faiss petsc

benchmarks for a future article.

HTML result view exported from: https://openbenchmarking.org/result/2305210-NE-FAISSPETS21&sro&gru.

faiss petscProcessorMotherboardChipsetMemoryDiskGraphicsMonitorNetworkOSKernelCompilerFile-SystemScreen Resolutionab2 x AMD EPYC 9654 96-Core @ 2.40GHz (192 Cores / 384 Threads)AMD Titanite_4G (RTI1004D BIOS)AMD Device 14a41520GB2 x 1920GB SAMSUNG MZWLJ1T9HBJR-00007ASPEEDVGA HDMIBroadcom NetXtreme BCM5720 PCIeFedora Linux 386.2.15-300.fc38.x86_64 (x86_64)GCC 13.1.1 20230511 + Clang 16.0.3 + LLVM 16.0.3xfs640x480OpenBenchmarking.orgKernel Details- Transparent Huge Pages: madviseCompiler Details- --build=x86_64-redhat-linux --disable-libunwind-exceptions --enable-__cxa_atexit --enable-bootstrap --enable-cet --enable-checking=release --enable-gnu-indirect-function --enable-gnu-unique-object --enable-initfini-array --enable-languages=c,c++,fortran,objc,obj-c++,ada,go,d,m2,lto --enable-libstdcxx-backtrace --enable-link-serialization=1 --enable-multilib --enable-offload-defaulted --enable-offload-targets=nvptx-none --enable-plugin --enable-shared --enable-threads=posix --mandir=/usr/share/man --with-arch_32=i686 --with-build-config=bootstrap-lto --with-gcc-major-version-only --with-libstdcxx-zoneinfo=/usr/share/zoneinfo --with-linker-hash-style=gnu --with-tune=generic --without-cuda-driver Processor Details- Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa101111 Python Details- Python 3.11.3Security Details- SELinux + 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

faiss petscpetsc: Streamsfaiss: bench_polysemous_sift1m - PQ baselinefaiss: bench_polysemous_sift1m - Polysemous 64faiss: bench_polysemous_sift1m - Polysemous 62faiss: bench_polysemous_sift1m - Polysemous 58faiss: bench_polysemous_sift1m - Polysemous 54faiss: bench_polysemous_sift1m - Polysemous 50faiss: bench_polysemous_sift1m - Polysemous 46faiss: bench_polysemous_sift1m - Polysemous 42faiss: bench_polysemous_sift1m - Polysemous 38faiss: bench_polysemous_sift1m - Polysemous 34faiss: bench_polysemous_sift1m - Polysemous 30faiss: demo_sift1Mab411642.12673.8556.095.1353.1971.9231.2380.9150.7840.7410.7270.721393640.3598OpenBenchmarking.org

PETSc

Test: Streams

OpenBenchmarking.orgMB/s, More Is BetterPETSc 3.19Test: Streamsab90K180K270K360K450KSE +/- 20373.49, N = 8411642.13393640.361. (CC) gcc options: -fPIC -O3 -O2 -lpthread -lm

Faiss

Test: bench_polysemous_sift1m - PQ baseline

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - PQ baselinea0.86741.73482.60223.46964.3373.8551. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 64

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 64a2468106.091. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 62

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 62a1.15542.31083.46624.62165.7775.1351. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 58

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 58a0.71931.43862.15792.87723.59653.1971. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 54

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 54a0.43270.86541.29811.73082.16351.9231. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 50

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 50a0.27860.55720.83581.11441.3931.2381. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 46

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 46a0.20590.41180.61770.82361.02950.9151. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 42

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 42a0.17640.35280.52920.70560.8820.7841. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 38

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 38a0.16670.33340.50010.66680.83350.7411. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 34

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 34a0.16360.32720.49080.65440.8180.7271. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Faiss

Test: bench_polysemous_sift1m - Polysemous 30

OpenBenchmarking.orgms per query, Fewer Is BetterFaiss 1.7.4Test: bench_polysemous_sift1m - Polysemous 30a0.16220.32440.48660.64880.8110.7211. (F9X) gfortran options: -O2 -frecursive -m64 -fopenmp -msse3 -mssse3 -msse4.1 -mavx -mavx2 -fno-tree-vectorize -lm -lpthread -lgfortran -lomp

Phoronix Test Suite v10.8.4