| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
[Huawei][AscendNPU IR] align ArithToAffine pattern with regbase version Offering: AscendNPU IR | 23 天前 | |
[Huawei][AscendNPU IR] adapt arith.mulext/mulextui conversion in arith-to-hfusion Offering: AscendNPU IR | 20 天前 | |
fix: preserve FMA through Ascend950 SIMD lowering | 3 天前 | |
[Huawei][AscendNPU IR] add pass: convert-arith-to-hivm-llvm from A5 | 1 个月前 | |
[SIMT] Adds support for permute instruction | 6 天前 | |
[Huawei][AscendNPU-IR] fix unknown loc | 17 天前 | |
[SIMT] Adds GPU To DPX conversions | 26 天前 | |
[AscendNPU IR] fix warnings: base specifier with no access specifier warnings in structs. | 8 个月前 | |
[AscendNPU-IR] fix: replace llvm_unreachable with report_fatal_error | 20 天前 | |
!2215 merge cp-remain-simd into master [Huawei][AscendNPU-IR] cherry-pick remaining simd commits Created-by: liupengcheng2012 Commit-by: QQ3646;Nikolay;lijiawei_123;liupengcheng2012;anna-antipina;shijingchang Merged-by: tanshengshun Description: --- title: "[AscendNPU IR]:" --- ## 描述 Description 请提供此 Pull Request 的高级别说明。 Please describe what this PR is about. ## 类型 Category - [ ] Bug修复 Bugfix - [ ] 新功能 New feature - [ ] 代码风格更新（格式化，局部变量）Style changes - [ ] 重构（即不是新增功能，也不是修改bug的代码变动）Refactor - [ ] 构建过程或辅助工具的变动 Build-related - [ ] 文档内容更新 Docs ## Checklist - [ ] 我的代码遵循这个项目的代码风格 My code follows the project's code style - [ ] 我已经自己测试过我的代码 I have self-tested my code - [ ] 我已经更新了相应的文档 I have updated the relevant documentation - [ ] 我已经根据需要更新了对应的变更日志 I have updated the changelog as needed - [ ] 我已经在标题中正确使用了类型标签（例如： feat:, fix:）I have correctly used type labels in the title (e.g., feat:, fix:) See merge request: Ascend/AscendNPU-IR!2215 | 16 天前 | |
[AscendNPU IR][Conversion] chore: port HFusionToVector pass onto master | 1 个月前 | |
[HFusion][HIVMAVE] Fix FP8 scalar argument lowering Carry internal VF FP8 scalar arguments as i8 payloads to avoid emitting unsupported typed FP8 literals at CCEC call boundaries. Lower FP8 scalar broadcasts through i8 vbr and restore the FP8 vector type with a bitcast. | 11 天前 | |
[Huawei][AscendNPU-IR] fix unknown loc | 17 天前 | |
[AscendNPU IR]: allow up to 3D GM to L1 copy on membase Update isCopyToCbuf to match load_gm_to_cbuf_{1,2,3}d template support and add FileCheck coverage for 1D/2D/3D GM to cbuf lowering. Co-authored-by: Cursor <cursoragent@cursor.com> | 6 天前 | |
!2545 merge refactor-memref-1n-conversion into master [AscendNPU-IR][Simd-Simt] convert MemRef via a 1:N descriptor Created-by: fabienzhao Commit-by: fabienzz Merged-by: ascend-robot Description: --- title: "[AscendNPU IR]:" --- ## 描述 Description 请提供此 Pull Request 的高级别说明。 Please describe what this PR is about. ## 类型 Category - [ ] Bug修复 Bugfix - [ ] 新功能 New feature - [ ] 代码风格更新（格式化，局部变量）Style changes - [ ] 重构（即不是新增功能，也不是修改bug的代码变动）Refactor - [ ] 构建过程或辅助工具的变动 Build-related - [ ] 文档内容更新 Docs ## Checklist - [ ] 我的代码遵循这个项目的代码风格 My code follows the project's code style - [ ] 我已经自己测试过我的代码 I have self-tested my code - [ ] 我已经更新了相应的文档 I have updated the relevant documentation - [ ] 我已经根据需要更新了对应的变更日志 I have updated the changelog as needed - [ ] 我已经在标题中正确使用了类型标签（例如： feat:, fix:）I have correctly used type labels in the title (e.g., feat:, fix:) See merge request: Ascend/AscendNPU-IR!2545 | 21 小时前 | |
Revert "!2568 merge fix/fp-bitwise-vv-broadcast into master" This reverts commit f6c77a82669a98186f90f7c783d2624286438b0c, reversing changes made to bb8d2496b6389ccbd25a79cd762635d4f17454f1. | 14 小时前 | |
[Huawei][AscendNPU IR] migrate LowerMemRefExt: SUFFIX SPLIT offset + sub-workspace walk Offset-handling: if (isRegBased) hardcoded size-2 else general for-loop. Sub-workspace walk: regbase-only, erases kSubWorkspace kernel args. | 1 个月前 | |
fix: preserve FMA through Ascend950 SIMD lowering | 3 天前 | |
[SIMT] Add Proton related conversions | 26 天前 | |
[AscendNPU IR] fix warnings: base specifier with no access specifier warnings in structs. | 8 个月前 | |
[Huawei][AscendNPU IR] guard TensorToHIVM InsertSliceSourceIndex by isMemBasedArch membase-only attr-setting, no-op on regbase. | 1 个月前 | |
[Huawei][AscendNPU IR] fix non-virtual destrutor Offering: Bisheng | 8 个月前 | |
[AscendNPU IR] Add HFusion Co-authored-by: chenxiangting<chenxiangting@huawei.com> | 10 个月前 | |
!2333 merge GM-Store-Master into master [Huawei][AscendNPU-IR][SIMT] Add support for using global scratch memory for staging in TileDotLoads and for Convert Layout ops Created-by: SanidhyaBarara Commit-by: Sanidhya Barara Merged-by: ascend-robot Description: --- title: "[AscendNPU IR]:" --- ## 描述 Description 请提供此 Pull Request 的高级别说明。 Add support for using global scratch memory for staging in TileDotLoads and for Convert Layout ops. Also adds a default false option to enable using global scratch. ## 类型 Category - 新功能 New feature - 代码风格更新（格式化，局部变量）Style changes - 重构（即不是新增功能，也不是修改bug的代码变动）Refactor ## Checklist - 我的代码遵循这个项目的代码风格 My code follows the project's code style - 我已经自己测试过我的代码 I have self-tested my code - 我已经更新了相应的文档 I have updated the relevant documentation - 我已经根据需要更新了对应的变更日志 I have updated the changelog as needed - 我已经在标题中正确使用了类型标签（例如： feat:, fix:）I have correctly used type labels in the title (e.g., feat:, fix:) See merge request: Ascend/AscendNPU-IR!2333 | 3 小时前 | |
[Huawei][AscendNPU-IR] fuse was_bool_to_int8 mask loads into plds Signed-off-by: lrb <liruibo6@h-partners.com> | 12 天前 | |
[Huawei][AscendNPU-IR] fix unknown loc | 17 天前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 23 天前 | ||
| 20 天前 | ||
| 3 天前 | ||
| 1 个月前 | ||
| 6 天前 | ||
| 17 天前 | ||
| 26 天前 | ||
| 8 个月前 | ||
| 20 天前 | ||
| 16 天前 | ||
| 1 个月前 | ||
| 11 天前 | ||
| 17 天前 | ||
| 6 天前 | ||
| 21 小时前 | ||
| 14 小时前 | ||
| 1 个月前 | ||
| 3 天前 | ||
| 26 天前 | ||
| 8 个月前 | ||
| 1 个月前 | ||
| 8 个月前 | ||
| 10 个月前 | ||
| 3 小时前 | ||
| 12 天前 | ||
| 17 天前 |