wgblas
    Preparing search index...

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

    Benchmark results for isamax on Nvidia Geforce Gtx 1650.

    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0205 0.0063 0.0169 0.0076 82.3%
    64 0.0209 0.0122 0.0166 0.0154 79.4%
    128 0.0203 0.0252 0.0168 0.0304 82.9%
    512 0.0209 0.0980 0.0171 0.1200 81.7%
    1024 0.0207 0.1978 0.0170 0.2406 82.2%
    4096 0.0206 0.7950 0.0182 0.9014 88.2%
    16384 0.0210 3.1267 0.0181 3.6152 86.5%
    65536 0.0215 12.1724 0.0181 14.4991 84.0%
    262144 0.0284 36.9425 0.0209 50.1423 73.7%
    1048576 0.0473 88.6520 0.0403 104.0667 85.2%
    4194304 0.1181 142.0450 0.1094 153.3006 92.7%
    16777216 0.4094 163.9040 0.4007 167.4640 97.9%

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

    isamax-default GB/s chart

    isamax-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 0.0210 0.0061 0.0138 0.0093 65.6%
    64 0.0205 0.0125 0.0165 0.0155 80.5%
    128 0.0206 0.0248 0.0162 0.0317 78.3%
    512 0.0216 0.0948 0.0166 0.1234 76.8%
    1024 0.0211 0.1942 0.0164 0.2505 77.5%
    4096 0.0208 0.7877 0.0175 0.9343 84.3%
    16384 0.0215 3.0476 0.0179 3.6637 83.2%
    65536 0.0273 9.5925 0.0193 13.5629 70.7%
    262144 0.0452 23.1986 0.0408 25.7105 90.2%
    1048576 0.1126 37.2364 0.1103 38.0305 97.9%
    4194304 0.3804 44.1004 0.3764 44.5747 98.9%

    isamax-stride4 GB/s chart

    isamax-stride4 ms chart

    Nvidia Geforce Gtx 1650 — stride = 32
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    32 0.0207 0.0062 0.0169 0.0076 81.2%
    64 0.0209 0.0123 0.0162 0.0158 77.7%
    128 0.0211 0.0243 0.0166 0.0308 78.9%
    512 0.0205 0.1000 0.0164 0.1252 79.9%
    1024 0.0209 0.1957 0.0172 0.2381 82.2%
    4096 0.0213 0.7682 0.0179 0.9176 83.7%
    16384 0.0293 2.2334 0.0223 2.9404 76.0%
    65536 0.0458 5.7287 0.0400 6.5510 87.4%
    262144 0.1145 9.1582 0.1074 9.7655 93.8%
    1048576 0.3861 10.8643 0.3768 11.1323 97.6%

    isamax-stride32 GB/s chart

    isamax-stride32 ms chart

    Nvidia Geforce Gtx 1650 — stride = 256
    n wgblas ms wgblas GB/s cuBLAS ms cuBLAS GB/s efficiency
    512 0.0213 0.0960 0.0164 0.1252 76.6%
    1024 0.0227 0.1803 0.0164 0.2490 72.4%
    4096 0.0258 0.6356 0.0181 0.9046 70.3%
    16384 0.0363 1.8076 0.0288 2.2756 79.4%
    65536 0.0645 4.0635 0.0508 5.1554 78.8%

    isamax-stride256 GB/s chart

    isamax-stride256 ms chart

    See also:

    Settings
    Member Visibility

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