| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 1 年前 | ||
add control of python GC before capture npugraph Co-authored-by: 周锐淇<zhouruiqi5@huawei.com> # message auto-generated for no-merge-commit merge: !27715 merge master into master add control of python GC before capture npugraph Created-by: rich9527 Commit-by: 周锐淇 Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > /kind task **What does this PR do / why do we need it**: add control of python GC before capture npugraph **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!27715 | 8 个月前 | |
add aot inductor for master Co-authored-by: ascend-robot<zhongyuanke@huawei.com> # message auto-generated for no-merge-commit merge: !28570 merge aoti_master into master add aot inductor for master Created-by: kkjocker Commit-by: ascend-robot Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line: > > /kind bug > /kind task > /kind feature **What does this PR do / why do we need it**: 为master新增aoti特性 **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!28570 | 8 个月前 | |
feat: Supports enabling environment variable log Co-authored-by: chenkun<chenkun82@huawei.com> # message auto-generated for no-merge-commit merge: !28110 merge master_feat_environ_log_v2 into master feat: Supports enabling environment variable log Created-by: kuhn7 Commit-by: chenkun Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** feature **What does this PR do / why do we need it**: Environment variable logging can be enabled by setting export TORCH_NPU_LOGS=env to help debug **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!28110 | 8 个月前 | |
[SilentCheck] fix with DTensor of grad Co-authored-by: wangchao430<wangchao430@huawei.com> # message auto-generated for no-merge-commit merge: merge v2.99.0_checkfix1 into master [SilentCheck] fix with DTensor of grad Created-by: wangchao430 Commit-by: wangchao430 Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line: > > /kind bug > /kind task > /kind feature **What does this PR do / why do we need it**: **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!25008 | 11 个月前 | |
Adapt parameter "device_id" for init_process_group Co-authored-by: l30044004<louyujing@huawei.com> # message auto-generated for no-merge-commit merge: !27487 merge master into master Adapt parameter "device_id" for init_process_group Created-by: louyujing Commit-by: l30044004 Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line: > > /kind bug > /kind task > /kind feature **What does this PR do / why do we need it**: **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # torch.distributed.init_process_group可传入参数device_id，如果传入的device_id包含cuda，在调用torch.distributed.barrier()时会获取process_group中包含cuda的device_id，导致在npu上运行报错； 因此，给torch.distributed.init_process_group添加装饰器，将参数device_id中的cuda替换为npu。 **Special notes for your reviewers**: See merge request: Ascend/pytorch!27487 | 8 个月前 | |
add aot inductor for master Co-authored-by: ascend-robot<zhongyuanke@huawei.com> # message auto-generated for no-merge-commit merge: !28570 merge aoti_master into master add aot inductor for master Created-by: kkjocker Commit-by: ascend-robot Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line: > > /kind bug > /kind task > /kind feature **What does this PR do / why do we need it**: 为master新增aoti特性 **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!28570 | 8 个月前 | |
Add register sharding strategy for npu_conv2d npu_grouped_matmul_add_ npu_apply_adam_w Co-authored-by: zhangguoguang<zhangguoguang2@huawei.com> # message auto-generated for no-merge-commit merge: !28586 merge add_register_sharding/2nd_master into master Add register sharding strategy for npu_conv2d npu_grouped_matmul_add_ npu_apply_adam_w Created-by: zhangguoguang Commit-by: zhangguoguang Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line: > > /kind bug > /kind task > /kind feature **What does this PR do / why do we need it**: **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!28586 | 8 个月前 | |
"npugraph_ex" backend supports options Co-authored-by: xuzhenhua2025<xuzhenhua4@huawei.com> # message auto-generated for no-merge-commit merge: !27876 merge master into master "npugraph_ex" backend supports options Created-by: xuzhenhua2025 Commit-by: xuzhenhua2025 Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** /kind feature **What does this PR do / why do we need it**: 1. "npugraph_ex" backend supports options: python model = torch.compile(model.npu(), backend="npugraph_ex", options={ # options: # "key1" = value1, # ... }) 2. "npugraph_ex" backend supports APIs: - torch.npu.npugraph_ex.inference.cache_compile - torch.npu.npugraph_ex.inference.readable_cache - torch.npu.npugraph_ex.ops.npu_print - torch.npu.npugraph_ex.ops.npu_create_tagged_event - torch.npu.npugraph_ex.ops.npu_tagged_event_record - torch.npu.npugraph_ex.ops.npu_tagged_event_wait - torch.npu.npugraph_ex.ops.npu_record_tagged_stream - torch.npu.npugraph_ex.scope.npu_stream_switch - torch.npu.npugraph_ex.scope.npu_wait_tensor - torch.npu.npugraph_ex.scope.limit_core_num **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!27876 | 8 个月前 | |
| 2 年前 | ||
Event supports cross-process and cross-device (IPC event) Co-authored-by: liujunzhu<liujunzhu@huawei.com> # message auto-generated for no-merge-commit merge: !28123 merge master into master Event supports cross-process and cross-device (IPC event) Created-by: liujunzhu Commit-by: liujunzhu Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line: > > /kind bug > /kind task /kind feature **What does this PR do / why do we need it**: event能力对齐cuda，支持跨进程与跨设备使用。 CUDA跨进程共享内存和跨设备拷贝内存时使用Event进行同步，而torch npu使用SynchronizeStream进行同步，希望torch_npu支持在跨设备和跨内存的场景也使用Event进行同步，提升整体性能。另外，需支持通过Python接口在进程间传递Event对象或Event handle并使用该Event进行进程间的同步。 相关场景如下： 1、跨进程使用Event：进程间传递interprocess=True的Event对象、Event的IPC Handle或torch.multiprocessing.reductions.reduce_event(event)的结果。 2、跨进程共享NPU内存：包括进程间通过参数或队列传递Tensor、进程间通过torch.multiprocessing.reductions.reduce_tensor()传递Tensor、进程间通过_share_npu_传递Tensor。 3、跨设备拷贝NPU内存：调用Tensor.to()或Tensor.copy_()接口。 不支持跨设备或跨进程使用interprocess=False的Event。 **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!28123 | 8 个月前 | |
Event supports cross-process and cross-device (IPC event) Co-authored-by: liujunzhu<liujunzhu@huawei.com> # message auto-generated for no-merge-commit merge: !28123 merge master into master Event supports cross-process and cross-device (IPC event) Created-by: liujunzhu Commit-by: liujunzhu Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line: > > /kind bug > /kind task /kind feature **What does this PR do / why do we need it**: event能力对齐cuda，支持跨进程与跨设备使用。 CUDA跨进程共享内存和跨设备拷贝内存时使用Event进行同步，而torch npu使用SynchronizeStream进行同步，希望torch_npu支持在跨设备和跨内存的场景也使用Event进行同步，提升整体性能。另外，需支持通过Python接口在进程间传递Event对象或Event handle并使用该Event进行进程间的同步。 相关场景如下： 1、跨进程使用Event：进程间传递interprocess=True的Event对象、Event的IPC Handle或torch.multiprocessing.reductions.reduce_event(event)的结果。 2、跨进程共享NPU内存：包括进程间通过参数或队列传递Tensor、进程间通过torch.multiprocessing.reductions.reduce_tensor()传递Tensor、进程间通过_share_npu_传递Tensor。 3、跨设备拷贝NPU内存：调用Tensor.to()或Tensor.copy_()接口。 不支持跨设备或跨进程使用interprocess=False的Event。 **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!28123 | 8 个月前 | |
| 1 年前 | ||
| 2 年前 | ||
memory optimaze Co-authored-by: tangmengcheng<tangmengcheng2@huawei.com> # message auto-generated for no-merge-commit merge: !27409 merge master into master memory optimaze Created-by: hhz0 Commit-by: tangmengcheng Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line: > > /kind bug > /kind task > /kind feature **What does this PR do / why do we need it**: 优化内存占用高的TraceViewParser和TraceStepTimeParser，两者在同一时间读取设备侧的json文件导致解析时内存的峰值较高，将TraceViewParser读取设备侧json文件的步骤前移至CANNTimelineParser，避免内存峰值过高导致oom **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!27409 | 8 个月前 | |
register strategies for npu custom ops Co-authored-by: jizewei<jizewei@huawei.com> # message auto-generated for no-merge-commit merge: !27012 merge master_register_strategy into master register strategy for npu custom ops Created-by: jizewei Commit-by: jizewei Merged-by: ascend-robot Description: <!-- Thanks for sending a pull request! --> **What type of PR is this?** > Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line: > > /kind bug > /kind task > /kind feature /kind feature **What does this PR do / why do we need it**: register sharding strategies for npu custom ops to support DTensor: 1. npu_add_rms_norm 2. npu_moe_token_permute, npu_moe_token_permute_grad 3. npu_moe_token_unpermute, npu_moe_token_unpermute_grad 4. npu_rotary_mul, npu_rotary_mul_backward **Which issue(s) this PR fixes**: <!-- *Automatically closes linked issue when PR is merged. Usage: Fixes #<issue number>, or Fixes (paste link of issue). --> Fixes # **Special notes for your reviewers**: See merge request: Ascend/pytorch!27012 | 9 个月前 | |
attention/ffn的aicpu算子提供torch_npu._afd的调用方式 | 8 个月前 | |
修改codecheck | 8 个月前 | |
| 1 年前 | ||
| 11 个月前 |