8 Commits

Author	SHA1	Message	Date
handsomezhuzhu	b8b4516b98	fix: replace ATB reshape_and_cache with pure PyTorch indexing	2026-02-10 19:56:47 +08:00
handsomezhuzhu	101435817a	fix: add initialize_cache method to NPU worker	2026-02-10 19:42:32 +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	c3631d65c2	fix: initialize TP/PP parallel groups after distributed environment	2026-02-10 19:14:29 +08:00
handsomezhuzhu	693e0a1d89	feat: add CUDA-to-NPU monkey patches for GPUModelRunner compatibility	2026-02-10 19:09:14 +08:00
handsomezhuzhu	0765fc9fd3	fix: pass world_size int to init_distributed_environment instead of vllm_config	2026-02-10 18:58:21 +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