CudaDMA
CudaDMA copied to clipboard
Emulating DMA Engines on GPUs for Performance and Portability
Does this work on tesla or pascal structure?
``` Enable multiple numbers of compile-time constants for CudaDMAHalo using template specialization. ``` Original issue reported on code.google.com by `[email protected]` on 11 Nov 2011 at 7:18
``` Recent performance optimizations applied to the CudaDMAStrided instance need to be added to the part of CudaDMAHalo that is similar to CudaDMAStrided. ``` Original issue reported on code.google.com by...