wgblas
    Preparing search index...

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

    Benchmark results for srot on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0069 0.0739 0.0029 0.1758 42.0%
    64 0.0073 0.1400 0.0028 0.3636 38.5%
    128 0.0074 0.2753 0.0028 0.7191 38.3%
    512 0.0072 1.1353 0.0029 2.8444 39.9%
    1024 0.0069 2.3649 0.0037 4.3761 54.0%
    4096 0.0074 8.8086 0.0035 18.6182 47.3%
    16384 0.0082 32.0000 0.0039 66.6016 48.0%
    65536 0.0102 102.4000 0.0070 150.3119 68.1%
    262144 0.0302 138.9210 0.0283 148.2715 93.7%
    1048576 0.1024 163.8400 0.1016 165.1821 99.2%
    4194304 0.3911 171.5813 0.3969 169.0842 101.5%
    16777216 1.5503 173.1466 1.5915 168.6644 102.7%

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

    srot-default GB/s chart

    srot-default ms chart

    • srot.js — WebGPU benchmark script
    • srot.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 0.0072 0.0714 0.0034 0.1509 47.3%
    64 0.0073 0.1407 0.0034 0.3005 46.8%
    128 0.0070 0.2929 0.0035 0.5792 50.6%
    512 0.0073 1.1179 0.0037 2.1880 51.1%
    1024 0.0073 2.2407 0.0038 4.3025 52.1%
    4096 0.0081 8.0630 0.0041 16.0000 50.4%
    16384 0.0103 25.5202 0.0066 39.4795 64.6%
    65536 0.0287 36.5714 0.0272 38.5053 95.0%
    262144 0.1010 41.5113 0.0996 42.1115 98.6%
    1048576 0.3908 42.9287 0.3871 43.3403 99.1%
    4194304 1.5422 43.5139 1.5391 43.6030 99.8%

    srot-stride4 GB/s chart

    srot-stride4 ms chart

    Nvidia Geforce Gtx 1650 — stride = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0061 0.0833 0.0030 0.1702 49.0%
    64 0.0061 0.1667 0.0030 0.3368 49.5%
    128 0.0061 0.3333 0.0032 0.6337 52.6%
    512 0.0063 1.3061 0.0042 1.9542 66.8%
    1024 0.0065 2.5222 0.0042 3.9084 64.5%
    4096 0.0095 6.9306 0.0057 11.4413 60.6%
    16384 0.0349 7.5122 0.0339 7.7393 97.1%
    65536 0.1411 7.4338 0.1331 7.8760 94.4%
    262144 0.5513 7.6083 0.5293 7.9245 96.0%
    1048576 2.2176 7.5656 2.1139 7.9367 95.3%

    srot-stride32 GB/s chart

    srot-stride32 ms chart

    Nvidia Geforce Gtx 1650 — stride = 256
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0063 1.2995 0.0042 1.9542 66.5%
    1024 0.0065 2.5222 0.0042 3.9084 64.5%
    4096 0.0106 6.1873 0.0090 7.3143 84.6%
    16384 0.0614 4.2667 0.0690 3.7979 112.3%
    65536 0.2888 3.6312 0.2950 3.5540 102.2%

    srot-stride256 GB/s chart

    srot-stride256 ms chart

    See also:

    Settings
    Member Visibility

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