Skip to content

[2026春季][T1-1-1] HyosungSink#1164

Draft
HyosungSink wants to merge 8 commits into
InfiniTensor:mainfrom
HyosungSink:2026-spring-HyosungSink-T1-1-1
Draft

[2026春季][T1-1-1] HyosungSink#1164
HyosungSink wants to merge 8 commits into
InfiniTensor:mainfrom
HyosungSink:2026-spring-HyosungSink-T1-1-1

Conversation

@HyosungSink
Copy link
Copy Markdown

@HyosungSink HyosungSink commented May 18, 2026

Summary

  • Connect T1-1-1 operators to the InfiniCore Python use_ntops path: rad2deg, copysign, lcm, lgamma, nextafter.
  • Add Python-level NVIDIA tests that verify use_ntops dispatch and PyTorch reference parity.
  • Merge the Iluvatar validation build adjustment into this PR branch.

Validation

  • NVIDIA L4/sm_89: source ../scripts/dev-env.sh && python test/infinicore/ops/rad2deg.py --nvidia -> passed
  • NVIDIA L4/sm_89: source ../scripts/dev-env.sh && python test/infinicore/ops/copysign.py --nvidia -> passed
  • NVIDIA L4/sm_89: source ../scripts/dev-env.sh && python test/infinicore/ops/lcm.py --nvidia -> passed
  • NVIDIA L4/sm_89: source ../scripts/dev-env.sh && python test/infinicore/ops/lgamma.py --nvidia -> passed
  • NVIDIA L4/sm_89: source ../scripts/dev-env.sh && python test/infinicore/ops/nextafter.py --nvidia -> passed

Notes

  • This PR only adds the Python use_ntops path; it does not add InfiniOP native, pybind, or C++ bridge implementations.
  • Screenshots for the validation commands will be added before marking ready for review.

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

Successfully merging this pull request may close these issues.

2 participants