ricky
ricky
From my rough benchmark test, the cgo call may lead performance problem in client program. Is there any better solution? @liu0x54
@Mergifyio backport branch-3.2
@jaogoy PTAL.
@mergifyio rebase
@mergifyio rebase
@mergifyio backport branch-3.3 branch-3.2 branch-3.1
> 从运行结果可以看到，BM_write_stream_then_read 跟 BM_write_streamed 运行耗时相近，似乎读对 steam 指令并没有影响 1. 从原始版本运行结果看，BM_write_stream_then_read 跟 BM_write_streamed 运行耗时相近，似乎读对 steam 指令并没有影响 2. 从删除omp parrallel for 版本运行结果看，BM_write_stream_then_read 运行耗时显著比 BM_write_streamed 长
> python 3.6.13 opencv-contrib-python==4.3.0.38 opencv-python==4.3.0.38 It works. Thanks
@alexey-milovidov Sorry for the late response. So should we implement vectorization that handles multiple elements at once? I misunderstood this issue. I'll try to modify the code.
@rschu1ze May I try on this?