wgblas
    Preparing search index...

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

    Benchmark results for idamax on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0219 0.0117 0.0168 0.0152 77.1%
    64 0.0216 0.0237 0.0163 0.0315 75.4%
    128 0.0220 0.0466 0.0168 0.0610 76.5%
    512 0.0223 0.1840 0.0162 0.2525 72.9%
    1024 0.0222 0.3686 0.0169 0.4858 75.9%
    4096 0.0219 1.4949 0.0182 1.8012 83.0%
    16384 0.0226 5.7935 0.0181 7.2240 80.2%
    65536 0.0243 21.5579 0.0182 28.7691 74.9%
    262144 0.0370 56.6430 0.0386 54.3416 104.2%
    1048576 0.0710 118.0829 0.0746 112.4358 105.0%
    4194304 0.2080 161.3566 0.2240 149.7859 107.7%
    16777216 0.7693 174.4755 0.8238 162.9330 107.1%

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

    idamax-default GB/s chart

    idamax-default ms chart

    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 0.0225 0.0114 0.0125 0.0204 55.8%
    64 0.0220 0.0233 0.0170 0.0302 77.1%
    128 0.0219 0.0467 0.0169 0.0604 77.4%
    512 0.0224 0.1830 0.0170 0.2411 75.9%
    1024 0.0222 0.3683 0.0169 0.4839 76.1%
    4096 0.0224 1.4629 0.0183 1.7918 81.6%
    16384 0.0243 5.3966 0.0226 5.7976 93.1%
    65536 0.0364 14.3845 0.0311 16.8473 85.4%
    262144 0.0715 29.3423 0.0641 32.6944 89.7%
    1048576 0.2106 39.8395 0.1984 42.2813 94.2%
    4194304 0.7808 42.9744 0.7499 44.7450 96.0%

    idamax-stride4 GB/s chart

    idamax-stride4 ms chart

    Nvidia Geforce Gtx 1650 — stride = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0213 0.0120 0.0367 0.0070 171.5%
    64 0.0211 0.0243 0.0369 0.0139 174.8%
    128 0.0208 0.0492 0.0355 0.0288 170.7%
    512 0.0212 0.1935 0.0356 0.1150 168.3%
    1024 0.0209 0.3920 0.0325 0.2520 155.6%
    4096 0.0249 1.3154 0.0258 1.2720 103.4%
    16384 0.0349 3.7509 0.0452 2.9019 129.3%
    65536 0.0690 7.5957 0.0639 8.2064 92.6%
    262144 0.2046 10.2512 0.9072 2.3116 443.5%

    idamax-stride32 GB/s chart

    idamax-stride32 ms chart

    Nvidia Geforce Gtx 1650 — stride = 256
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0217 0.1891 0.0165 0.2483 76.1%
    1024 0.0212 0.3867 0.0163 0.5025 77.0%
    4096 0.0256 1.2776 0.1873 0.1749 730.5%
    16384 0.0450 2.9122 0.0295 4.4498 65.4%
    65536 0.1011 5.1873 0.0508 10.3239 50.2%

    idamax-stride256 GB/s chart

    idamax-stride256 ms chart

    See also:

    Settings
    Member Visibility

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