composable_kernel

composable_kernel copied to clipboard

Add padding to weight-preshuffled

1

1.K padding 2.N padding

ltqin

## Proposed changes - add preshuffle gemm fp16 - same logic as fp8 and tune configs to reach 2.5TB/s ## Checklist Please put an `x` into the boxes that apply....

coderfeli

[Issue]: GPU verification of example problems fails

1

### Problem Description Running example binaries generated within composable_kernel/build/bin with GPU verification fails. CPU verification succeeds. See below:  PS: ROCm version 6.3.0 not 6.0.0 ### Operating System Ubuntu 22.04.2...

mysoreanoop

## Proposed changes Please describe the motivation behind the pull request, whether it enables a new feature or fixes a bug. If there are associated pull requests or issues, please...

mtgu0705

Add readme of fmha

1

## Proposed changes Add document of FMHA kernel ## Checklist Please put an `x` into the boxes that apply. You can also fill these out after creating the PR. If...

asleepzzz

## Proposed changes Please describe the motivation behind the pull request, whether it enables a new feature or fixes a bug. If there are associated pull requests or issues, please...

mozga-amd

## Proposed changes As titled ## Checklist Please put an `x` into the boxes that apply. You can also fill these out after creating the PR. If you're not sure,...

tenpercent

moe_sorting kernel num_tokens > 13K compute error. reproducible in aiter, can't reproducible in tile_example_moe_sorting

carlushuang

[CK_TILE] implement basic flatmm

2

## Proposed changes **add basic flatmm based on ck_tile:** - flatmm is placed in a seperate example folder - flatmm is using dependent kernel and pipeline and block function -...

feifei14119

## Proposed changes Please describe the motivation behind the pull request, whether it enables a new feature or fixes a bug. If there are associated pull requests or issues, please...

mozga-amd