Repository navigation

Avoid repeated CUDA device-property queries during convolution plan rebuilds - #33776

Draft

Tianlei Wu (tianleiwu) with Copilot wants to merge 6 commits into

mainfrom

copilot/performance-fix-cuda-conv

Draft

Avoid repeated CUDA device-property queries during convolution plan rebuilds#33776
Tianlei Wu (tianleiwu) with Copilot wants to merge 6 commits into
mainfrom
copilot/performance-fix-cuda-conv

Commits on Oct 30, 2026
- Initial plan
  Copilot
  authored
  62b0c96
- Reuse cached CUDA SM version when rebuilding convolution frontend graphs
  
  Copilot
  and
  Tianlei Wu (tianleiwu)
  authored
  01ffa24
- Pass cached CUDA properties to generation helpers and avoid full occupancy queries
  
  Copilot
  and
  Tianlei Wu (tianleiwu)
  authored
  d328bed
- Fix dynamic CUDA convolution test output-name argument
  
  Copilot
  and
  Tianlei Wu (tianleiwu)
  authored
  c0e505b
Commits on Nov 10, 2026
- Merge origin/main into copilot/performance-fix-cuda-conv
  
  Tianlei Wu (tianleiwu)
  and
  Copilot
  committed
  0d2f60c
- test(cuda): drop ConvTranspose fused-bias case
  
  Tianlei Wu (tianleiwu)
  and
  Copilot
  committed
  9c12f94