Commits · d84b00c7debde0a4ca11ff053304fd1989853375 · gaoqiong / composable_kernel

"vscode:/vscode.git/clone" did not exist on "6e65d027c3422c672104ae8e4a7ad9998cf0db33"

10 Aug, 2023 3 commits
- remove v2r2 · d84b00c7
  letaoqin authored Aug 10, 2023
  
  d84b00c7
- grouped gemm add bias · 192dca60
  letaoqin authored Aug 10, 2023
  
  192dca60
- add bias to batched gemm · 53a74710
  letaoqin authored Aug 10, 2023
  
  53a74710
09 Aug, 2023 3 commits
- group change to multiple D · b8f08e67
  letaoqin authored Aug 09, 2023
  
  b8f08e67
- gridwise change to multiple D · fa066d60
  letaoqin authored Aug 09, 2023
  
  fa066d60
- format · ee275d4d
  letaoqin authored Aug 09, 2023
  
  ee275d4d
08 Aug, 2023 7 commits
- fix to block_work_idx_m · 1fa679c6
  letaoqin authored Aug 08, 2023
  
  1fa679c6
- Merge branch 'mha-train-develop' into mha-train-bias · d8e77f4a
  letaoqin authored Aug 08, 2023
  
  d8e77f4a
- batched gemm add do vector load · d173a2cb
  letaoqin authored Aug 08, 2023
  
  d173a2cb
- add verify d0 vector load · 9679ba63
  letaoqin authored Aug 08, 2023
  
  9679ba63
- add d vector load template parameters · 5d6bfabb
  letaoqin authored Aug 08, 2023
  
  5d6bfabb
- Merge pull request #834 from ROCmSoftwarePlatform/mha-train-develop-tinyfix · d20c472f
  Dan Yao authored Aug 08, 2023
```
Remove un-necessary including in the FlashAttention device operators
```
  d20c472f
- Remove un-necessary including in the FlashAttention device operators · ab7ef6b8
  Qianfeng Zhang authored Aug 08, 2023
  
  ab7ef6b8
07 Aug, 2023 1 commit
- group gemm add kernel · 8e7b98eb
  letaoqin authored Aug 07, 2023
  
  8e7b98eb
04 Aug, 2023 5 commits
- add parameters device · b158c537
  letaoqin authored Aug 04, 2023
  
  b158c537
- start add group · 81f4481c
  letaoqin authored Aug 04, 2023
  
  81f4481c
- rename file · cf9e7d5b
  letaoqin authored Aug 04, 2023
  
  cf9e7d5b
- change name to NumD0Tensor · 499dfe39
  letaoqin authored Aug 04, 2023
  
  499dfe39
- redefine interface · 381a7317
  letaoqin authored Aug 04, 2023
  
  381a7317
03 Aug, 2023 3 commits
- add dim=32 64 · 6fa4feac
  letaoqin authored Aug 03, 2023
  
  6fa4feac
- change bias pos · 9e49c2bf
  letaoqin authored Aug 03, 2023
  
  9e49c2bf
- add get_lane_local_1d_id · 1ab31830
  danyao12 authored Aug 03, 2023
  
  1ab31830
02 Aug, 2023 2 commits
- v2 add bias · de6af588
  letaoqin authored Aug 02, 2023
  
  de6af588
- Merge branch 'mha-train-develop' into mha-train-bias · 8e26a612
  letaoqin authored Aug 02, 2023
  
  8e26a612
01 Aug, 2023 1 commit
- fix input null to gridwise gemm · 927cfb23
  ltqin authored Aug 01, 2023
  
  927cfb23
31 Jul, 2023 2 commits
- add code to device · 18b1aa04
  ltqin authored Jul 31, 2023
  
  18b1aa04
- start forward bias · 5a72d8d6
  ltqin authored Jul 31, 2023
  
  5a72d8d6
28 Jul, 2023 4 commits
- Merge branch 'develop' into mha-train-develop · a59e8d48
  ltqin authored Jul 28, 2023
  
  a59e8d48
- fix name · f0984ca5
  ltqin authored Jul 28, 2023
  
  f0984ca5
- bwd qloop 2 kernels update mask · 7b915a10
  danyao12 authored Jul 28, 2023
  
  7b915a10
- Merge pull request #811 from ROCmSoftwarePlatform/attn-train-develop-qloop-mask · 2018bd28
  Dan Yao authored Jul 28, 2023
```
Add another mask(upper tringle from bottom right) to flash attetion
```
  2018bd28
27 Jul, 2023 9 commits
- Add s_nops after v_dot to avoid hazard (#808) · 7761e523
  Bartłomiej Kocot authored Jul 27, 2023
```
* Add s_nops after v_dot to avoid hazard

* Fix builtin for inner_produxt fp16

* Skip inline version to builtin

* Add comments regarding isa

* Fix comment regarding s_nop
```
  7761e523
- device_grouped_mha_fwd_xdl_cshuffle_v2 ignore · 120760d6
  ltqin authored Jul 27, 2023
  
  120760d6
- change IsTileSkippable parameters name · a822937a
  ltqin authored Jul 27, 2023
  
  a822937a
- Merge branch 'mha-train-develop' into attn-train-develop-qloop-mask · b0ae034e
  ltqin authored Jul 27, 2023
  
  b0ae034e
- add comments for MaskUpperTriangleFromBottomRightPredicate · 07206672
  ltqin authored Jul 27, 2023
  
  07206672
- fix offset name · fbe34dd4
  ltqin authored Jul 27, 2023
  
  fbe34dd4
- Merge branch 'mha-train-develop' of... · a0044283
  ltqin authored Jul 27, 2023
```
Merge branch 'mha-train-develop' of https://github.com/ROCmSoftwarePlatform/composable_kernel into mha-train-develop
```
  a0044283
- v2 ignore parameter · 94a39b18
  ltqin authored Jul 27, 2023
  
  94a39b18
- adjust naming · 159e74fa
  danyao12 authored Jul 27, 2023
  
  159e74fa