wgblas
    Preparing search index...

    Module benchmarks/nvidia-geforce-gtx-1650/sasum

    Benchmark results for sasum on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0202 0.0063 0.0166 0.0077 82.1%
    64 0.0196 0.0131 0.0181 0.0141 92.6%
    128 0.0190 0.0269 0.0180 0.0284 94.7%
    512 0.0196 0.1046 0.0175 0.1173 89.2%
    1024 0.0197 0.2081 0.0179 0.2294 90.7%
    4096 0.0196 0.8339 0.0179 0.9143 91.2%
    16384 0.0197 3.3247 0.0175 3.7441 88.8%
    65536 0.0205 12.8000 0.0181 14.4863 88.4%
    262144 0.0272 38.5960 0.0183 57.3368 67.3%
    1048576 0.0461 91.0538 0.0438 95.7778 95.1%
    4194304 0.1166 143.8968 0.1736 96.6251 148.9%
    16777216 0.4083 164.3600 2.7608 24.3075 676.2%

    Efficiency = wgblas GB/s ÷ cuBLAS GB/s × 100. 100% means parity with cuBLAS; values above 100% mean wgblas achieved greater throughput.

    sasum-default GB/s chart

    sasum-default ms chart

    • sasum.js — WebGPU benchmark script
    • sasum.c — CUDA / cuBLAS reference script

    Unless noted otherwise, every result above uses unit stride (incx = incy = 1) — the normal case, and the coalesced, best-case GPU access pattern. Real usage sometimes passes a non-unit stride (e.g. operating on a row or column of a larger matrix, where incx = lda), which breaks memory coalescing and costs measurably more. This section sweeps a few representative strides to characterize that cost separately, collapsed below by default — expand a stride to see its table and chart.

    Nvidia Geforce Gtx 1650 — stride = 4
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0198 0.0065 0.0181 0.0071 91.2%
    64 0.0198 0.0129 0.0181 0.0141 91.7%
    128 0.0198 0.0259 0.0181 0.0283 91.5%
    512 0.0199 0.1031 0.0180 0.1135 90.9%
    1024 0.0200 0.2051 0.0180 0.2270 90.4%
    4096 0.0203 0.8076 0.0181 0.9062 89.1%
    16384 0.0208 3.1556 0.0181 3.6248 87.1%
    65536 0.0263 9.9538 0.0190 13.8145 72.1%
    262144 0.0444 23.6081 0.0363 28.8705 81.8%
    1048576 0.1118 37.5188 0.1041 40.2988 93.1%
    4194304 0.3795 44.2064 0.3733 44.9435 98.4%

    sasum-stride4 GB/s chart

    sasum-stride4 ms chart

    Nvidia Geforce Gtx 1650 — stride = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0202 0.0063 0.0179 0.0071 89.1%
    64 0.0202 0.0127 0.0172 0.0148 85.6%
    128 0.0205 0.0250 0.0170 0.0301 83.1%
    512 0.0201 0.1021 0.0174 0.1179 86.6%
    1024 0.0201 0.2035 0.0172 0.2375 85.7%
    4096 0.0205 0.7988 0.0181 0.9030 88.5%
    16384 0.0285 2.3011 0.0239 2.7453 83.8%
    65536 0.0445 5.8851 0.0394 6.6494 88.5%
    262144 0.1131 9.2696 0.1067 9.8284 94.3%
    1048576 0.3848 10.9013 0.3732 11.2392 97.0%

    sasum-stride32 GB/s chart

    sasum-stride32 ms chart

    Nvidia Geforce Gtx 1650 — stride = 256
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0209 0.0980 0.0181 0.1133 86.5%
    1024 0.0232 0.1769 0.0181 0.2265 78.1%
    4096 0.0241 0.6795 0.0176 0.9318 72.9%
    16384 0.0338 1.9376 0.0212 3.0867 62.8%
    65536 0.0600 4.3714 0.0500 5.2479 83.3%

    sasum-stride256 GB/s chart

    sasum-stride256 ms chart

    See also:

    Settings
    Member Visibility

    On This Page
    Nvidia Geforce Gtx 1650 — wgblas vs cuBLASSee alsoStride sweep