gfx942 80cu HHS NN GridBased logic update

ROCm / hipBLASLt

hipBLASLt is a library that provides general matrix-matrix operations with a flexible API and extends functionalities beyond a traditional BLAS library

https://rocm.docs.amd.com/projects/hipBLASLt/en/latest/index.html

MIT License

49 stars 80 forks source link

gfx942 80cu HHS NN GridBased logic update #1112

Closed nakajee closed 2 weeks ago

nakajee commented 2 weeks ago

Reduced the number of sizes and kernels Before: size: 6855 kernel: 2974 After: size: 3391 kernel: 1807

nakajee commented 2 weeks ago

Closing this for now