Hello everyone,I’m trying to using stream to overlap data transfer and kernel execution.
I have run the same exe on two different computers, one installed RTX2060 and another installed RTX2080.
I found that while execution on RTX2060, the data transfer and kernel execution can’t overlapped. The DtoH memory copies began after all kernel executed completely, but transfer and execution overlapped on 2080.
I thought the settings of GPU on RTX2060 may be wrong, but I don’t know how can I fix it.
I really appreciate if anyone have ideas.