| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
catlass英文合入 Co-authored-by: y60124828<yanglou1@h-partners.com> # message auto-generated for no-merge-commit merge: !1024 merge master into master catlass英文合入 Created-by: yanglu-1 Commit-by: y60124828 Merged-by: cann-robot Description: ## 描述 catlass英文合入 ## 关联的Issue https://gitcode.com/cann/catlass/issues/398 ## 原因 <!--说明此次改动的目的、解决的问题等，应与类型标签匹配 --> ## 测试 <!--描述进行了哪些测试来验证你的改动。包括但不限于二级冒烟、算子泛化等。--> ## 文档更新 <!--如果这个PR包含文档的更新，请在这里指出。例如：更新了README.md文件。--> ## 类型标签 <!-- [x] 表示选中 --> - [ ] Bug修复 - [ ] 新特性 - [ ] 性能优化 - [ ] 文档更新 - [ ] 其他，请描述： See merge request: cann/catlass!1024 | 1 个月前 | |
docs: 修正 dispatch_policies.md 中 ENABLE_UNIT_FLAG 宏名拼写 Co-authored-by: yuantao_<yuantao313@outlook.com> # message auto-generated for no-merge-commit merge: !1090 merge fix/doc-unit-flag into master docs: 修正 dispatch_policies.md 中 ENABLE_UNIT_FLAG 宏名拼写 Created-by: yuantao_ Commit-by: yuantao_ Merged-by: cann-robot Description: ### 问题描述 修复 issue #367： docs/zh/2_Design/01_kernel_design/03_dispatch_policies.md 及英文版中宏名误写为 ENABLE_UINT_FLAG（描述文本为 uintflag），实际代码 include/catlass/gemm/dispatch_policy.hpp 中为 ENABLE_UNIT_FLAG / unitflag。 ### 修改内容 - docs/zh/2_Design/01_kernel_design/03_dispatch_policies.md：ENABLE_UINT_FLAG → ENABLE_UNIT_FLAG（8 处）、uintflag → unitflag（4 处） - docs/en/2_Design/01_kernel_design/03_dispatch_policies.md：同上（8 + 4 处） Fixes #367 See merge request: cann/catlass!1090 | 26 天前 | |
feat: Add template definitions for JD.com's core retail operations. Co-authored-by: ext.wangguangcai1<ext.wangguangcai1@jd.com> # message auto-generated for no-merge-commit merge: !1161 merge master into master feat: Add template definitions for JD.com's core retail operations. Created-by: ware2009 Commit-by: ext.wangguangcai1 Merged-by: cann-robot Description: ## 描述 <!--在这里详细描述你的改动，包括改动的原因和所采取的方法。--> 增加京东使用的一些自定义模板类型 ## 关联的Issue <!-- 如果这个PR是为了解决特定的Issue，请在这里提供Issue链接。--> ## 原因 <!--说明此次改动的目的、解决的问题等，应与类型标签匹配 --> 1.主要增加x_flash_attention_infer模板和x_attention模板，支撑推荐推理和传统推荐。 ## 测试 <!--描述进行了哪些测试来验证你的改动。包括但不限于二级冒烟、算子泛化等。--> 使用实例： https://github.com/xLLM-AI/xllm-ops/tree/main/xllm_ops/x_attention https://github.com/xLLM-AI/xllm-ops/tree/main/xllm_ops/x_flash_attention_infer 测试用例运行情况：[root@A03-R40-I191-99-4100006 python_test]# pytest test_x_attention_with_pa.py ============================================================================================================================================================== test session starts ============================================================================================================================================================== platform linux -- Python 3.11.15, pytest-8.3.2, pluggy-1.6.0 rootdir: /export/home/ext.wangguangcai1/code/github/xllm-ops/test/python_test plugins: forked-1.6.0, xdist-3.6.1, anyio-4.14.2 collected 1920 items test_x_attention_with_pa.py ............................................................................................................................................................................................................................................................................................................. [ 15%] ......................................................................................................................................................................................................................................................................................................................................... [ 32%] ......................................................................................................................................................................................................................................................................................................................................... [ 49%] ......................................................................................................................................................................................................................................................................................................................................... [ 67%] ......................................................................................................................................................................................................................................................................................................................................... [ 84%] ............................................................................................................................................................................................................................................................................................................... [100%] ======================================================================================================================================================= 1920 passed in 1516.38s (0:25:16) ======================================================================================================================================================= [root@A03-R40-I191-99-4100006 python_test]# ls RegisterOps.cpp pytorch_npu_helper.hpp test_causal_conv1d_qkv.py test_gamma_add_rms_norm.py test_layer_norm_fwd.py test_moe_gating_top_k_hash.py test_multi_latent_attention.py test_replace_token.py test_x_attention.py pycache setup.py test_causal_conv1d_update.py test_hc_post.py test_lightning_indexer_quant.py test_moe_grouped_matmul.py test_onerec_final_beam_select.py test_reshape_and_cache_a5.py test_x_attention_with_pa.py build test_add_rms_norm_bias.py test_compressor.py test_hc_pre_inv_rms.py test_lightning_indexer_quant_metadata.py test_moe_grouped_matmul_swiglu_quant.py test_pp_matmul_opt.py test_rms_norm_dynamic_quant.py test_x_flash_attention_infer.py build_and_run.sh test_beam_search.py test_convert_kv_cache_format.py test_hc_pre_sinkhorn.py test_mega_chunk_gdn.py test_moe_grouped_matmul_swiglu_quant_v2.py test_quant_lightning_indexer.py test_scatter_nd_update_v2.py custom_ops.egg-info test_beam_search_group.py test_dequant_swiglu_quant.py test_index_group_matmul.py test_moe_fused_add_topk.py test_moe_init_routing_custom.py test_quant_lightning_indexer_metadata.py test_select_unshared_kv.py custom_ops.py test_cache_unshared_kv.py test_dispatch_ffn_combine.py test_inplace_partial_rotary_mul.py test_moe_fused_reducesum_div.py test_moe_init_routing_v3.py test_rec_constrained_topk.py test_sparse_attn_sharedkv.py dist test_causal_conv1d.py test_dispatch_gmm_combine_decode.py test_laser_attention.py test_moe_gating_top_k.py test_mtp_prepare_next_draft.py test_recurrent_gated_delta_rule.py test_sparse_attn_sharedkv_metadata.py [root@A03-R40-I191-99-4100006 python_test]# [root@A03-R40-I191-99-4100006 python_test]# [root@A03-R40-I191-99-4100006 python_test]# ls RegisterOps.cpp pytorch_npu_helper.hpp test_causal_conv1d_qkv.py test_gamma_add_rms_norm.py test_layer_norm_fwd.py test_moe_gating_top_k_hash.py test_multi_latent_attention.py test_replace_token.py test_x_attention.py pycache setup.py test_causal_conv1d_update.py test_hc_post.py test_lightning_indexer_quant.py test_moe_grouped_matmul.py test_onerec_final_beam_select.py test_reshape_and_cache_a5.py test_x_attention_with_pa.py build test_add_rms_norm_bias.py test_compressor.py test_hc_pre_inv_rms.py test_lightning_indexer_quant_metadata.py test_moe_grouped_matmul_swiglu_quant.py test_pp_matmul_opt.py test_rms_norm_dynamic_quant.py test_x_flash_attention_infer.py build_and_run.sh test_beam_search.py test_convert_kv_cache_format.py test_hc_pre_sinkhorn.py test_mega_chunk_gdn.py test_moe_grouped_matmul_swiglu_quant_v2.py test_quant_lightning_indexer.py test_scatter_nd_update_v2.py custom_ops.egg-info test_beam_search_group.py test_dequant_swiglu_quant.py test_index_group_matmul.py test_moe_fused_add_topk.py test_moe_init_routing_custom.py test_quant_lightning_indexer_metadata.py test_select_unshared_kv.py custom_ops.py test_cache_unshared_kv.py test_dispatch_ffn_combine.py test_inplace_partial_rotary_mul.py test_moe_fused_reducesum_div.py test_moe_init_routing_v3.py test_rec_constrained_topk.py test_sparse_attn_sharedkv.py dist test_causal_conv1d.py test_dispatch_gmm_combine_decode.py test_laser_attention.py test_moe_gating_top_k.py test_mtp_prepare_next_draft.py test_recurrent_gated_delta_rule.py test_sparse_attn_sharedkv_metadata.py [root@A03-R40-I191-99-4100006 python_test]# pytest test_x_flash_attention_infer.py ============================================================================================================================================================== test session starts ============================================================================================================================================================== platform linux -- Python 3.11.15, pytest-8.3.2, pluggy-1.6.0 rootdir: /export/home/ext.wangguangcai1/code/github/xllm-ops/test/python_test plugins: forked-1.6.0, xdist-3.6.1, anyio-4.14.2 collected 3 items test_x_flash_attention_infer.py ... [100%] =============================================================================================================================================================== 3 passed in 6.89s =============================================================================================================================================================== [root@A03-R40-I191-99-4100006 python_test]# [root@A03-R40-I191-99-4100006 python_test]# [root@A03-R40-I191-99-4100006 python_test]# pytest test_x_attention.py ============================================================================================================================================================== test session starts ============================================================================================================================================================== platform linux -- Python 3.11.15, pytest-8.3.2, pluggy-1.6.0 rootdir: /export/home/ext.wangguangcai1/code/github/xllm-ops/test/python_test plugins: forked-1.6.0, xdist-3.6.1, anyio-4.14.2 collected 18 items test_x_attention.py .................. [100%] ========================================================================================================================================================= 18 passed in 69.31s (0:01:09) ========================================================================================================================================================= [root@A03-R40-I191-99-4100006 python_test]# ls ## 文档更新 <!--如果这个PR包含文档的更新，请在这里指出。例如：更新了README.md文件。--> ## 类型标签 <!-- [x] 表示选中 --> - [ ] Bug修复 - [x] 新特性 - [ ] 性能优化 - [ ] 文档更新 - [ ] 其他，请描述： See merge request: cann/catlass!1161 | 13 天前 |