reed

https://reed-lau.github.io/

Beijing China Ph.D. of Geophysics, HPC

Results 5 comments of


                                            reed

add publication: ‘EVT: Accelerating Deep Learning Training with Epilo…

@hwu36 could you help with reviewing it?

move the functions from rmw.h into multiple separate files

> each containing only functions of a specific category that sounds great! I just mean, `event` is not special enough.

[FEA] Add Prefetching Hints Support for Global Memory Loading

you mean 3.5？（since 3.4 have already released）

[BUG] tensor.hpp must be included before mma_atom.hpp

the dependency can be summarized as the following table: |File|Includes|Depends|Define|End includes| |-|-|-|-|-| |tensor.hpp|NA|NA|Tensor|gemm.hpp| |gemm.hpp|tensor.hpp/mma_atom.hpp|Tensor/MMA_Atom/ThrMMA|gemm func|NA| |mma_atom.hpp|tensor.hpp|Tensor|MMA_Atom/ThrMMA|NA| when we include tensor.hpp, after the preprocess, the code as: ``` Tensor; // include...

[BUG] tensor.hpp must be included before mma_atom.hpp

> @reed-lau It sounds like you would prefer Solution 2; is that right? yes.