Commits · ad141d07228459d32f344e2bc121296665badd10 · OpenDAS / vllm_cscc

13 Oct, 2025 3 commits
- 去掉all2all ep相关代码 · 0b467604
  王敏 authored Oct 13, 2025
  
  0b467604
- update the envs for moe_sum and moe_align · 766663e6
  zhuwenwen authored Oct 13, 2025
  
  766663e6
- optimize the implementation of moe_sum · 06a1bee2
  zhuwenwen authored Oct 13, 2025
  
  06a1bee2
10 Oct, 2025 1 commit
- token split by token adapt to pd separation & p2p can be used async · 1a135a9d
  maxiao1 authored Oct 10, 2025
  
  1a135a9d
09 Oct, 2025 1 commit
- Revert "Merge branch 'v0.9.2-dev-lmcache-pd' into 'v0.9.2-dev'" · 7b30ecd8
  zhuwenwen authored Oct 09, 2025
```
This reverts merge request !219
```
  7b30ecd8
30 Sep, 2025 1 commit
- [feat]优化mori计算逻辑，支持cudagraph，按照bs*ep_size截断fused_moe的输入，共享专家不tp切分，去掉最后的allreduce · d2e57a90
  王敏 authored Sep 30, 2025
  
  d2e57a90
24 Sep, 2025 3 commits
- update VLLM_USE_OPT_CAT · 4d97c5fc
  zhuwenwen authored Sep 24, 2025
  
  4d97c5fc
- [kernel] add lightop's moe_sum(mul+add) fusion operator for deepseek · 8d2cac26
  zhuwenwen authored Sep 24, 2025
```
[FIX] 修复mtp和VLLM_USE_TRITON_CAT不能一起开的bug
```
  8d2cac26
- add env · c6b7a44b
  yangshj1 authored Sep 24, 2025
  
  c6b7a44b
22 Sep, 2025 1 commit
- deepseek-r1-w4a8 mlp/moe调用silu-mul-quant融合 · 3964e019
  wujl5 authored Sep 22, 2025
  
  3964e019
18 Sep, 2025 1 commit
- [kernel] add VLLM_USE_DEEPSEEK_MOE_SUM_MUL_AND to use lightop's moe_sum fusion... · 0975d9e8
  zhuwenwen authored Sep 18, 2025
```
[kernel] add VLLM_USE_DEEPSEEK_MOE_SUM_MUL_AND to use lightop's moe_sum fusion operator for deepseek
```
  0975d9e8
17 Sep, 2025 1 commit
- merge 092-dev分支近期修改 · f9f1887d
  王敏 authored Sep 17, 2025
  
  f9f1887d
14 Sep, 2025 1 commit
- deepseek-r1-w4a8使用rmsquant融合算子及横向融合 · fc443d52
  wujl5 authored Sep 14, 2025
  
  fc443d52
13 Sep, 2025 3 commits
- add VLLM_USE_MERGE_ATTN_STATES_OPT to control merge_attn_states support · 2071c380
  zhuwenwen authored Sep 13, 2025
  
  2071c380
- set VLLM_HAS_CONTEXT_DEFAULT=1 · 48b4c41d
  zhuwenwen authored Sep 13, 2025
  
  48b4c41d
- update triton cat scheduling · b1329ff2
  zhuwenwen authored Sep 13, 2025
```
update the default values of VLLM_USE_TRITON_CAT and VLLM_USE_LIGHT_OP to True
```
  b1329ff2
09 Sep, 2025 1 commit
- add VLLM_USE_TRITON_CAT to opt torch cat · 379315a0
  zhuwenwen authored Sep 09, 2025
  
  379315a0
07 Sep, 2025 1 commit
- add VLLM_USE_LIGHT_OP to optimize moe_align_block_size and moe_fused_gate · a54ab95d
  zhuwenwen authored Sep 07, 2025
  
  a54ab95d
04 Sep, 2025 1 commit
- 1.优化大EP，合入grouped gemm · d997afc4
  王敏 authored Sep 04, 2025
```
2.解决mtp >1 大EP推理all gather卡住问题
```
  d997afc4
01 Sep, 2025 1 commit
- [feat]添加all2all ep · ffa325e0
  王敏 authored Sep 01, 2025
  
  ffa325e0
19 Aug, 2025 1 commit
- set VLLM_USE_GLOBAL_CACHE13=False · f3e13c54
  zhuwenwen authored Aug 19, 2025
  
  f3e13c54
08 Aug, 2025 1 commit
- feat:新增VLLM_USE_GLOBAL_CACHE13 设置moe使用全局变量的cache13 · 333104ab
  jujl1 authored Jul 31, 2025
  
  333104ab
06 Aug, 2025 1 commit
- update VLLM_FLASH_ATTN_V1 to VLLM_USE_FLASH_ATTN_PA · 88dbf92c
  zhuwenwen authored Aug 06, 2025
  
  88dbf92c
04 Aug, 2025 1 commit
- add tbo on v1 engine · 20e75ed6
  lizhigong authored Aug 02, 2025
  
  20e75ed6
25 Jul, 2025 1 commit
- set VLLM_USE_V1=1 · e4a84fdc
  zhuwenwen authored Jul 25, 2025
  
  e4a84fdc
17 Jul, 2025 1 commit
- set default block_size and pa · b61f7a69
  zhuwenwen authored Jul 17, 2025
  
  b61f7a69
10 Jul, 2025 1 commit
- add apex rmsnorm · 3e0a595d
  zhuwenwen authored Jul 10, 2025
  
  3e0a595d
03 Jul, 2025 2 commits
- update VLLM_FLASH_ATTN_BACKEND to VLLM_FLASH_ATTN_V1 · 64fc5a29
  zhuwenwen authored Jul 03, 2025
  
  64fc5a29
- update VLLM_USE_FLASH_ATTN_BACKEND to VLLM_USE_FLASH_ATTN_PA · e036115e
  zhuwenwen authored Jul 03, 2025
  
  e036115e
02 Jul, 2025 2 commits
- set VLLM_USE_NN=1 · 058b32ae
  zhuwenwen authored Jul 02, 2025
  
  058b32ae
- [Refactor] Remove Unused Env `VLLM_ENABLE_MOE_ALIGN_BLOCK_SIZE_TRITON` (#20334) · 9dae7d46
  Wentao Ye authored Jul 01, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  9dae7d46
01 Jul, 2025 1 commit
- [CPU] Update custom ops for the CPU backend (#20255) · 6cc1e7d9
  Li, Jiang authored Jul 01, 2025
```
Signed-off-by: jiang1.li <jiang1.li@intel.com>
```
  6cc1e7d9
27 Jun, 2025 2 commits
- update cutlass fa and pa · 4ddb5447
  zhuwenwen authored Jun 27, 2025
  
  4ddb5447
- [Feature] add quick all reduce (#19744) · 0740e29b
  li haoyang authored Jun 27, 2025
```
Signed-off-by: ilmarkov <imarkov@redhat.com>
Signed-off-by: Haoyang Li <Haoyang.Li@amd.com>
Co-authored-by: ilmarkov <imarkov@redhat.com>
```
  0740e29b
25 Jun, 2025 3 commits
- [TPU] Add TPU specific var VLLM_TPU_MOST_MODEL_LEN (#19919) · 2d7620c3
  Chenyaaang authored Jun 25, 2025
```
Signed-off-by: Chenyaaang <chenyangli@google.com>
```
  2d7620c3
- [Quantization] Add compressed-tensors emulations support for NVFP4 (#19879) · 02c97d9a
  Dipika Sikka authored Jun 25, 2025
```
Signed-off-by: Dipika Sikka <dipikasikka1@gmail.com>
Signed-off-by: Dipika <dipikasikka1@gmail.com>
```
  02c97d9a
- [Kernels][Bugfix] Use torch op for all kernels in FusedMoE forward. Add... · 015fab8c
  bnellnm authored Jun 25, 2025
```
[Kernels][Bugfix] Use torch op for all kernels in FusedMoE forward.  Add additional testing for cudagraphs. (#19717)
Signed-off-by: Bill Nell <bnell@redhat.com>
```
  015fab8c
23 Jun, 2025 1 commit
- [Misc] Configurable timeout for execute_model RPC calls via env var (#19544) · f39ab2d4
  jinqinn authored Jun 23, 2025
```
Signed-off-by: jinqinn <goodqinjin@163.com>
```
  f39ab2d4
22 Jun, 2025 1 commit
- [BugFix] Add an env to disable moe chunking to work around compile incompatibility (#19642) · 33d51f59
  Ye (Charlotte) Qi authored Jun 22, 2025
```
Signed-off-by: Ye (Charlotte) Qi <yeq@meta.com>
```
  33d51f59
20 Jun, 2025 1 commit
- Export NaNs in logits to scheduler_stats if output is corrupted (#18777) · 2e3e3c86
  Vlad Tiberiu Mihailescu authored Jun 20, 2025
```
Signed-off-by: Vlad Mihailescu <vtmihailescu@gmail.com>
```
  2e3e3c86