Ampere Altra Mt Jade vs. Xeon vs. EPYC Benchmarks Benchmarks for a future article. Compared to prior out-of-the-box Ubuntu results, this round of testing is with the performance governor forced on all CPUs. Also with the CFLAGS/CXXFLAGS being catered on each server to the processor microarchitecture under test. Ampere Altra "Boost" run is when force enabling CPUFreq boost on Ubuntu.
HTML result view exported from: https://openbenchmarking.org/result/2012106-HA-AABOOST0459 .
Ampere Altra Mt Jade vs. Xeon vs. EPYC Benchmarks Processor Motherboard Chipset Memory Disk Graphics Monitor Network OS Kernel Display Server Display Driver Compiler File-System Screen Resolution Desktop Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune Ampere Altra ARMv8 Neoverse-N1 @ 3.00GHz (160 Cores) WIWYNN Mt.Jade (1.1.20201019 BIOS) Ampere Computing LLC Device e100 16 x 32 GB DDR4-3200MT/s Samsung M393A4K40DB3-CWE 3841GB Micron_9300_MTFDHAL3T8TDP ASPEED VE228 Mellanox MT28908 + Intel I210 Ubuntu 20.10 5.8.0-29-generic (aarch64) X Server 1.20.9 modesetting 1.20.9 GCC 10.2.0 ext4 1920x1080 2 x Intel Xeon Platinum 8280 @ 4.00GHz (56 Cores / 112 Threads) GIGABYTE MD61-SC2-00 v01000100 (T15 BIOS) Intel Sky Lake-E DMI3 Registers 12 x 32 GB DDR4-2933MT/s HMA84GR7CJR4N-WM 2 x Intel X722 for 1GbE + 2 x QLogic FastLinQ QL41000 10/25/40/50GbE 5.8.0-29-generic (x86_64) GNOME Shell 3.38.1 X Server 1.20.8 modesetting 1.20.8 1024x768 2 x AMD EPYC 7742 64-Core @ 2.25GHz (128 Cores / 256 Threads) AMD DAYTONA_X (RDY1006G BIOS) AMD Starship/Matisse 16 x 32 GB DDR4-3200MT/s 36ASF4G72PZ-3G2E2 VE228 2 x Mellanox MT27710 1920x1080 Ampere Altra ARMv8 Neoverse-N1 @ 3.30GHz (160 Cores) WIWYNN Mt.Jade (1.1.20201019 BIOS) Ampere Computing LLC Device e100 502GB 3841GB Micron_9300_MTFDHAL3T8TDP + 960GB SAMSUNG MZ1LB960HAJQ-00007 Mellanox MT28908 + Intel I210 5.10.0-051000rc6daily20201206-generic (aarch64) 20201206 X Server 1.20.9 modesetting 1.20.9 OpenBenchmarking.org Environment Details - Ampere Altra 2P: CXXFLAGS="-O3 -march=armv8.2-a -mtune=neoverse-n1" CFLAGS="-O3 -march=armv8.2-a -mtune=neoverse-n1" - Xeon Platinum 8280 2P: CXXFLAGS="-O3 -march=native" CFLAGS="-O3 -march=native" - EPYC 7742 2P: CXXFLAGS="-O3 -march=native" CFLAGS="-O3 -march=native" - Ampere Altra 29 - nomarch/mtune: CXXFLAGS=-O3 CFLAGS=-O3 - Ampere Altra 2P - Just march: CXXFLAGS="-O3 -march=armv8.2-a" CFLAGS="-O3 -march=armv8.2-a" - Ampere Altra 2P - Just mtune: CXXFLAGS="-O3 -mtune=neoverse-n1" CFLAGS="-O3 -mtune=neoverse-n1" Compiler Details - Ampere Altra 2P: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --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++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-nls --enable-objc-gc=auto --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 --with-target-system-zlib=auto -v - Xeon Platinum 8280 2P: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --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++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none=/build/gcc-10-JvwpWM/gcc-10-10.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-10-JvwpWM/gcc-10-10.2.0/debian/tmp-gcn/usr,hsa --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=auto --with-tune=generic --without-cuda-driver -v - EPYC 7742 2P: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --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++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none=/build/gcc-10-JvwpWM/gcc-10-10.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-10-JvwpWM/gcc-10-10.2.0/debian/tmp-gcn/usr,hsa --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=auto --with-tune=generic --without-cuda-driver -v - Ampere Altra 29 - nomarch/mtune: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --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++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-nls --enable-objc-gc=auto --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 --with-target-system-zlib=auto -v - Ampere Altra 2P - Just march: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --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++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-nls --enable-objc-gc=auto --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 --with-target-system-zlib=auto -v - Ampere Altra 2P - Just mtune: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --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++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-nls --enable-objc-gc=auto --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 --with-target-system-zlib=auto -v Disk Details - Ampere Altra 2P: NONE / relatime,rw / Block Size: 4096 - Xeon Platinum 8280 2P: NONE / errors=remount-ro,relatime,rw / Block Size: 4096 - EPYC 7742 2P: NONE / errors=remount-ro,relatime,rw / Block Size: 4096 Processor Details - Ampere Altra 2P: Scaling Governor: cppc_cpufreq performance (Boost: Enabled) - Xeon Platinum 8280 2P: Scaling Governor: intel_pstate performance - CPU Microcode: 0x5003003 - EPYC 7742 2P: Scaling Governor: acpi-cpufreq performance - CPU Microcode: 0x8301034 - Ampere Altra 29 - nomarch/mtune: Scaling Governor: cppc_cpufreq performance (Boost: Enabled) - Ampere Altra 2P - Just march: Scaling Governor: cppc_cpufreq performance (Boost: Enabled) - Ampere Altra 2P - Just mtune: Scaling Governor: cppc_cpufreq performance (Boost: Enabled) Java Details - Ampere Altra 2P, Xeon Platinum 8280 2P, EPYC 7742 2P: OpenJDK Runtime Environment (build 11.0.9.1+1-Ubuntu-0ubuntu1.20.10) Python Details - Ampere Altra 2P: Python 3.8.6 - Xeon Platinum 8280 2P: Python 2.7.18 + Python 3.8.6 - EPYC 7742 2P: Python 2.7.18 + Python 3.8.6 - Ampere Altra 29 - nomarch/mtune: Python 3.8.6 - Ampere Altra 2P - Just march: Python 3.8.6 - Ampere Altra 2P - Just mtune: Python 3.8.6 Security Details - Ampere Altra 2P: 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 - Xeon Platinum 8280 2P: itlb_multihit: KVM: Mitigation of VMX disabled + 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 Enhanced IBRS IBPB: conditional RSB filling + srbds: Not affected + tsx_async_abort: Mitigation of TSX disabled - EPYC 7742 2P: 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: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected - Ampere Altra 29 - nomarch/mtune: 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 - Ampere Altra 2P - Just march: 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 - Ampere Altra 2P - Just mtune: 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
Ampere Altra Mt Jade vs. Xeon vs. EPYC Benchmarks astcenc: Exhaustive basis: UASTC Level 3 build-ffmpeg: Time To Compile build-llvm: Time To Compile build-php: Time To Compile compress-lz4: 1 - Compression Speed compress-lz4: 9 - Compression Speed compress-lz4: 9 - Decompression Speed coremark: CoreMark Size 666 - Iterations Per Second hint: FLOAT hpcg: m-queens: Time To Solve numpy: openssl: RSA 4096-bit Performance rodinia: OpenMP LavaMD pgbench: 1000 - 250 - Read Write pgbench: 1000 - 250 - Read Write - Average Latency phpbench: PHP Benchmark Suite pybench: Total For Average Test Times redis: GET rocksdb: Rand Read tnn: CPU - MobileNet v2 tnn: CPU - SqueezeNet v1.1 blender: Classroom - CPU-Only blender: Barbershop - CPU-Only compress-lz4: 1 - Decompression Speed Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 25.05 11.509 20.060 281.122 54.162 9460.62 44.69 9548.0 3705087.522089 355417184.82856 44.8021 6.257 237.16 8573.4 31.771 23154 10.857 529344 1304 1846629.65 890738361 317.810 290.800 65.33 219.31 52.20 19.626 19.959 225.600 38.376 5417.76 47.68 6361.8 1844054.141034 382187740.79717 28.8477 17.984 329.32 13716.6 56.423 37660 6.657 652140 1096 2488006.9 247216815 337.357 326.466 84.56 162.32 21.60 13.613 16.403 203.918 39.368 9477.91 44.35 10378.5 4002710.553017 291664268.00303 32.3962 7.693 300.56 25075.0 31.317 24214 10.618 515977 1173 1973922.33 482174282 332.587 300.670 44.96 131.02 25.16 11.357 18.961 291.600 60.311 9546.04 44.88 9461.6 3963890.483245 352844884.35696 44.6980 6.511 240.40 8667.4 31.527 23431 10.750 519124 1307 2035777.13 907899101 320.310 276.386 65.61 223.20 10349.8 25.19 11.414 19.131 286.200 61.993 9525.57 44.71 9293.0 4021112.879475 353246293.48793 44.6129 6.416 239.61 8636.2 30.654 25034 10.035 522951 1308 1980047.51 902448814 336.109 267.823 65.71 222.63 10625.0 25.45 11.526 19.791 286.956 61.943 9365.54 44.94 9151.3 4111356.659856 352166438.37569 44.6199 6.547 242.32 8687.7 28.964 23582 10.669 520476 1308 2082583.42 902181816 318.105 250.017 65.77 10286.0 OpenBenchmarking.org
ASTC Encoder Preset: Exhaustive OpenBenchmarking.org Seconds, Fewer Is Better ASTC Encoder 2.0 Preset: Exhaustive Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 12 24 36 48 60 SE +/- 0.06, N = 3 SE +/- 0.02, N = 3 SE +/- 0.01, N = 3 SE +/- 0.04, N = 3 SE +/- 0.04, N = 3 SE +/- 0.07, N = 3 25.05 52.20 21.60 25.16 25.19 25.45 -mfpmath=sse -mavx2 -mpopcnt -mfpmath=sse -mavx2 -mpopcnt 1. (CXX) g++ options: -std=c++14 -fvisibility=hidden -O3 -flto -lpthread
ASTC Encoder CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 37.0 291.0 405.0 Xeon Platinum 8280 2P 28.9 328.0 382.7 EPYC 7742 2P 103.6 332.9 453.7 Ampere Altra 29 - nomarch/mtune 35.0 284.0 403.0 OpenBenchmarking.org Watts, Fewer Is Better ASTC Encoder 2.0 CPU Power Consumption Monitor 120 240 360 480 600
ASTC Encoder CPU Temperature Monitor Min Avg Max Ampere Altra 2P 47.0 74.4 88.0 Xeon Platinum 8280 2P 37.0 62.3 69.0 EPYC 7742 2P 54.0 64.0 70.0 Ampere Altra 29 - nomarch/mtune 44.0 73.6 88.0 OpenBenchmarking.org Celsius, Fewer Is Better ASTC Encoder 2.0 CPU Temperature Monitor 20 40 60 80 100
Basis Universal Settings: UASTC Level 3 OpenBenchmarking.org Seconds, Fewer Is Better Basis Universal 1.12 Settings: UASTC Level 3 Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 5 10 15 20 25 SE +/- 0.07, N = 4 SE +/- 0.00, N = 3 SE +/- 0.07, N = 4 SE +/- 0.04, N = 5 SE +/- 0.20, N = 3 SE +/- 0.09, N = 3 11.51 19.63 13.61 11.36 11.41 11.53 1. (CXX) g++ options: -std=c++11 -fvisibility=hidden -fPIC -fno-strict-aliasing -O3 -rdynamic -lm -lpthread
Basis Universal CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 37.0 120.3 321.0 Xeon Platinum 8280 2P 28.6 239.4 425.8 EPYC 7742 2P 98.9 219.5 434.3 Ampere Altra 29 - nomarch/mtune 36.0 122.6 318.0 OpenBenchmarking.org Watts, Fewer Is Better Basis Universal 1.12 CPU Power Consumption Monitor 110 220 330 440 550
Basis Universal CPU Temperature Monitor Min Avg Max Ampere Altra 2P 47.0 54.5 75.0 Xeon Platinum 8280 2P 37.0 51.1 66.0 EPYC 7742 2P 53.8 59.3 66.8 Ampere Altra 29 - nomarch/mtune 48.0 55.3 76.0 OpenBenchmarking.org Celsius, Fewer Is Better Basis Universal 1.12 CPU Temperature Monitor 20 40 60 80 100
Timed FFmpeg Compilation Time To Compile OpenBenchmarking.org Seconds, Fewer Is Better Timed FFmpeg Compilation 4.2.2 Time To Compile Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 5 10 15 20 25 SE +/- 0.33, N = 15 SE +/- 0.13, N = 3 SE +/- 0.04, N = 3 SE +/- 0.25, N = 15 SE +/- 0.32, N = 3 SE +/- 0.36, N = 15 20.06 19.96 16.40 18.96 19.13 19.79
Timed FFmpeg Compilation CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 34.0 131.3 364.0 Xeon Platinum 8280 2P 28.8 216.1 466.2 EPYC 7742 2P 118.2 217.9 457.2 Ampere Altra 29 - nomarch/mtune 35.0 136.4 362.0 OpenBenchmarking.org Watts, Fewer Is Better Timed FFmpeg Compilation 4.2.2 CPU Power Consumption Monitor 120 240 360 480 600
Timed FFmpeg Compilation CPU Temperature Monitor Min Avg Max Ampere Altra 2P 46.0 55.6 77.0 Xeon Platinum 8280 2P 32.0 45.1 59.0 EPYC 7742 2P 57.0 61.5 69.0 Ampere Altra 29 - nomarch/mtune 47.0 56.8 78.0 OpenBenchmarking.org Celsius, Fewer Is Better Timed FFmpeg Compilation 4.2.2 CPU Temperature Monitor 20 40 60 80 100
Timed LLVM Compilation Time To Compile OpenBenchmarking.org Seconds, Fewer Is Better Timed LLVM Compilation 10.0 Time To Compile Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 60 120 180 240 300 SE +/- 3.72, N = 4 SE +/- 1.51, N = 3 SE +/- 0.64, N = 3 SE +/- 3.51, N = 3 SE +/- 4.06, N = 4 SE +/- 3.14, N = 3 281.12 225.60 203.92 291.60 286.20 286.96
Timed LLVM Compilation CPU Power Consumption Monitor OpenBenchmarking.org Watts, Fewer Is Better Timed LLVM Compilation 10.0 CPU Power Consumption Monitor Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 80 160 240 320 400 Min: 38 / Avg: 145.71 / Max: 341 Min: 29.29 / Avg: 275.86 / Max: 466.34 Min: 108.16 / Avg: 257.9 / Max: 457.83 Min: 37 / Avg: 146.08 / Max: 338
Timed LLVM Compilation CPU Temperature Monitor OpenBenchmarking.org Celsius, Fewer Is Better Timed LLVM Compilation 10.0 CPU Temperature Monitor Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 16 32 48 64 80 Min: 48 / Avg: 57.25 / Max: 77 Min: 38 / Avg: 57.29 / Max: 71 Min: 54.25 / Avg: 61.8 / Max: 71.25 Min: 48 / Avg: 59.5 / Max: 82
Timed PHP Compilation Time To Compile OpenBenchmarking.org Seconds, Fewer Is Better Timed PHP Compilation 7.4.2 Time To Compile Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 14 28 42 56 70 SE +/- 0.49, N = 3 SE +/- 0.23, N = 3 SE +/- 0.03, N = 3 SE +/- 1.41, N = 15 SE +/- 1.31, N = 15 SE +/- 1.52, N = 12 54.16 38.38 39.37 60.31 61.99 61.94
Timed PHP Compilation CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 37.0 102.6 312.0 Xeon Platinum 8280 2P 28.8 142.2 466.0 EPYC 7742 2P 99.2 179.9 454.9 Ampere Altra 29 - nomarch/mtune 37.0 105.3 336.0 OpenBenchmarking.org Watts, Fewer Is Better Timed PHP Compilation 7.4.2 CPU Power Consumption Monitor 120 240 360 480 600
Timed PHP Compilation CPU Temperature Monitor Min Avg Max Ampere Altra 2P 47.0 52.8 67.0 Xeon Platinum 8280 2P 35.0 41.7 61.0 EPYC 7742 2P 54.8 57.7 65.3 Ampere Altra 29 - nomarch/mtune 49.0 54.6 69.0 OpenBenchmarking.org Celsius, Fewer Is Better Timed PHP Compilation 7.4.2 CPU Temperature Monitor 20 40 60 80 100
LZ4 Compression Compression Level: 1 - Compression Speed OpenBenchmarking.org MB/s, More Is Better LZ4 Compression 1.9.3 Compression Level: 1 - Compression Speed Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 2K 4K 6K 8K 10K SE +/- 20.24, N = 3 SE +/- 16.87, N = 3 SE +/- 19.87, N = 3 SE +/- 109.92, N = 3 SE +/- 88.78, N = 3 SE +/- 107.16, N = 3 9460.62 5417.76 9477.91 9546.04 9525.57 9365.54 1. (CC) gcc options: -O3
LZ4 Compression Compression Level: 9 - Compression Speed OpenBenchmarking.org MB/s, More Is Better LZ4 Compression 1.9.3 Compression Level: 9 - Compression Speed Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 11 22 33 44 55 SE +/- 0.19, N = 3 SE +/- 0.04, N = 3 SE +/- 0.59, N = 5 SE +/- 0.16, N = 3 SE +/- 0.24, N = 3 SE +/- 0.01, N = 3 44.69 47.68 44.35 44.88 44.71 44.94 1. (CC) gcc options: -O3
LZ4 Compression Compression Level: 9 - Decompression Speed OpenBenchmarking.org MB/s, More Is Better LZ4 Compression 1.9.3 Compression Level: 9 - Decompression Speed Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 2K 4K 6K 8K 10K SE +/- 36.79, N = 3 SE +/- 29.10, N = 3 SE +/- 46.44, N = 5 SE +/- 124.78, N = 3 SE +/- 69.18, N = 3 SE +/- 55.01, N = 3 9548.0 6361.8 10378.5 9461.6 9293.0 9151.3 1. (CC) gcc options: -O3
LZ4 Compression Compression Level: 9 - Decompression Speed OpenBenchmarking.org MB/s Per Watt, More Is Better LZ4 Compression 1.9.3 Compression Level: 9 - Decompression Speed Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 30 60 90 120 150 124.17 67.45 67.36 119.26
LZ4 Compression CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 34.0 76.9 89.0 Xeon Platinum 8280 2P 28.6 94.3 100.4 EPYC 7742 2P 94.5 154.1 169.2 Ampere Altra 29 - nomarch/mtune 38.0 79.3 93.0 OpenBenchmarking.org Watts, Fewer Is Better LZ4 Compression 1.9.3 CPU Power Consumption Monitor 50 100 150 200 250
LZ4 Compression CPU Temperature Monitor Min Avg Max Ampere Altra 2P 47.0 50.0 51.0 Xeon Platinum 8280 2P 30.0 36.5 38.0 EPYC 7742 2P 54.5 58.9 61.5 Ampere Altra 29 - nomarch/mtune 48.0 49.9 50.0 OpenBenchmarking.org Celsius, Fewer Is Better LZ4 Compression 1.9.3 CPU Temperature Monitor 20 40 60 80 100
Coremark CoreMark Size 666 - Iterations Per Second OpenBenchmarking.org Iterations/Sec, More Is Better Coremark 1.0 CoreMark Size 666 - Iterations Per Second Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 900K 1800K 2700K 3600K 4500K SE +/- 113984.01, N = 15 SE +/- 9763.66, N = 3 SE +/- 877.48, N = 3 SE +/- 59006.00, N = 15 SE +/- 67060.40, N = 3 SE +/- 3957.68, N = 3 3705087.52 1844054.14 4002710.55 3963890.48 4021112.88 4111356.66 -march=armv8.2-a -mtune=neoverse-n1 -march=native -march=native -march=armv8.2-a -mtune=neoverse-n1 1. (CC) gcc options: -O2 -O3 -lrt" -lrt
Coremark CoreMark Size 666 - Iterations Per Second OpenBenchmarking.org Iterations/Sec Per Watt, More Is Better Coremark 1.0 CoreMark Size 666 - Iterations Per Second Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 4K 8K 12K 16K 20K 18385.43 5733.84 11499.36 19152.74
Coremark CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 35.0 201.5 304.0 Xeon Platinum 8280 2P 28.7 321.6 437.6 EPYC 7742 2P 109.3 348.1 434.2 Ampere Altra 29 - nomarch/mtune 34.0 207.0 308.0 OpenBenchmarking.org Watts, Fewer Is Better Coremark 1.0 CPU Power Consumption Monitor 110 220 330 440 550
Coremark CPU Temperature Monitor Min Avg Max Ampere Altra 2P 48.0 64.7 76.0 Xeon Platinum 8280 2P 34.0 58.8 71.0 EPYC 7742 2P 54.8 65.6 72.8 Ampere Altra 29 - nomarch/mtune 47.0 65.8 76.0 OpenBenchmarking.org Celsius, Fewer Is Better Coremark 1.0 CPU Temperature Monitor 20 40 60 80 100
Hierarchical INTegration Test: FLOAT OpenBenchmarking.org QUIPs, More Is Better Hierarchical INTegration 1.0 Test: FLOAT Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 80M 160M 240M 320M 400M SE +/- 964159.32, N = 3 SE +/- 896467.10, N = 3 SE +/- 3770188.53, N = 3 SE +/- 778890.43, N = 3 SE +/- 729646.49, N = 3 SE +/- 143355.25, N = 3 355417184.83 382187740.80 291664268.00 352844884.36 353246293.49 352166438.38 -march=armv8.2-a -mtune=neoverse-n1 -march=armv8.2-a -mtune=neoverse-n1 1. (CC) gcc options: -O3 -march=native -lm
Hierarchical INTegration Test: FLOAT OpenBenchmarking.org QUIPs Per Watt, More Is Better Hierarchical INTegration 1.0 Test: FLOAT Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 1000K 2000K 3000K 4000K 5000K 4660815.60 4009517.27 1925493.78 4409863.29
Hierarchical INTegration CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 35.0 76.3 84.0 Xeon Platinum 8280 2P 28.4 95.3 104.0 EPYC 7742 2P 111.4 151.5 156.1 Ampere Altra 29 - nomarch/mtune 38.0 80.0 88.0 OpenBenchmarking.org Watts, Fewer Is Better Hierarchical INTegration 1.0 CPU Power Consumption Monitor 40 80 120 160 200
Hierarchical INTegration CPU Temperature Monitor Min Avg Max Ampere Altra 2P 46.0 49.0 49.0 Xeon Platinum 8280 2P 31.0 33.3 34.0 EPYC 7742 2P 55.5 59.0 60.5 Ampere Altra 29 - nomarch/mtune 48.0 51.4 52.0 OpenBenchmarking.org Celsius, Fewer Is Better Hierarchical INTegration 1.0 CPU Temperature Monitor 20 40 60 80 100
High Performance Conjugate Gradient OpenBenchmarking.org GFLOP/s, More Is Better High Performance Conjugate Gradient 3.1 Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 10 20 30 40 50 SE +/- 0.03, N = 3 SE +/- 0.02, N = 3 SE +/- 0.07, N = 3 SE +/- 0.06, N = 3 SE +/- 0.03, N = 3 SE +/- 0.03, N = 3 44.80 28.85 32.40 44.70 44.61 44.62 1. (CXX) g++ options: -O3 -ffast-math -ftree-vectorize -pthread -lmpi_cxx -lmpi
High Performance Conjugate Gradient OpenBenchmarking.org GFLOP/s Per Watt, More Is Better High Performance Conjugate Gradient 3.1 Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 0.0293 0.0586 0.0879 0.1172 0.1465 0.13 0.08 0.08 0.12
High Performance Conjugate Gradient CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 37.0 352.3 423.0 Xeon Platinum 8280 2P 28.5 348.8 389.4 EPYC 7742 2P 111.5 417.9 438.1 Ampere Altra 29 - nomarch/mtune 39.0 360.7 429.0 OpenBenchmarking.org Watts, Fewer Is Better High Performance Conjugate Gradient 3.1 CPU Power Consumption Monitor 110 220 330 440 550
High Performance Conjugate Gradient CPU Temperature Monitor Min Avg Max Ampere Altra 2P 46.0 81.6 88.0 Xeon Platinum 8280 2P 30.0 62.6 69.0 EPYC 7742 2P 57.3 66.6 72.0 Ampere Altra 29 - nomarch/mtune 48.0 83.6 90.0 OpenBenchmarking.org Celsius, Fewer Is Better High Performance Conjugate Gradient 3.1 CPU Temperature Monitor 20 40 60 80 100
m-queens Time To Solve OpenBenchmarking.org Seconds, Fewer Is Better m-queens 1.2 Time To Solve Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 4 8 12 16 20 SE +/- 0.065, N = 8 SE +/- 0.069, N = 3 SE +/- 0.042, N = 6 SE +/- 0.052, N = 6 SE +/- 0.088, N = 15 SE +/- 0.087, N = 15 6.257 17.984 7.693 6.511 6.416 6.547 -march=armv8.2-a -mtune=neoverse-n1 -march=armv8.2-a -mtune=neoverse-n1 1. (CXX) g++ options: -fopenmp -O3 -O2 -march=native
m-queens CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 37.0 185.7 327.0 Xeon Platinum 8280 2P 28.4 333.7 410.6 EPYC 7742 2P 112.8 306.6 429.6 Ampere Altra 29 - nomarch/mtune 38.0 193.1 333.0 OpenBenchmarking.org Watts, Fewer Is Better m-queens 1.2 CPU Power Consumption Monitor 110 220 330 440 550
m-queens CPU Temperature Monitor Min Avg Max Ampere Altra 2P 50.0 66.8 81.0 Xeon Platinum 8280 2P 32.0 55.9 67.0 EPYC 7742 2P 54.3 62.9 72.3 Ampere Altra 29 - nomarch/mtune 51.0 65.1 81.0 OpenBenchmarking.org Celsius, Fewer Is Better m-queens 1.2 CPU Temperature Monitor 20 40 60 80 100
Numpy Benchmark OpenBenchmarking.org Score, More Is Better Numpy Benchmark Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 70 140 210 280 350 SE +/- 0.21, N = 3 SE +/- 0.43, N = 3 SE +/- 0.51, N = 3 SE +/- 2.99, N = 3 SE +/- 0.82, N = 3 SE +/- 0.54, N = 3 237.16 329.32 300.56 240.40 239.61 242.32
Numpy Benchmark OpenBenchmarking.org Score Per Watt, More Is Better Numpy Benchmark Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 0.765 1.53 2.295 3.06 3.825 2.93 3.40 1.94 2.91
Numpy Benchmark CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 35.0 80.9 93.0 Xeon Platinum 8280 2P 28.5 97.0 101.3 EPYC 7742 2P 103.7 155.0 166.9 Ampere Altra 29 - nomarch/mtune 39.0 82.7 96.0 OpenBenchmarking.org Watts, Fewer Is Better Numpy Benchmark CPU Power Consumption Monitor 50 100 150 200 250
Numpy Benchmark CPU Temperature Monitor Min Avg Max Ampere Altra 2P 48.0 51.0 52.0 Xeon Platinum 8280 2P 32.0 36.4 38.0 EPYC 7742 2P 54.8 56.6 59.5 Ampere Altra 29 - nomarch/mtune 49.0 51.7 52.0 OpenBenchmarking.org Celsius, Fewer Is Better Numpy Benchmark CPU Temperature Monitor 16 32 48 64 80
OpenSSL RSA 4096-bit Performance OpenBenchmarking.org Signs Per Second, More Is Better OpenSSL 1.1.1 RSA 4096-bit Performance Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 5K 10K 15K 20K 25K SE +/- 45.96, N = 3 SE +/- 22.54, N = 3 SE +/- 2.28, N = 3 SE +/- 0.22, N = 3 SE +/- 31.83, N = 3 SE +/- 0.32, N = 3 8573.4 13716.6 25075.0 8667.4 8636.2 8687.7 -march=armv8.2-a -mtune=neoverse-n1 -m64 -march=native -m64 -march=native -march=armv8.2-a -mtune=neoverse-n1 1. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl
OpenSSL RSA 4096-bit Performance OpenBenchmarking.org Signs Per Second Per Watt, More Is Better OpenSSL 1.1.1 RSA 4096-bit Performance Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 16 32 48 64 80 46.85 39.61 72.01 46.83
OpenSSL CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 39.0 183.0 227.0 Xeon Platinum 8280 2P 28.8 346.3 460.8 EPYC 7742 2P 96.1 348.2 409.5 Ampere Altra 29 - nomarch/mtune 38.0 185.1 228.0 OpenBenchmarking.org Watts, Fewer Is Better OpenSSL 1.1.1 CPU Power Consumption Monitor 120 240 360 480 600
OpenSSL CPU Temperature Monitor Min Avg Max Ampere Altra 2P 48.0 63.8 69.0 Xeon Platinum 8280 2P 32.0 59.3 69.0 EPYC 7742 2P 56.8 66.1 70.8 Ampere Altra 29 - nomarch/mtune 48.0 64.0 69.0 OpenBenchmarking.org Celsius, Fewer Is Better OpenSSL 1.1.1 CPU Temperature Monitor 20 40 60 80 100
Rodinia Test: OpenMP LavaMD OpenBenchmarking.org Seconds, Fewer Is Better Rodinia 3.1 Test: OpenMP LavaMD Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 13 26 39 52 65 SE +/- 0.80, N = 15 SE +/- 0.10, N = 3 SE +/- 0.15, N = 3 SE +/- 0.70, N = 15 SE +/- 0.82, N = 15 SE +/- 0.15, N = 3 31.77 56.42 31.32 31.53 30.65 28.96 1. (CXX) g++ options: -O2 -lOpenCL
Rodinia CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 37.0 246.4 440.0 Xeon Platinum 8280 2P 29.0 351.3 466.0 EPYC 7742 2P 107.2 355.0 452.8 Ampere Altra 29 - nomarch/mtune 37.0 248.1 436.0 OpenBenchmarking.org Watts, Fewer Is Better Rodinia 3.1 CPU Power Consumption Monitor 120 240 360 480 600
Rodinia CPU Temperature Monitor Min Avg Max Ampere Altra 2P 49.0 71.6 97.0 Xeon Platinum 8280 2P 35.0 63.2 73.0 EPYC 7742 2P 54.8 65.0 73.5 Ampere Altra 29 - nomarch/mtune 51.0 72.4 97.0 OpenBenchmarking.org Celsius, Fewer Is Better Rodinia 3.1 CPU Temperature Monitor 20 40 60 80 100
PostgreSQL pgbench Scaling Factor: 1000 - Clients: 250 - Mode: Read Write OpenBenchmarking.org TPS, More Is Better PostgreSQL pgbench 13.0 Scaling Factor: 1000 - Clients: 250 - Mode: Read Write Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 8K 16K 24K 32K 40K SE +/- 247.30, N = 12 SE +/- 366.26, N = 9 SE +/- 1158.09, N = 12 SE +/- 375.16, N = 12 SE +/- 59.38, N = 3 SE +/- 283.73, N = 12 23154 37660 24214 23431 25034 23582 -march=armv8.2-a -mtune=neoverse-n1 -march=native -march=native -march=armv8.2-a -mtune=neoverse-n1 1. (CC) gcc options: -fno-strict-aliasing -fwrapv -O3 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm
PostgreSQL pgbench Scaling Factor: 1000 - Clients: 250 - Mode: Read Write - Average Latency OpenBenchmarking.org ms, Fewer Is Better PostgreSQL pgbench 13.0 Scaling Factor: 1000 - Clients: 250 - Mode: Read Write - Average Latency Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 3 6 9 12 15 SE +/- 0.124, N = 12 SE +/- 0.062, N = 9 SE +/- 0.483, N = 12 SE +/- 0.181, N = 12 SE +/- 0.024, N = 3 SE +/- 0.132, N = 12 10.857 6.657 10.618 10.750 10.035 10.669 -march=armv8.2-a -mtune=neoverse-n1 -march=native -march=native -march=armv8.2-a -mtune=neoverse-n1 1. (CC) gcc options: -fno-strict-aliasing -fwrapv -O3 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm
PostgreSQL pgbench CPU Power Consumption Monitor OpenBenchmarking.org Watts, Fewer Is Better PostgreSQL pgbench 13.0 CPU Power Consumption Monitor Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 60 120 180 240 300 Min: 34 / Avg: 83.83 / Max: 208 Min: 28.69 / Avg: 111.71 / Max: 293.1 Min: 101.94 / Avg: 170.82 / Max: 334.86 Min: 34 / Avg: 87.67 / Max: 205
PostgreSQL pgbench CPU Temperature Monitor OpenBenchmarking.org Celsius, Fewer Is Better PostgreSQL pgbench 13.0 CPU Temperature Monitor Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 13 26 39 52 65 Min: 48 / Avg: 51.47 / Max: 58 Min: 31 / Avg: 33.72 / Max: 45 Min: 54.75 / Avg: 60.57 / Max: 69.5 Min: 49 / Avg: 52.79 / Max: 60
PHPBench PHP Benchmark Suite OpenBenchmarking.org Score, More Is Better PHPBench 0.8.1 PHP Benchmark Suite Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 140K 280K 420K 560K 700K SE +/- 658.53, N = 3 SE +/- 716.89, N = 4 SE +/- 5524.79, N = 3 SE +/- 3173.76, N = 3 SE +/- 595.52, N = 3 SE +/- 3585.35, N = 3 529344 652140 515977 519124 522951 520476
PHPBench PHP Benchmark Suite OpenBenchmarking.org Score Per Watt, More Is Better PHPBench 0.8.1 PHP Benchmark Suite Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 2K 4K 6K 8K 10K 7387.18 7783.17 3455.11 7113.99
PHPBench CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 35.0 71.7 82.0 Xeon Platinum 8280 2P 28.6 83.8 99.2 EPYC 7742 2P 108.6 149.3 155.3 Ampere Altra 29 - nomarch/mtune 36.0 73.0 85.0 OpenBenchmarking.org Watts, Fewer Is Better PHPBench 0.8.1 CPU Power Consumption Monitor 40 80 120 160 200
PHPBench CPU Temperature Monitor Min Avg Max Ampere Altra 2P 48.0 49.6 50.0 Xeon Platinum 8280 2P 32.0 33.1 34.0 EPYC 7742 2P 56.5 58.1 59.0 Ampere Altra 29 - nomarch/mtune 49.0 50.9 52.0 OpenBenchmarking.org Celsius, Fewer Is Better PHPBench 0.8.1 CPU Temperature Monitor 16 32 48 64 80
PyBench Total For Average Test Times OpenBenchmarking.org Milliseconds, Fewer Is Better PyBench 2018-02-16 Total For Average Test Times Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 300 600 900 1200 1500 SE +/- 1.20, N = 3 SE +/- 0.88, N = 3 SE +/- 5.29, N = 3 SE +/- 4.58, N = 3 SE +/- 4.70, N = 3 1304 1096 1173 1307 1308 1308
PyBench CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 37.0 73.3 82.0 Xeon Platinum 8280 2P 28.5 88.8 99.6 EPYC 7742 2P 113.6 150.2 155.4 Ampere Altra 29 - nomarch/mtune 35.0 75.3 86.0 OpenBenchmarking.org Watts, Fewer Is Better PyBench 2018-02-16 CPU Power Consumption Monitor 40 80 120 160 200
PyBench CPU Temperature Monitor Min Avg Max Ampere Altra 2P 47.0 48.9 50.0 Xeon Platinum 8280 2P 30.0 32.1 33.0 EPYC 7742 2P 53.8 55.3 58.3 Ampere Altra 29 - nomarch/mtune 48.0 50.7 51.0 OpenBenchmarking.org Celsius, Fewer Is Better PyBench 2018-02-16 CPU Temperature Monitor 16 32 48 64 80
Redis Test: GET OpenBenchmarking.org Requests Per Second, More Is Better Redis 6.0.9 Test: GET Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 500K 1000K 1500K 2000K 2500K SE +/- 25760.99, N = 15 SE +/- 4383.55, N = 5 SE +/- 34420.11, N = 15 SE +/- 25905.52, N = 15 SE +/- 20696.26, N = 15 SE +/- 16735.15, N = 3 1846629.65 2488006.90 1973922.33 2035777.13 1980047.51 2082583.42 -march=armv8.2-a -mtune=neoverse-n1 -march=native -march=native -march=armv8.2-a -mtune=neoverse-n1 1. (CXX) g++ options: -MM -MT -g3 -fvisibility=hidden -O3
Redis Test: GET OpenBenchmarking.org Requests Per Second Per Watt, More Is Better Redis 6.0.9 Test: GET Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 20K 40K 60K 80K 100K 46643.73 79620.82 16146.33 49532.84
Redis CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 34.0 39.6 78.0 Xeon Platinum 8280 2P 28.3 31.2 59.2 EPYC 7742 2P 102.2 122.3 150.7 Ampere Altra 29 - nomarch/mtune 34.0 41.1 81.0 OpenBenchmarking.org Watts, Fewer Is Better Redis 6.0.9 CPU Power Consumption Monitor 40 80 120 160 200
Redis CPU Temperature Monitor Min Avg Max Ampere Altra 2P 44.0 44.5 47.0 Xeon Platinum 8280 2P 25.0 26.0 26.0 EPYC 7742 2P 51.0 52.8 55.3 Ampere Altra 29 - nomarch/mtune 46.0 46.7 49.0 OpenBenchmarking.org Celsius, Fewer Is Better Redis 6.0.9 CPU Temperature Monitor 15 30 45 60 75
Facebook RocksDB Test: Random Read OpenBenchmarking.org Op/s, More Is Better Facebook RocksDB 6.3.6 Test: Random Read Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 200M 400M 600M 800M 1000M SE +/- 10379193.81, N = 15 SE +/- 385263.55, N = 3 SE +/- 5369165.24, N = 3 SE +/- 10981934.32, N = 12 SE +/- 9762288.83, N = 15 SE +/- 11322408.14, N = 15 890738361 247216815 482174282 907899101 902448814 902181816 -march=armv8-a+crc -march=armv8-a+crc -march=armv8-a+crc -march=armv8-a+crc 1. (CXX) g++ options: -O3 -march=native -std=c++11 -fno-builtin-memcmp -fno-rtti -rdynamic -lpthread
Facebook RocksDB Test: Random Read OpenBenchmarking.org Op/s Per Watt, More Is Better Facebook RocksDB 6.3.6 Test: Random Read Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune 500K 1000K 1500K 2000K 2500K 2204937.01 642060.42 1129501.20 2187339.52
Facebook RocksDB CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 36.0 404.0 462.0 Xeon Platinum 8280 2P 28.9 385.0 466.1 EPYC 7742 2P 98.9 426.9 454.4 Ampere Altra 29 - nomarch/mtune 37.0 415.1 474.0 OpenBenchmarking.org Watts, Fewer Is Better Facebook RocksDB 6.3.6 CPU Power Consumption Monitor 120 240 360 480 600
Facebook RocksDB CPU Temperature Monitor Min Avg Max Ampere Altra 2P 47.0 95.5 104.0 Xeon Platinum 8280 2P 34.0 67.8 73.0 EPYC 7742 2P 56.8 66.1 68.8 Ampere Altra 29 - nomarch/mtune 46.0 98.4 106.0 OpenBenchmarking.org Celsius, Fewer Is Better Facebook RocksDB 6.3.6 CPU Temperature Monitor 20 40 60 80 100
TNN Target: CPU - Model: MobileNet v2 OpenBenchmarking.org ms, Fewer Is Better TNN 0.2.3 Target: CPU - Model: MobileNet v2 Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 70 140 210 280 350 SE +/- 1.09, N = 3 SE +/- 0.62, N = 3 SE +/- 3.93, N = 3 SE +/- 5.02, N = 15 SE +/- 4.73, N = 15 SE +/- 1.47, N = 3 317.81 337.36 332.59 320.31 336.11 318.11 -march=armv8.2-a -mtune=neoverse-n1 - MIN: 297.54 / MAX: 422.5 -march=native - MIN: 335.4 / MAX: 423.46 -march=native - MIN: 321.8 / MAX: 864.72 MIN: 293.05 / MAX: 697.95 -march=armv8.2-a - MIN: 303.63 / MAX: 487.84 -mtune=neoverse-n1 - MIN: 298.96 / MAX: 416.05 1. (CXX) g++ options: -O3 -fopenmp -pthread -fvisibility=hidden -rdynamic -ldl
TNN CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 38.0 79.8 170.0 Xeon Platinum 8280 2P 28.4 110.3 139.5 EPYC 7742 2P 102.7 190.7 209.5 Ampere Altra 29 - nomarch/mtune 38.0 83.0 161.0 OpenBenchmarking.org Watts, Fewer Is Better TNN 0.2.3 CPU Power Consumption Monitor 60 120 180 240 300
TNN CPU Temperature Monitor Min Avg Max Ampere Altra 2P 48.0 50.6 52.0 Xeon Platinum 8280 2P 28.0 31.9 34.0 EPYC 7742 2P 55.0 60.4 63.5 Ampere Altra 29 - nomarch/mtune 51.0 54.4 57.0 OpenBenchmarking.org Celsius, Fewer Is Better TNN 0.2.3 CPU Temperature Monitor 20 40 60 80 100
TNN Target: CPU - Model: SqueezeNet v1.1 OpenBenchmarking.org ms, Fewer Is Better TNN 0.2.3 Target: CPU - Model: SqueezeNet v1.1 Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 70 140 210 280 350 SE +/- 3.77, N = 15 SE +/- 0.04, N = 3 SE +/- 0.45, N = 3 SE +/- 2.78, N = 15 SE +/- 0.26, N = 3 SE +/- 0.89, N = 3 290.80 326.47 300.67 276.39 267.82 250.02 -march=armv8.2-a -mtune=neoverse-n1 - MIN: 249.46 / MAX: 443.15 -march=native - MIN: 326.02 / MAX: 327.08 -march=native - MIN: 298.85 / MAX: 303.55 MIN: 250.72 / MAX: 300.03 -march=armv8.2-a - MIN: 266.94 / MAX: 268.91 -mtune=neoverse-n1 - MIN: 248.44 / MAX: 252.17 1. (CXX) g++ options: -O3 -fopenmp -pthread -fvisibility=hidden -rdynamic -ldl
TNN CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 34.0 72.0 84.0 Xeon Platinum 8280 2P 28.4 85.4 96.9 EPYC 7742 2P 104.0 146.4 153.8 Ampere Altra 29 - nomarch/mtune 36.0 75.5 92.0 OpenBenchmarking.org Watts, Fewer Is Better TNN 0.2.3 CPU Power Consumption Monitor 40 80 120 160 200
TNN CPU Temperature Monitor Min Avg Max Ampere Altra 2P 47.0 49.5 50.0 Xeon Platinum 8280 2P 30.0 31.0 32.0 EPYC 7742 2P 53.5 60.6 66.5 Ampere Altra 29 - nomarch/mtune 49.0 51.4 53.0 OpenBenchmarking.org Celsius, Fewer Is Better TNN 0.2.3 CPU Temperature Monitor 20 40 60 80 100
Blender Blend File: Classroom - Compute: CPU-Only OpenBenchmarking.org Seconds, Fewer Is Better Blender 2.83.5 Blend File: Classroom - Compute: CPU-Only Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 20 40 60 80 100 SE +/- 0.20, N = 3 SE +/- 0.16, N = 3 SE +/- 0.31, N = 3 SE +/- 0.15, N = 3 SE +/- 0.17, N = 3 SE +/- 0.16, N = 3 65.33 84.56 44.96 65.61 65.71 65.77
Blender Blend File: Barbershop - Compute: CPU-Only OpenBenchmarking.org Seconds, Fewer Is Better Blender 2.83.5 Blend File: Barbershop - Compute: CPU-Only Ampere Altra 2P Xeon Platinum 8280 2P EPYC 7742 2P Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march 50 100 150 200 250 SE +/- 2.76, N = 9 SE +/- 0.12, N = 3 SE +/- 1.78, N = 3 SE +/- 2.83, N = 3 SE +/- 2.70, N = 6 219.31 162.32 131.02 223.20 222.63
Blender CPU Power Consumption Monitor Min Avg Max Ampere Altra 2P 37.0 283.5 337.0 Ampere Altra 29 - nomarch/mtune 38.0 290.1 344.0 OpenBenchmarking.org Watts, Fewer Is Better Blender 2.83.5 CPU Power Consumption Monitor 80 160 240 320 400
Blender CPU Temperature Monitor Min Avg Max Ampere Altra 2P 47.0 72.2 80.0 Ampere Altra 29 - nomarch/mtune 48.0 74.4 81.0 OpenBenchmarking.org Celsius, Fewer Is Better Blender 2.83.5 CPU Temperature Monitor 20 40 60 80 100
Blender CPU Power Consumption Monitor OpenBenchmarking.org Watts, Fewer Is Better Blender 2.83.5 CPU Power Consumption Monitor Ampere Altra 2P Ampere Altra 29 - nomarch/mtune 60 120 180 240 300 Min: 36 / Avg: 196.59 / Max: 315 Min: 37 / Avg: 196.07 / Max: 312
Blender CPU Temperature Monitor OpenBenchmarking.org Celsius, Fewer Is Better Blender 2.83.5 CPU Temperature Monitor Ampere Altra 2P Ampere Altra 29 - nomarch/mtune 15 30 45 60 75 Min: 47 / Avg: 61.7 / Max: 78 Min: 49 / Avg: 63.95 / Max: 81
LZ4 Compression Compression Level: 1 - Decompression Speed OpenBenchmarking.org MB/s, More Is Better LZ4 Compression 1.9.3 Compression Level: 1 - Decompression Speed Ampere Altra 29 - nomarch/mtune Ampere Altra 2P - Just march Ampere Altra 2P - Just mtune 2K 4K 6K 8K 10K SE +/- 270.35, N = 3 SE +/- 21.91, N = 3 SE +/- 223.50, N = 3 10349.8 10625.0 10286.0 1. (CC) gcc options: -O3
LZ4 Compression Compression Level: 1 - Decompression Speed OpenBenchmarking.org MB/s Per Watt, More Is Better LZ4 Compression 1.9.3 Compression Level: 1 - Decompression Speed Ampere Altra 29 - nomarch/mtune 30 60 90 120 150 128.20
LZ4 Compression CPU Power Consumption Monitor Min Avg Max Ampere Altra 29 - nomarch/mtune 39.0 80.7 93.0 OpenBenchmarking.org Watts, Fewer Is Better LZ4 Compression 1.9.3 CPU Power Consumption Monitor 20 40 60 80 100
LZ4 Compression CPU Temperature Monitor Min Avg Max Ampere Altra 29 - nomarch/mtune 48.0 50.8 51.0 OpenBenchmarking.org Celsius, Fewer Is Better LZ4 Compression 1.9.3 CPU Temperature Monitor 15 30 45 60 75
Meta Performance Per Watts Performance Per Watts OpenBenchmarking.org Performance Per Watts, More Is Better Meta Performance Per Watts Performance Per Watts Ampere Altra 29 - nomarch/mtune 30K 60K 90K 120K 150K 161724.28
CPU Power Consumption Monitor Phoronix Test Suite System Monitoring OpenBenchmarking.org Watts CPU Power Consumption Monitor Phoronix Test Suite System Monitoring Ampere Altra 29 - nomarch/mtune 80 160 240 320 400 Min: 34 / Avg: 156.74 / Max: 474
CPU Temperature Monitor Phoronix Test Suite System Monitoring OpenBenchmarking.org Celsius CPU Temperature Monitor Phoronix Test Suite System Monitoring Ampere Altra 29 - nomarch/mtune 20 40 60 80 100 Min: 44 / Avg: 60.98 / Max: 106
Phoronix Test Suite v10.8.4