wgblas
    Preparing search index...

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

    Benchmark results for drotm on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0089 0.1155 0.0041 0.2500 46.2%
    64 0.0086 0.2379 0.0041 0.5020 47.4%
    128 0.0088 0.4672 0.0041 0.9961 46.9%
    512 0.0087 1.8755 0.0044 3.6835 50.9%
    1024 0.0087 3.7509 0.0046 7.1111 52.7%
    4096 0.0091 14.3719 0.0045 28.9470 49.6%
    16384 0.0122 42.9463 0.0082 64.0000 67.1%
    65536 0.0246 85.3333 0.0213 98.5504 86.6%
    262144 0.0659 127.2853 0.0731 114.8244 110.9%
    1048576 0.2328 144.1540 0.2744 122.2686 117.9%
    4194304 0.8984 149.3884 1.0840 123.8207 120.6%
    16777216 3.1595 169.9251 3.4607 155.1349 109.5%

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

    drotm-default GB/s chart

    drotm-default ms chart

    • drotm.js — WebGPU benchmark script
    • drotm.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 0.0075 0.1362 0.0035 0.2949 46.2%
    64 0.0071 0.2870 0.0035 0.5792 49.6%
    128 0.0072 0.5727 0.0037 1.1082 51.7%
    512 0.0074 2.2213 0.0039 4.1967 52.9%
    1024 0.0075 4.3667 0.0039 8.3934 52.0%
    4096 0.0089 14.6547 0.0044 30.0073 48.8%
    16384 0.0161 32.6375 0.0097 54.2517 60.2%
    65536 0.0671 31.2597 0.0555 37.7729 82.8%
    262144 0.2740 30.6100 0.2109 39.7670 77.0%
    1048576 1.0916 30.7379 0.8298 40.4379 76.0%
    4194304 4.3705 30.7102 3.3073 40.5821 75.7%

    drotm-stride4 GB/s chart

    drotm-stride4 ms chart

    Nvidia Geforce Gtx 1650 — stride = 5
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0074 0.1385 0.0036 0.2883 48.0%
    64 0.0072 0.2857 0.0035 0.5818 49.1%
    128 0.0072 0.5651 0.0036 1.1429 49.4%
    512 0.0074 2.2165 0.0039 4.2490 52.2%
    1024 0.0076 4.3025 0.0040 8.1920 52.5%
    4096 0.0096 13.6990 0.0044 29.8978 45.8%
    16384 0.0245 21.4170 0.0186 28.1512 76.1%
    65536 0.0921 22.7714 0.0696 30.1176 75.6%
    262144 0.3763 22.2931 0.2694 31.1353 71.6%
    1048576 1.5050 22.2959 1.0660 31.4755 70.8%

    drotm-stride5 GB/s chart

    drotm-stride5 ms chart

    Nvidia Geforce Gtx 1650 — stride = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0076 0.1350 0.0036 0.2844 47.5%
    64 0.0076 0.2689 0.0037 0.5541 48.5%
    128 0.0076 0.5378 0.0038 1.0847 49.6%
    512 0.0081 2.0237 0.0045 3.6312 55.7%
    1024 0.0083 3.9385 0.0045 7.3143 53.8%
    4096 0.0205 6.3801 0.0067 19.6923 32.4%
    16384 0.0715 7.3290 0.0433 12.1183 60.5%
    65536 0.2785 7.5307 0.1826 11.4844 65.6%
    262144 1.1304 7.4207 0.7412 11.3169 65.6%

    drotm-stride32 GB/s chart

    drotm-stride32 ms chart

    Nvidia Geforce Gtx 1650 — stride = 33
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    64 0.0123 0.1667 0.0036 0.5740 29.0%
    128 0.0125 0.3274 0.0037 1.1034 29.7%
    512 0.0139 1.1784 0.0043 3.8496 30.6%
    1024 0.0148 2.2117 0.0044 7.5294 29.4%
    4096 0.0284 4.6178 0.0100 13.0446 35.4%
    16384 0.0842 6.2273 0.0491 10.6736 58.3%
    65536 0.3412 6.1467 0.2122 9.8825 62.2%
    262144 1.5528 5.4023 0.8791 9.5421 56.6%

    drotm-stride33 GB/s chart

    drotm-stride33 ms chart

    Nvidia Geforce Gtx 1650 — stride = 255
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0160 1.0230 0.0043 3.8209 26.8%
    1024 0.0170 1.9321 0.0043 7.6992 25.1%
    4096 0.0450 2.9112 0.0127 10.3565 28.1%
    16384 0.1659 3.1596 0.0726 7.2176 43.8%
    65536 2.8159 0.7448 0.3072 6.8267 10.9%

    drotm-stride255 GB/s chart

    drotm-stride255 ms chart

    Nvidia Geforce Gtx 1650 — stride = 256
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0145 1.1302 0.0044 3.6835 30.7%
    1024 0.0164 2.0000 0.0044 7.4203 27.0%
    4096 0.0389 3.3684 0.0101 12.9211 26.1%
    16384 0.1355 3.8696 0.0714 7.3438 52.7%
    65536 2.7089 0.7742 0.2980 7.0382 11.0%

    drotm-stride256 GB/s chart

    drotm-stride256 ms chart

    See also:

    Settings
    Member Visibility

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