|
1 | 1 | # Release note |
2 | 2 |
|
| 3 | +## v0.9.1rc3 - 2025.08.22 |
| 4 | + |
| 5 | +This is the 3rd release candidate of v0.9.1 for vLLM Ascend. Please follow the [official doc](https://vllm-ascend.readthedocs.io/en/v0.9.1-dev/) to get started. |
| 6 | + |
| 7 | +### Core |
| 8 | + |
| 9 | +- MTP supports V1 scheduler [#2371](https://github.com/vllm-project/vllm-ascend/pull/2371) |
| 10 | +- Add LMhead TP communication groups [#1956](https://github.com/vllm-project/vllm-ascend/pull/1956) |
| 11 | +- Fix the bug that qwen3 moe doesn't work with aclgraph [#2478](https://github.com/vllm-project/vllm-ascend/pull/2478) |
| 12 | +- Fix `grammar_bitmask` IndexError caused by outdated `apply_grammar_bitmask` method [#2314](https://github.com/vllm-project/vllm-ascend/pull/2314) |
| 13 | +- Remove `chunked_prefill_for_mla` [#2177](https://github.com/vllm-project/vllm-ascend/pull/2177) |
| 14 | +- Fix bugs and refactor cached mask generation logic [#2326](https://github.com/vllm-project/vllm-ascend/pull/2326) |
| 15 | +- Fix configuration check logic about ascend scheduler [#2327](https://github.com/vllm-project/vllm-ascend/pull/2327) |
| 16 | +- Cancel the verification between deepseek-mtp and non-ascend scheduler in disaggregated-prefill deployment [#2368](https://github.com/vllm-project/vllm-ascend/pull/2368) |
| 17 | +- Fix issue that failed with ray distributed backend [#2306](https://github.com/vllm-project/vllm-ascend/pull/2306) |
| 18 | +- Fix incorrect req block length in ascend scheduler [#2394](https://github.com/vllm-project/vllm-ascend/pull/2394) |
| 19 | +- Fix header include issue in rope [#2398](https://github.com/vllm-project/vllm-ascend/pull/2398) |
| 20 | +- Fix mtp config bug [#2412](https://github.com/vllm-project/vllm-ascend/pull/2412) |
| 21 | +- Fix error info and adapt `attn_metedata` refactor [#2402](https://github.com/vllm-project/vllm-ascend/pull/2402) |
| 22 | +- Fix torchair runtime errror caused by configuration mismtaches and `.kv_cache_bytes` file missing [#2312](https://github.com/vllm-project/vllm-ascend/pull/2312) |
| 23 | +- Move `with_prefill` allreduce from cpu to npu [#2230](https://github.com/vllm-project/vllm-ascend/pull/2230) |
| 24 | + |
| 25 | +### Docs |
| 26 | + |
| 27 | +- Add document for deepseek large EP [#2339](https://github.com/vllm-project/vllm-ascend/pull/2339) |
| 28 | + |
| 29 | +### Known Issues |
| 30 | + |
| 31 | +- `test_aclgraph.py` failed with `"full_cuda_graph": True` on A2 (910B1) [#2182](https://github.com/vllm-project/vllm-ascend/issues/2182) |
| 32 | + |
3 | 33 | ## v0.9.1rc2 - 2025.08.06 |
4 | 34 | This is the 2nd release candidate of v0.9.1 for vLLM Ascend. Please follow the [official doc](https://vllm-ascend.readthedocs.io/en/v0.9.1-dev/) to get started. |
5 | 35 |
|
@@ -120,7 +150,7 @@ This is the 1st release candidate of v0.9.2 for vLLM Ascend. Please follow the [ |
120 | 150 | - Fix the accuracy problem with deploy models with parallel parameters. [#1678](https://github.com/vllm-project/vllm-ascend/pull/1678) |
121 | 151 | - The pre-built wheel package now requires lower version of glibc. Users can use it by `pip install vllm-ascend` directly. [#1582](https://github.com/vllm-project/vllm-ascend/pull/1582) |
122 | 152 |
|
123 | | -## Other |
| 153 | +### Other |
124 | 154 | - Official doc has been updated for better read experience. For example, more deployment tutorials are added, user/developer docs are updated. More guide will coming soon. |
125 | 155 | - Fix accuracy problem for deepseek V3/R1 models with torchair graph in long sequence predictions. [#1331](https://github.com/vllm-project/vllm-ascend/pull/1331) |
126 | 156 | - A new env variable `VLLM_ENABLE_FUSED_EXPERTS_ALLGATHER_EP` has been added. It enables the fused allgather-experts kernel for Deepseek V3/R1 models. The default value is `0`. [#1335](https://github.com/vllm-project/vllm-ascend/pull/1335) |
|
0 commit comments