| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
[Feature][ops]将 fused infer attention score 算子迁移到 MindIE-SD Co-authored-by: jiajin_qi<qijiajin1@huawei.com> # message auto-generated for no-merge-commit merge: !436 merge dev into dev [Feature][ops]将 fused infer attention score 算子迁移到 MindIE-SD Created-by: jiajin_qi Commit-by: jiajin_qi Merged-by: ascend-robot Description: Which issue(s) this PR fixes or accomplishes Fix part of #ISSUE 198 Purpose 为 incre_flash_attention 算子新增 host 侧完整 tiling 实现，支撑推理场景的 tiling 数据结构与计算流程。 主要新增内容（位于 csrc/ops/incre_flash_attention/op_host/）： tiling 数据结构定义：incre_flash_attention_tiling.h、incre_flash_attention_tiling_base.h、incre_flash_attention_tiling_struct.h，覆盖 Base / Core / SingleCore / SplitKV / InitOutput / Prefix 等参数 主 tiling 计算逻辑：incre_flash_attention_tiling.cpp，含对齐、分核、sparse 处理等基础工具 MLA（Multi-head Latent Attention）支持：incre_flash_attention_tiling_mla.h，定义 MLA 专属 Base/Core/SingleCore 参数与 AIC 核数约束 V2 版本实现：incre_flash_attention_tiling_v2.cpp、incre_flash_attention_tiling_v2.h，扩展 prefix 与 V2 tiling data tiling 校验：incre_flash_attention_tiling_check.cpp 上下文与实现细节：incre_flash_attention_tiling_context.h、incre_flash_attention_tiling_impl.h Test Plan 编译算子组件： bash build/build_ops.sh bash build/build_ascendc_ops.sh 运行 fused infer attention score UT： bash tests/ops/fused_infer_attention_score/run_fia_arch35_ut.sh 运行 Python 侧 fused infer attention score 测试： pytest tests/ops/fused_infer_attention_score/test_fused_infer_attention_score_v2.py Wan2.2 5B TI2V w8a8f8推理测试 Test Report 待补充： bash build/build_ops.sh: ok bash build/build_ascendc_ops.sh: ok bash tests/ops/fused_infer_attention_score/run_fia_arch35_ut.sh: passed pytest tests/ops/fused_infer_attention_score/test_fused_infer_attention_score_v2.py: passed Wan2.2 5B TI2V w8a8f8推理测试正常 See merge request: Ascend/MindIE-SD!436 | 30 天前 | |
[Feature][ops]迁移 fused infer attention score 算子到 MindIE-SD Co-authored-by: HAAZZZEEEE<wang-taicheng@qq.com> # message auto-generated for no-merge-commit merge: !407 merge haze/mig_fia_to_mindiesd into dev [Feature][ops]迁移 fused infer attention score 算子到 MindIE-SD Created-by: HAAZZZEEEE Commit-by: HAAZZZEEEE Merged-by: lanwangli Description: # Which issue(s) this PR fixes or accomplishes Fix part of #ISSUE 198 # Purpose 本 PR 将 fused infer attention score 算子迁移到 MindIE-SD。 主要变更： - 新增 csrc/ops/fused_infer_attention_score/ 下的算子实现，包括 op api、op host tiling/checker、op kernel、fallback graph 路径和 UT 框架。 - 新增 csrc/ops/common/、csrc/ops/incre_flash_attention/、csrc/ops/prompt_flash_attention/ 下的 FIA/flash-attention 公共 host 与 kernel 工具。 - 接入 fused infer attention score plugin，并更新相关 CMake/build 脚本。 - 新增 Python 层入口 mindiesd/layers/flash_attn/fused_infer_attention_score.py。 - 新增 tests/ops/fused_infer_attention_score/ 下的测试入口和 UT 脚本。 - 同步更新中英文 quantization 文档。 # Test Plan - 编译算子组件： - bash build/build_ops.sh - bash build/build_ascendc_ops.sh - 运行 fused infer attention score UT： - bash tests/ops/fused_infer_attention_score/run_fia_arch35_ut.sh - 运行 Python 侧 fused infer attention score 测试： - pytest tests/ops/fused_infer_attention_score/test_fused_infer_attention_score_v2.py - Wan2.2 5B TI2V w8a8f8推理测试 # Test Report 待补充： - bash build/build_ops.sh: ok - bash build/build_ascendc_ops.sh: ok - bash tests/ops/fused_infer_attention_score/run_fia_arch35_ut.sh: passed - pytest tests/ops/fused_infer_attention_score/test_fused_infer_attention_score_v2.py: passed - Wan2.2 5B TI2V w8a8f8推理测试正常 See merge request: Ascend/MindIE-SD!407 | 30 天前 |