9 Commits

Author	SHA1	Message	Date
handsomezhuzhu	5337842e92	fix: pure pytorch reshape_and_cache + _npu_flash_attention prefill	2026-02-10 20:33:14 +08:00
handsomezhuzhu	30cf7ccd1f	fix: revert to _npu_reshape_and_cache (contiguous) and _npu_flash_attention	2026-02-10 20:29:18 +08:00
handsomezhuzhu	a58c3fe973	fix: correct layout for npu_incre_flash_attention (BNSD requires B,H,1,D)	2026-02-10 20:23:03 +08:00
handsomezhuzhu	e7655a0745	fix: proper PrefillNoCache detection, fallback to npu_fusion_attention for chunked prefill (CANN compat)	2026-02-10 20:14:42 +08:00
handsomezhuzhu	810a2ef757	refactor: align attention with Huawei vllm-ascend - reshape_and_cache with kv_cache[0]/[1], _get_fia_params, npu_fused_infer_attention_score for chunked prefill, add actual_seq_lengths_q	2026-02-10 20:06:52 +08:00
handsomezhuzhu	b8b4516b98	fix: replace ATB reshape_and_cache with pure PyTorch indexing	2026-02-10 19:56:47 +08:00
handsomezhuzhu	7120cd803b	fix: KV cache shape needs leading 2 dim for key+value pair	2026-02-10 19:27:10 +08:00
handsomezhuzhu	a274fd82ad	fix: accept cache_dtype_str in get_kv_cache_shape	2026-02-10 19:23:20 +08:00
handsomezhuzhu	e75504df72	feat: initial vllm-npu-plugin for Ascend NPU adaptation ... - NPUPlatform: device management, HCCL process group, config adaptation - AscendAttentionBackend: npu_fusion_attention (prefill) + npu_incre_flash_attention (decode) - NPUCommunicator: HCCL-based distributed communication - NPUWorker: NPU device init, memory profiling - Custom ops: SiluAndMul, RMS norm, rotary embedding - Plugin registered via vllm.platform_plugins entry point Based on vllm-ascend official pattern, targeting Ascend 910B	2026-02-10 11:06:01 +08:00