wgblas
    Preparing search index...

    Module benchmarks/nvidia-geforce-gtx-1650/dnrm2

    Benchmark results for dnrm2 on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0410 0.0063 0.1733 0.0015 416.7%
    64 0.0406 0.0126 0.1788 0.0029 435.3%
    128 0.0412 0.0249 0.1839 0.0056 444.3%
    512 0.0411 0.0998 0.1905 0.0215 464.0%
    1024 0.0411 0.1995 0.1935 0.0423 471.5%
    4096 0.0417 0.7865 0.2137 0.1534 512.7%
    16384 0.0442 2.9660 0.2340 0.5601 529.5%
    65536 0.0548 9.5757 0.0794 6.6025 145.0%
    262144 0.1032 20.3244 0.0929 22.5675 90.1%
    1048576 0.2501 33.5458 0.1521 55.1534 60.8%
    4194304 0.9116 36.8102 0.7697 43.5935 84.4%
    16777216 2.7197 49.3500 1.3886 96.6563 51.1%

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

    dnrm2-default GB/s chart

    dnrm2-default ms chart

    • dnrm2.js — WebGPU benchmark script
    • dnrm2.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 0.0319 0.0080 0.0428 0.0060 133.6%
    64 0.0317 0.0161 0.0428 0.0119 135.7%
    128 0.0319 0.0321 0.0510 0.0201 159.8%
    512 0.0319 0.1284 0.0509 0.0805 159.5%
    1024 0.0319 0.2572 0.0510 0.1607 160.0%
    4096 0.0325 1.0079 0.0591 0.5544 181.8%
    16384 0.0364 3.6009 0.0754 1.7389 207.1%
    65536 0.0566 9.2617 0.0756 6.9350 133.6%
    262144 0.1051 19.9562 0.1141 18.3780 108.6%
    1048576 0.2910 28.8229 0.2450 34.2448 84.2%
    4194304 0.9891 33.9257 0.7815 42.9357 79.0%

    dnrm2-stride4 GB/s chart

    dnrm2-stride4 ms chart

    Nvidia Geforce Gtx 1650 — stride = 5
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0410 0.0063 0.0427 0.0060 104.2%
    64 0.0410 0.0125 0.0426 0.0120 104.1%
    128 0.0403 0.0254 0.0507 0.0202 125.9%
    512 0.0409 0.1001 0.0507 0.0808 123.9%
    1024 0.0416 0.1967 0.0508 0.1614 121.9%
    4096 0.0418 0.7838 0.0592 0.5537 141.6%
    16384 0.0449 2.9205 0.0755 1.7367 168.2%
    65536 0.0691 7.5905 0.0760 6.8985 110.0%
    262144 0.1101 19.0429 0.1194 17.5676 108.4%
    1048576 0.3015 27.8270 0.2895 28.9758 96.0%

    dnrm2-stride5 GB/s chart

    dnrm2-stride5 ms chart

    Nvidia Geforce Gtx 1650 — stride = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0466 0.0055 0.0426 0.0060 91.5%
    64 0.0466 0.0110 0.0426 0.0120 91.5%
    128 0.0467 0.0219 0.0510 0.0201 109.1%
    512 0.0466 0.0879 0.0518 0.0790 111.2%
    1024 0.0480 0.1707 0.0509 0.1611 106.0%
    4096 0.0552 0.5938 0.0590 0.5552 107.0%
    16384 0.0660 1.9874 0.0875 1.4979 132.7%
    65536 0.1056 4.9656 0.1058 4.9566 100.2%
    262144 0.2508 8.3608 0.2163 9.6961 86.2%

    dnrm2-stride32 GB/s chart

    dnrm2-stride32 ms chart

    Nvidia Geforce Gtx 1650 — stride = 33
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    64 0.0490 0.0104 0.0425 0.0120 87.0%
    128 0.0488 0.0210 0.0507 0.0202 103.8%
    512 0.0492 0.0833 0.0504 0.0812 102.6%
    1024 0.0492 0.1667 0.0508 0.1614 103.3%
    4096 0.0573 0.5714 0.0589 0.5564 102.7%
    16384 0.0704 1.8605 0.0868 1.5098 123.2%
    65536 0.1087 4.8238 0.1051 4.9898 96.7%
    262144 0.2499 8.3908 0.1719 12.2018 68.8%

    dnrm2-stride33 GB/s chart

    dnrm2-stride33 ms chart

    Nvidia Geforce Gtx 1650 — stride = 255
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0493 0.0830 0.0508 0.0806 103.0%
    1024 0.0508 0.1613 0.0509 0.1610 100.2%
    4096 0.0569 0.5761 0.0591 0.5541 104.0%
    16384 0.0798 1.6423 0.1051 1.2467 131.7%
    65536 0.1432 3.6612 0.1451 3.6144 101.3%

    dnrm2-stride255 GB/s chart

    dnrm2-stride255 ms chart

    Nvidia Geforce Gtx 1650 — stride = 256
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0498 0.0823 0.0507 0.0808 101.9%
    1024 0.0511 0.1604 0.0506 0.1619 99.0%
    4096 0.0582 0.5634 0.0589 0.5559 101.4%
    16384 0.0740 1.7709 0.0959 1.3662 129.6%
    65536 0.1279 4.0980 0.1222 4.2896 95.5%

    dnrm2-stride256 GB/s chart

    dnrm2-stride256 ms chart

    See also:

    Settings
    Member Visibility

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