Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Roadmap #34

Open
18 of 32 tasks
FFFrog opened this issue Jul 30, 2024 · 0 comments
Open
18 of 32 tasks

Roadmap #34

FFFrog opened this issue Jul 30, 2024 · 0 comments
Assignees

Comments

@FFFrog
Copy link
Collaborator

FFFrog commented Jul 30, 2024

  • [元昊] autoload torch.npu & torch_npu.npu
  • [元昊] NPUGuardImpl.cpp 内部注册拆分 Refactor register #47
  • [元昊] DeviceProp 下移到 csrc/npu下(可以参考XPU实现)
  • [元昊] AclInterface.h 删除 Remove AclInterface.h/.cpp #46
  • [元昊] 新增 torch_backend._C 的 pyi文件(代码提示)feat: add torch_backend/_C/__init__.pyi #79
  • [逢春] StreamGuard.h 按需重构
  • [逢春] HasCompatibleShallowCopyType.cpp Warning定位(这个可以改成设备无关,不通过dispatcher调用)
  • [逢春] Refactor AutocastMode.cpp(官方提供了默认列表,对比差异,如果一样可以直接调用宏)
  • [逢春] npu init & finalize 重构
  • [泽升] torch_npu/optim 删除
  • [泽升] testing删除以及新的test framework
  • [泽升] namespace 设计
  • [贞斌] NPUPluggableAllocator.h删除
  • [贞斌] torch_npu/csrc/Module.cpp 单独拆出来内存相关
  • [贞斌] NPU LOG重新设计,ERROR打印错误栈
  • [佳伟] aten 重构
  • [佳伟] torch_npu/utils refactor
  • [佳伟] npu+op-plugin 新建仓库
  • [佳伟] NPUGuard.h 移到 csrc/npu
  • [贞斌] autocast 设备相关api,兼容社区最新api
  • [待定] deviceguard api完备度缺失,后续考虑对标CUDA(torch.Stream("npu:0"))
  • [泽升] REAMDE.md
  • [泽升] LICENSE update
  • [泽升] torch_npu/utils 无用代码删除
  • [ALL] 文档
  • [贞斌] torch_npu/csrc/Module.cpp内容未拆分完全
  • [待定] torch_npu/npu/utils.py内容未拆分
  • [逢春] maybe_initialize_npu删除,替换成device_lazy_init
  • [待定] torch_npu 移除npu前缀
  • [待定] _npu_is_bf16_supported 新增
  • [待定] 补齐 CachingAllocator NPUCachingAllocator NPUCachingHostAllocator NPUCachingAllocatorHelper 宏 PR
  • [待定] 新增 提供device name的API
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants