Intel Kernel Scaling Optimizations On AMD Genoa

AMD EPYC 9654 benchmarks by Michael Larabel for a future article.

HTML result view exported from: https://openbenchmarking.org/result/2304019-NE-INTELKERN69&grs.

ProcessorMotherboardChipsetMemoryDiskGraphicsNetworkOSKernelDisplay ServerCompilerFile-SystemScreen ResolutionClear Linux 3 Threads 6 Threads 12 Threads 24 Threads 48 Threads 96 Threads 192 Threads 384 ThreadsAMD EPYC 9654 96-Core @ 2.40GHz (3 Cores)AMD Titanite_4G (RTI1004D BIOS)AMD Device 14a41520GB2 x 1920GB SAMSUNG MZWLJ1T9HBJR-00007ASPEEDBroadcom NetXtreme BCM5720 PCIeClear Linux OS 386606.2.8-1293.native (x86_64)X ServerGCC 12.2.1 20230323 releases/gcc-12.2.0-616-g1b6b7f214c + Clang 15.0.7 + LLVM 15.0.7ext4800x600AMD EPYC 9654 96-Core @ 2.40GHz (6 Cores)AMD EPYC 9654 96-Core @ 2.40GHz (12 Cores)AMD EPYC 9654 96-Core @ 2.40GHz (24 Cores)AMD EPYC 9654 96-Core @ 2.40GHz (48 Cores)AMD EPYC 9654 96-Core @ 2.40GHz (96 Cores)2 x AMD EPYC 9654 96-Core @ 2.40GHz (192 Cores)2 x AMD EPYC 9654 96-Core @ 2.40GHz (192 Cores / 384 Threads)OpenBenchmarking.orgKernel Details- Transparent Huge Pages: alwaysEnvironment Details- FFLAGS="-g -O3 -feliminate-unused-debug-types -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -m64 -fasynchronous-unwind-tables -Wp,-D_REENTRANT -ftree-loop-distribute-patterns -Wl,-z,now -Wl,-z,relro -malign-data=abi -fno-semantic-interposition -ftree-vectorize -ftree-loop-vectorize -Wl,--enable-new-dtags" CXXFLAGS="-g -O3 -feliminate-unused-debug-types -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -Wformat -Wformat-security -m64 -fasynchronous-unwind-tables -Wp,-D_REENTRANT -ftree-loop-distribute-patterns -Wl,-z,now -Wl,-z,relro -fno-semantic-interposition -ffat-lto-objects -fno-trapping-math -Wl,-sort-common -Wl,--enable-new-dtags -mrelax-cmpxchg-loop -fvisibility-inlines-hidden -Wl,--enable-new-dtags -std=gnu++17" FCFLAGS="-g -O3 -feliminate-unused-debug-types -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -m64 -fasynchronous-unwind-tables -Wp,-D_REENTRANT -ftree-loop-distribute-patterns -Wl,-z,now -Wl,-z,relro -malign-data=abi -fno-semantic-interposition -ftree-vectorize -ftree-loop-vectorize -Wl,-sort-common -Wl,--enable-new-dtags" CFLAGS="-g -O3 -feliminate-unused-debug-types -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -Wformat -Wformat-security -m64 -fasynchronous-unwind-tables -Wp,-D_REENTRANT -ftree-loop-distribute-patterns -Wl,-z,now -Wl,-z,relro -fno-semantic-interposition -ffat-lto-objects -fno-trapping-math -Wl,-sort-common -Wl,--enable-new-dtags -mrelax-cmpxchg-loop" THEANO_FLAGS="floatX=float32,openmp=true,gcc.cxxflags="-ftree-vectorize -mavx"" Compiler Details- --build=x86_64-generic-linux --disable-libmpx --disable-libunwind-exceptions --disable-multiarch --disable-vtable-verify --disable-werror --enable-__cxa_atexit --enable-bootstrap --enable-cet --enable-clocale=gnu --enable-default-pie --enable-gnu-indirect-function --enable-gnu-indirect-function --enable-host-shared --enable-languages=c,c++,fortran,go,jit --enable-ld=default --enable-libstdcxx-pch --enable-linux-futex --enable-lto --enable-multilib --enable-plugin --enable-shared --enable-threads=posix --exec-prefix=/usr --includedir=/usr/include --target=x86_64-generic-linux --with-arch=x86-64-v3 --with-gcc-major-version-only --with-glibc-version=2.35 --with-gnu-ld --with-isl --with-pic --with-ppl=yes --with-tune=sapphirerapids --with-zstd Processor Details- Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa101111Python Details- Python 3.11.2Security Details- itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Retpolines IBPB: conditional IBRS_FW STIBP: always-on RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected

onnx: fcn-resnet101-11 - CPU - Standardstress-ng: Mallocstress-ng: Context Switchingpgbench: 1000 - 1000 - Read Writepgbench: 1000 - 1000 - Read Write - Average Latencypgbench: 1000 - 800 - Read Writepgbench: 1000 - 800 - Read Write - Average Latencymemcached: 1:10pgbench: 1000 - 500 - Read Write - Average Latencypgbench: 1000 - 500 - Read Writerocksdb: Rand Readrocksdb: Read Rand Write Randmemcached: 1:5dragonflydb: 200 - 1:1gromacs: MPI CPU - water_GMX50_barerocksdb: Read While Writingdragonflydb: 200 - 1:5daphne: OpenMP - Points2Imagedragonflydb: 50 - 1:1dragonflydb: 50 - 1:5rocksdb: Update Randmysqlslap: 1024mysqlslap: 2048openvkl: vklBenchmark ISPCbuild-llvm: Ninjablosc: blosclz bitshuffleonnx: super-resolution-10 - CPU - Standardrocksdb: Rand Fillmysqlslap: 4096onnx: ArcFace ResNet-100 - CPU - Standardmysqlslap: 8192daphne: OpenMP - NDT Mappingblosc: blosclz shufflestress-ng: Semaphoresstress-ng: Pollmemcached: 1:100pgbench: 1000 - 800 - Read Onlypgbench: 1000 - 800 - Read Only - Average Latencypgbench: 1000 - 1000 - Read Onlypgbench: 1000 - 1000 - Read Only - Average Latencymysqlslap: 512onnx: GPT-2 - CPU - Standarddaphne: OpenMP - Euclidean Clusteronnx: super-resolution-10 - CPU - Standardonnx: ArcFace ResNet-100 - CPU - Standardonnx: fcn-resnet101-11 - CPU - Standardonnx: GPT-2 - CPU - Standardpgbench: 1000 - 500 - Read Only - Average Latencypgbench: 1000 - 500 - Read OnlyClear Linux 3 Threads 6 Threads 12 Threads 24 Threads 48 Threads 96 Threads 192 Threads 384 Threads0.638233746208.281598345.831133288.2501210766.080737199.5535.9091392625468762648929756342.521395674.140.5773846071528577.2255251.8460077731675279.581846162.52350254277268484281.6476726.142.050280031226413.03712581497.7214864.8274742.51207642.31731869.931315186.0831235088.09854598.36611580.0523.780376.70241566.8310.16363.4691441451.222052245686.963157050.102038549.0592190836.5261444643.6919.029262765056662212134061525644.942838943.131.0308964002991716.5956002.3221564042940540.923122983.64627962500363922301.1469456.580.1559116300930722.75913041693.3619957.9568067.97433822.351460141.392767492.8912666693.75054696.10681784.3012.475043.9369818.29210.4021.7122919942.148618631852.176465487.693148231.7663476823.0102231958.8412.5393987910145604316948292036420.913344564.181.98715673903544883.1549536.0795348493466341.223577457.306153746696141771191.61212388.2118.48887502149726.11304031845.2524813.71140550.30855199.572418070.754791761.6704583792.182690121.6111876.908.4390538.2934465.4148.220160.9585217783.6797629798829.2012990582.764723221.1725159315.5073257852.778.1806113719893008125566483318242.604338063.674.07928165364485799.5441767.7007082964403726.764558138.92746843880778326642.75716925.6122.641109220554130.18604641936.4630906.42441085.441741737.993753535.619016990.8878450181.184941132.4051908.988.1534133.2021271.7557.551210.49910032375.51762101730074.7225991362.226894814.5057356010.8764907076.236.3487876039204094734225433988403.905763835.977.19343538215965009.4037128.0392505755423657.305646481.60965084934871573374.52618569.1120.736127197357429.88374501625.8632904.24869808.413427214.564227589.8813877120.57712925210.775948130.7361901.438.2824933.5251181.2367.648400.33814827356.16716259187800.4848227338.578076012.386827019.6754894915.905.7108756263654868936521654000099.8410.828578718332428.1720708946281327.926525049.871210840906848864270.04920160.1121.149135760952531.93894371101.0730049.09675139.276738430.404533164.7210320930.77610780860.928901128.1281860.388.2547731.3153162.1517.803810.45611017484.86504455696317.9997933563.125540718.1486011413.4294273958.348.47559595129174157020199522591881.0619.355710640621042.2048971576070240.206279681.055120076636701154202.86512893.2122.38652363065222.9787538872.1719541.318664446.0313429013.794723725.639238330.8718583791.167663122.36941737.148.1707443.5198205.5508.270660.5049970794.61936605020044.78128197099.806471615.4536701111.9394517240.606.90572418113378407217671852621772.8619.3761218322514661.9421912604766356767051396192.72711685.393.719148887767215.9714613903.8415348.238002306.1627368220.788688154.178979050.8959134611.10067796.91381667.1410.673762.6187216.48110.31670.4981011900OpenBenchmarking.org

ONNX Runtime

Model: fcn-resnet101-11 - Device: CPU - Executor: Standard

MinAvgMaxClear Linux0.63.66.2OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.14Model: fcn-resnet101-11 - Device: CPU - Executor: Standard3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads3691215

Stress-NG

Test: Malloc

MinAvgMaxClear Linux746208182882102605020045OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.15.06Test: Malloc3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads160M320M480M640M800M

Stress-NG

Test: Context Switching

MinAvgMaxClear Linux159834640570104128197100OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.15.06Test: Context Switching3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads40M80M120M160M200M

PostgreSQL

Scaling Factor: 1000 - Clients: 1000 - Mode: Read Write

MinAvgMaxClear Linux113324753380760OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 1000 - Mode: Read Write3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads20K40K60K80K100K

PostgreSQL

Scaling Factor: 1000 - Clients: 1000 - Mode: Read Write - Average Latency

MinAvgMaxClear Linux12.431.388.3OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 1000 - Mode: Read Write - Average Latency3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads20406080100

PostgreSQL

Scaling Factor: 1000 - Clients: 800 - Mode: Read Write

MinAvgMaxClear Linux121075047082701OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 800 - Mode: Read Write3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads20K40K60K80K100K

PostgreSQL

Scaling Factor: 1000 - Clients: 800 - Mode: Read Write - Average Latency

MinAvgMaxClear Linux9.723.466.1OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 800 - Mode: Read Write - Average Latency3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads20406080100

Memcached

Set To Get Ratio: 1:10

MinAvgMaxClear Linux73720032831064907076OpenBenchmarking.orgOps/sec, More Is BetterMemcached 1.6.19Set To Get Ratio: 1:103 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads1.3M2.6M3.9M5.2M6.5M

PostgreSQL

Scaling Factor: 1000 - Clients: 500 - Mode: Read Write - Average Latency

MinAvgMaxClear Linux5.712.935.9OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 500 - Mode: Read Write - Average Latency3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads1020304050

PostgreSQL

Scaling Factor: 1000 - Clients: 500 - Mode: Read Write

MinAvgMaxClear Linux139265494487562OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 500 - Mode: Read Write3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads20K40K60K80K100K

RocksDB

Test: Random Read

MinAvgMaxClear Linux254687624788170981291741570OpenBenchmarking.orgOp/s, More Is BetterRocksDB 8.0Test: Random Read3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads400M800M1200M1600M2000M

RocksDB

Test: Read Random Write Random

MinAvgMaxClear Linux64892921219573652165OpenBenchmarking.orgOp/s, More Is BetterRocksDB 8.0Test: Read Random Write Random3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads1000K2000K3000K4000K5000K

Memcached

Set To Get Ratio: 1:5

MinAvgMaxClear Linux75634326048514000100OpenBenchmarking.orgOps/sec, More Is BetterMemcached 1.6.19Set To Get Ratio: 1:53 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads1.1M2.2M3.3M4.4M5.5M

Dragonflydb

Clients: 200 - Set To Get Ratio: 1:1

MinAvgMaxClear Linux139567435362165763836OpenBenchmarking.orgOps/sec, More Is BetterDragonflydb 0.6Clients: 200 - Set To Get Ratio: 1:13 Threads6 Threads12 Threads24 Threads48 Threads1.4M2.8M4.2M5.6M7M

GROMACS

Implementation: MPI CPU - Input: water_GMX50_bare

MinAvgMaxClear Linux0.68.119.4OpenBenchmarking.orgNs Per Day, More Is BetterGROMACS 2023Implementation: MPI CPU - Input: water_GMX50_bare3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads612182430

RocksDB

Test: Read While Writing

MinAvgMaxClear Linux384607438694612183225OpenBenchmarking.orgOp/s, More Is BetterRocksDB 8.0Test: Read While Writing3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads4M8M12M16M20M

Dragonflydb

Clients: 200 - Set To Get Ratio: 1:5

MinAvgMaxClear Linux152857737031975965009OpenBenchmarking.orgOps/sec, More Is BetterDragonflydb 0.6Clients: 200 - Set To Get Ratio: 1:53 Threads6 Threads12 Threads24 Threads48 Threads1.6M3.2M4.8M6.4M8M

Darmstadt Automotive Parallel Heterogeneous Suite

Backend: OpenMP - Kernel: Points2Image

MinAvgMaxClear Linux146623847756002OpenBenchmarking.orgTest Cases Per Minute, More Is BetterDarmstadt Automotive Parallel Heterogeneous Suite 2021.11.02Backend: OpenMP - Kernel: Points2Image3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads14K28K42K56K70K

Dragonflydb

Clients: 50 - Set To Get Ratio: 1:1

MinAvgMaxClear Linux167528043230166281328OpenBenchmarking.orgOps/sec, More Is BetterDragonflydb 0.6Clients: 50 - Set To Get Ratio: 1:13 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads1.6M3.2M4.8M6.4M8M

Dragonflydb

Clients: 50 - Set To Get Ratio: 1:5

MinAvgMaxClear Linux184616345079946525050OpenBenchmarking.orgOps/sec, More Is BetterDragonflydb 0.6Clients: 50 - Set To Get Ratio: 1:53 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads2M4M6M8M10M

RocksDB

Test: Update Random

MinAvgMaxClear Linux3502546881251210840OpenBenchmarking.orgOp/s, More Is BetterRocksDB 8.0Test: Update Random3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads400K800K1200K1600K2000K

MariaDB

Clients: 1024

MinAvgMaxClear Linux277688934OpenBenchmarking.orgQueries Per Second, More Is BetterMariaDB 11.0.1Clients: 10243 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads2004006008001000

MariaDB

Clients: 2048

MinAvgMaxClear Linux268640871OpenBenchmarking.orgQueries Per Second, More Is BetterMariaDB 11.0.1Clients: 20483 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads2004006008001000

OpenVKL

Benchmark: vklBenchmark ISPC

MinAvgMaxClear Linux485791396OpenBenchmarking.orgItems / Sec, More Is BetterOpenVKL 1.3.1Benchmark: vklBenchmark ISPC3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads400800120016002000

Timed LLVM Compilation

Build System: Ninja

MinAvgMaxClear Linux19311824282OpenBenchmarking.orgSeconds, Fewer Is BetterTimed LLVM Compilation 16.0Build System: Ninja3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads11002200330044005500

C-Blosc

Test: blosclz bitshuffle

MinAvgMaxClear Linux67261360120160OpenBenchmarking.orgMB/s, More Is BetterC-Blosc 2.3Test: blosclz bitshuffle3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads5K10K15K20K25K

ONNX Runtime

Model: super-resolution-10 - Device: CPU - Executor: Standard

MinAvgMaxClear Linux42.1102.7122.6OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.14Model: super-resolution-10 - Device: CPU - Executor: Standard3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads4080120160200

RocksDB

Test: Random Fill

MinAvgMaxClear Linux4888779465801357609OpenBenchmarking.orgOp/s, More Is BetterRocksDB 8.0Test: Random Fill3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads400K800K1200K1600K2000K

MariaDB

Clients: 4096

MinAvgMaxClear Linux264504672OpenBenchmarking.orgQueries Per Second, More Is BetterMariaDB 11.0.1Clients: 40963 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads2004006008001000

ONNX Runtime

Model: ArcFace ResNet-100 - Device: CPU - Executor: Standard

MinAvgMaxClear Linux13.024.131.9OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.14Model: ArcFace ResNet-100 - Device: CPU - Executor: Standard3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads918273645

MariaDB

Clients: 8192

MinAvgMaxClear Linux258.0433.4613.0OpenBenchmarking.orgQueries Per Second, More Is BetterMariaDB 11.0.1Clients: 81923 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads160320480640800

Darmstadt Automotive Parallel Heterogeneous Suite

Backend: OpenMP - Kernel: NDT Mapping

MinAvgMaxClear Linux87214341936OpenBenchmarking.orgTest Cases Per Minute, More Is BetterDarmstadt Automotive Parallel Heterogeneous Suite 2021.11.02Backend: OpenMP - Kernel: NDT Mapping3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads5001000150020002500

C-Blosc

Test: blosclz shuffle

MinAvgMaxClear Linux148652354832904OpenBenchmarking.orgMB/s, More Is BetterC-Blosc 2.3Test: blosclz shuffle3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads8K16K24K32K40K

Stress-NG

Test: Semaphores

MinAvgMaxClear Linux274743945451838002306OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.15.06Test: Semaphores3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads10M20M30M40M50M

Stress-NG

Test: Poll

MinAvgMaxClear Linux207642677516027368221OpenBenchmarking.orgBogo Ops/s, More Is BetterStress-NG 0.15.06Test: Poll3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads7M14M21M28M35M

Memcached

Set To Get Ratio: 1:100

MinAvgMaxClear Linux73187038170328688154OpenBenchmarking.orgOps/sec, More Is BetterMemcached 1.6.19Set To Get Ratio: 1:1003 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads2M4M6M8M10M

PostgreSQL

Scaling Factor: 1000 - Clients: 800 - Mode: Read Only

MinAvgMaxClear Linux1315187538361387712OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 800 - Mode: Read Only3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads400K800K1200K1600K2000K

PostgreSQL

Scaling Factor: 1000 - Clients: 800 - Mode: Read Only - Average Latency

MinAvgMaxClear Linux0.61.86.1OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 800 - Mode: Read Only - Average Latency3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads3691215

PostgreSQL

Scaling Factor: 1000 - Clients: 1000 - Mode: Read Only

MinAvgMaxClear Linux1235087295031292521OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 1000 - Mode: Read Only3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads400K800K1200K1600K2000K

PostgreSQL

Scaling Factor: 1000 - Clients: 1000 - Mode: Read Only - Average Latency

MinAvgMaxClear Linux0.82.48.1OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 1000 - Mode: Read Only - Average Latency3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads3691215

MariaDB

Clients: 512

MinAvgMaxClear Linux545739948OpenBenchmarking.orgQueries Per Second, More Is BetterMariaDB 11.0.1Clients: 5123 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads2004006008001000

ONNX Runtime

Model: GPT-2 - Device: CPU - Executor: Standard

MinAvgMaxClear Linux96.1115.8132.4OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.14Model: GPT-2 - Device: CPU - Executor: Standard3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads4080120160200

Darmstadt Automotive Parallel Heterogeneous Suite

Backend: OpenMP - Kernel: Euclidean Cluster

MinAvgMaxClear Linux158017901909OpenBenchmarking.orgTest Cases Per Minute, More Is BetterDarmstadt Automotive Parallel Heterogeneous Suite 2021.11.02Backend: OpenMP - Kernel: Euclidean Cluster3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads5001000150020002500

ONNX Runtime

Model: super-resolution-10 - Device: CPU - Executor: Standard

MinAvgMaxClear Linux8.211.023.8OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.14Model: super-resolution-10 - Device: CPU - Executor: Standard3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads816243240

ONNX Runtime

Model: ArcFace ResNet-100 - Device: CPU - Executor: Standard

MinAvgMaxClear Linux31.345.476.7OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.14Model: ArcFace ResNet-100 - Device: CPU - Executor: Standard3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads20406080100

ONNX Runtime

Model: fcn-resnet101-11 - Device: CPU - Executor: Standard

MinAvgMaxClear Linux1624861567OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.14Model: fcn-resnet101-11 - Device: CPU - Executor: Standard3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads400800120016002000

ONNX Runtime

Model: GPT-2 - Device: CPU - Executor: Standard

MinAvgMaxClear Linux7.68.810.4OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.14Model: GPT-2 - Device: CPU - Executor: Standard3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads48121620

PostgreSQL

Scaling Factor: 1000 - Clients: 500 - Mode: Read Only - Average Latency

MinAvgMaxClear Linux0.31.13.5OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 500 - Mode: Read Only - Average Latency3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads1.813.625.437.249.05

PostgreSQL

Scaling Factor: 1000 - Clients: 500 - Mode: Read Only

MinAvgMaxClear Linux1441458193271482735OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 1000 - Clients: 500 - Mode: Read Only3 Threads6 Threads12 Threads24 Threads48 Threads96 Threads192 Threads384 Threads400K800K1200K1600K2000K

Phoronix Test Suite v10.8.4