsi_mod.use_blk_mq=Y - Transparent Huge Pages: always- t2d-standard-32: scsi_mod.use_blk_mq=Y - Transparent Huge Pages: always- m6g.2xlarge: Transparent Huge Pages: always- m6g.8xlarge: Transparent Huge Pages: alwaysCompiler Details- t2d-standard-8: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++ --enable-libmpx --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib --with-tune=generic --without-cuda-driver -v - t2d-standard-32: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++ --enable-libmpx --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib --with-tune=generic --without-cuda-driver -v - m6g.2xlarge: --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 - m6g.8xlarge: --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 Processor Details- t2d-standard-8, t2d-standard-32: CPU Microcode: 0x1000065Java Details- t2d-standard-8, t2d-standard-32: OpenJDK Runtime Environment (build 11.0.12+7-post-Debian-2deb10u1)Python Details- t2d-standard-8: Python 3.7.3- t2d-standard-32: Python 3.7.3- m6g.2xlarge: Python 2.7.16 + Python 3.7.3- m6g.8xlarge: Python 2.7.16 + Python 3.7.3Security Details- t2d-standard-8: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional IBRS_FW STIBP: disabled RSB filling + srbds: Not affected + tsx_async_abort: Not affected - t2d-standard-32: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional IBRS_FW STIBP: disabled RSB filling + srbds: Not affected + tsx_async_abort: Not affected - m6g.2xlarge: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected - m6g.8xlarge: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Google TauVM AMD EPYC vs. M6g Graviton2 Benchmarksjohn-the-ripper: Blowfishpgbench: 100 - 500 - Read Only - Average Latencypgbench: 100 - 500 - Read Onlypgbench: 100 - 250 - Read Onlypgbench: 100 - 250 - Read Only - Average Latencygraphics-magick: Enhancedstress-ng: Cryptocompress-7zip: Compress Speed Testgraphics-magick: Sharpenopenssl: SHA256graphics-magick: Swirlbuild-nodejs: Time To Compilestress-ng: Vector Mathdav1d: Summer Nature 4Kbasis: UASTC Level 3basis: UASTC Level 2build2: Time To Compilebuild-ffmpeg: Time To Compilebuild-gdb: Time To Compilejohn-the-ripper: MD5t2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge107963.4881437631522891.6421491650.0238830896659560627340554.72320807.29136.9674.77137.681139.92571.04471.620812530427320.8266053736179610.4055736218.86133228346257398356531248165.28378739.94326.4824.76114.96557.69830.70843.817321033351836.07382474878732.8451031257.5727799745527759170269761.41717255.1778.2199.64450.229183.07684.158100.206185258206481.4393475444066650.6154015030.58109417267220784599501006234.10269034.73198.2631.59219.17778.06132.67163.432721783OpenBenchmarking.org

John The Ripper

Test: Blowfish

OpenBenchmarking.orgReal C/S, More Is BetterJohn The Ripper 1.9.0-jumbo-1Test: Blowfisht2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge9K18K27K36K45KSE +/- 5.36, N = 3SE +/- 31.94, N = 3SE +/- 0.67, N = 3SE +/- 2.19, N = 31079642732518320648-m64-m641. (CC) gcc options: -lssl -lcrypto -fopenmp -lgmp -pthread -lm -lz -ldl -lcrypt

PostgreSQL pgbench

Scaling Factor: 100 - Clients: 500 - Mode: Read Only - Average Latency

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL pgbench 14.0Scaling Factor: 100 - Clients: 500 - Mode: Read Only - Average Latencyt2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge246810SE +/- 0.061, N = 11SE +/- 0.005, N = 3SE +/- 0.075, N = 12SE +/- 0.010, N = 33.4880.8266.0731.4391. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

PostgreSQL pgbench

Scaling Factor: 100 - Clients: 500 - Mode: Read Only

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 14.0Scaling Factor: 100 - Clients: 500 - Mode: Read Onlyt2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge130K260K390K520K650KSE +/- 2355.50, N = 11SE +/- 3614.22, N = 3SE +/- 997.61, N = 12SE +/- 2376.63, N = 3143763605373824743475441. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

PostgreSQL pgbench

Scaling Factor: 100 - Clients: 250 - Mode: Read Only

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 14.0Scaling Factor: 100 - Clients: 250 - Mode: Read Onlyt2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge130K260K390K520K650KSE +/- 453.87, N = 3SE +/- 3588.30, N = 3SE +/- 593.35, N = 3SE +/- 1401.95, N = 3152289617961878734066651. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

PostgreSQL pgbench

Scaling Factor: 100 - Clients: 250 - Mode: Read Only - Average Latency

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL pgbench 14.0Scaling Factor: 100 - Clients: 250 - Mode: Read Only - Average Latencyt2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge0.64011.28021.92032.56043.2005SE +/- 0.005, N = 3SE +/- 0.002, N = 3SE +/- 0.019, N = 3SE +/- 0.002, N = 31.6420.4052.8450.6151. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

GraphicsMagick

Operation: Enhanced

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.33Operation: Enhancedt2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge120240360480600SE +/- 0.33, N = 3149573103401-lSM -lICE -lX11 -lxml2-lSM -lICE -lX11 -lxml2-lxml21. (CC) gcc options: -fopenmp -O2 -pthread -ljpeg -lz -lm -lpthread

Stress-NG

Test: Crypto

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.13.02Test: Cryptot2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge13002600390052006500SE +/- 5.68, N = 3SE +/- 2.08, N = 3SE +/- 0.71, N = 3SE +/- 2.69, N = 31650.026218.861257.575030.581. (CC) gcc options: -O2 -std=gnu99 -lm -lcrypt -lrt -lz -ldl -pthread -lc -latomic

7-Zip Compression

Compress Speed Test

OpenBenchmarking.orgMIPS, More Is Better7-Zip Compression 16.02Compress Speed Testt2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge30K60K90K120K150KSE +/- 581.53, N = 15SE +/- 1357.41, N = 3SE +/- 77.93, N = 3SE +/- 186.53, N = 338830133228277991094171. (CXX) g++ options: -pipe -lpthread

GraphicsMagick

Operation: Sharpen

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.33Operation: Sharpent2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge801602403204008934674267-lSM -lICE -lX11 -lxml2-lSM -lICE -lX11 -lxml2-lxml21. (CC) gcc options: -fopenmp -O2 -pthread -ljpeg -lz -lm -lpthread

OpenSSL

Algorithm: SHA256

OpenBenchmarking.orgbyte/s, More Is BetterOpenSSL 3.0Algorithm: SHA256t2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge6000M12000M18000M24000M30000MSE +/- 51724089.12, N = 3SE +/- 4017105.90, N = 3SE +/- 13414421.31, N = 3SE +/- 55812357.58, N = 3665956062725739835653552775917022078459950-m64-m641. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl

GraphicsMagick

Operation: Swirl

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.33Operation: Swirlt2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge30060090012001500SE +/- 1.53, N = 3SE +/- 2.67, N = 3SE +/- 2.33, N = 334012482691006-lSM -lICE -lX11 -lxml2-lSM -lICE -lX11 -lxml2-lxml21. (CC) gcc options: -fopenmp -O2 -pthread -ljpeg -lz -lm -lpthread

Timed Node.js Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Node.js Compilation 15.11Time To Compilet2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge160320480640800SE +/- 1.86, N = 3SE +/- 0.89, N = 3SE +/- 3.14, N = 3SE +/- 0.29, N = 3554.72165.28761.42234.10

Stress-NG

Test: Vector Math

OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.13.02Test: Vector Matht2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge20K40K60K80K100KSE +/- 58.31, N = 3SE +/- 12.58, N = 3SE +/- 7.25, N = 3SE +/- 28.36, N = 320807.2978739.9417255.1769034.731. (CC) gcc options: -O2 -std=gnu99 -lm -lcrypt -lrt -lz -ldl -pthread -lc -latomic

dav1d

Video Input: Summer Nature 4K

OpenBenchmarking.orgFPS, More Is Betterdav1d 0.9.2Video Input: Summer Nature 4Kt2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge70140210280350SE +/- 0.15, N = 3SE +/- 1.06, N = 3SE +/- 0.07, N = 3SE +/- 0.03, N = 3136.96326.4878.21198.26MIN: 128.38 / MAX: 155.91MIN: 254.49 / MAX: 365.12MIN: 72.84 / MAX: 84.32MIN: 135.94 / MAX: 210.431. (CC) gcc options: -lm -pthread

Basis Universal

Settings: UASTC Level 3

OpenBenchmarking.orgSeconds, Fewer Is BetterBasis Universal 1.13Settings: UASTC Level 3t2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge20406080100SE +/- 0.24, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 374.7724.7699.6431.591. (CXX) g++ options: -std=c++11 -fvisibility=hidden -fPIC -fno-strict-aliasing -O3 -rdynamic -lm -lpthread

Basis Universal

Settings: UASTC Level 2

OpenBenchmarking.orgSeconds, Fewer Is BetterBasis Universal 1.13Settings: UASTC Level 2t2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge1122334455SE +/- 0.06, N = 3SE +/- 0.03, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 337.6814.9750.2319.181. (CXX) g++ options: -std=c++11 -fvisibility=hidden -fPIC -fno-strict-aliasing -O3 -rdynamic -lm -lpthread

Build2

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterBuild2 0.13Time To Compilet2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge4080120160200SE +/- 0.54, N = 3SE +/- 0.06, N = 3SE +/- 1.20, N = 3SE +/- 0.06, N = 3139.9357.70183.0878.06

Timed FFmpeg Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed FFmpeg Compilation 4.4Time To Compilet2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge20406080100SE +/- 0.62, N = 15SE +/- 0.24, N = 3SE +/- 0.17, N = 3SE +/- 0.03, N = 371.0430.7184.1632.67

Timed GDB GNU Debugger Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed GDB GNU Debugger Compilation 10.2Time To Compilet2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge20406080100SE +/- 0.43, N = 3SE +/- 0.09, N = 3SE +/- 0.13, N = 3SE +/- 0.04, N = 371.6243.82100.2163.43

John The Ripper

Test: MD5

OpenBenchmarking.orgReal C/S, More Is BetterJohn The Ripper 1.9.0-jumbo-1Test: MD5t2d-standard-8t2d-standard-32m6g.2xlargem6g.8xlarge700K1400K2100K2800K3500KSE +/- 1883.13, N = 3SE +/- 1201.85, N = 3SE +/- 37.20, N = 3SE +/- 670.59, N = 38125303210333185258721783-m64-m641. (CC) gcc options: -lssl -lcrypto -fopenmp -lgmp -pthread -lm -lz -ldl -lcrypt

Phoronix Test Suite v10.8.5