wgblas
    Preparing search index...

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

    Benchmark results for sscal on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0070 0.0364 0.0030 0.0860 42.4%
    64 0.0068 0.0751 0.0031 0.1649 45.6%
    128 0.0070 0.1471 0.0031 0.3351 43.9%
    512 0.0072 0.5689 0.0032 1.2995 43.8%
    1024 0.0069 1.1852 0.0031 2.6667 44.4%
    4096 0.0072 4.5714 0.0034 9.5701 47.8%
    16384 0.0080 16.3840 0.0037 35.6174 46.0%
    65536 0.0102 51.5220 0.0051 102.4000 50.3%
    262144 0.0203 103.3691 0.0126 166.3350 62.1%
    1048576 0.0661 126.8234 0.0589 142.3922 89.1%
    4194304 0.2449 137.0061 0.2235 150.1505 91.2%
    16777216 0.8372 160.3174 0.8833 151.9455 105.5%

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

    sscal-default GB/s chart

    sscal-default ms chart

    • sscal.js — WebGPU benchmark script
    • sscal.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 0.0070 0.0367 0.0034 0.0755 48.6%
    64 0.0071 0.0719 0.0033 0.1538 46.8%
    128 0.0070 0.1468 0.0034 0.2991 49.1%
    512 0.0072 0.5714 0.0035 1.1636 49.1%
    1024 0.0072 1.1353 0.0038 2.1333 53.2%
    4096 0.0076 4.3207 0.0038 8.5690 50.4%
    16384 0.0084 15.6935 0.0048 27.3067 57.5%
    65536 0.0144 36.2879 0.0095 55.0723 65.9%
    262144 0.0532 39.3846 0.0512 40.9600 96.2%
    1048576 0.1971 42.5628 0.1950 43.0273 98.9%
    4194304 0.7743 43.3350 0.7705 43.5509 99.5%

    sscal-stride4 GB/s chart

    sscal-stride4 ms chart

    Nvidia Geforce Gtx 1650 — stride = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0070 0.0364 0.0028 0.0909 40.0%
    64 0.0072 0.0713 0.0028 0.1860 38.3%
    128 0.0074 0.1388 0.0030 0.3459 40.1%
    512 0.0074 0.5553 0.0035 1.1797 47.1%
    1024 0.0077 1.0579 0.0034 2.4381 43.4%
    4096 0.0085 3.8496 0.0041 7.9073 48.7%
    16384 0.0205 6.4000 0.0167 7.8618 81.4%
    65536 0.0677 7.7393 0.0635 8.2581 93.7%
    262144 0.2601 8.0630 0.2476 8.4710 95.2%
    1048576 1.0230 8.2003 0.9807 8.5539 95.9%

    sscal-stride32 GB/s chart

    sscal-stride32 ms chart

    Nvidia Geforce Gtx 1650 — stride = 256
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0076 0.5412 0.0033 1.2549 43.1%
    1024 0.0078 1.0535 0.0033 2.4615 42.8%
    4096 0.0085 3.8352 0.0041 8.0000 47.9%
    16384 0.0307 4.2667 0.0306 4.2823 99.6%
    65536 0.1428 3.6711 0.1390 3.7708 97.4%

    sscal-stride256 GB/s chart

    sscal-stride256 ms chart

    See also:

    Settings
    Member Visibility

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