Skip to content

v1.1.0

Choose a tag to compare

View all tags
@pgumienny-nvda pgumienny-nvda released this 22 May 20:15
· 7 commits to main since this release
v1.1.0
ff08fe0
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Added

  • Heatmap plotter
  • Support for CUDA Error Log Management
  • Retry mechanism for CUDA multicast allocations
  • Nvloom_cli argument to set number of samples in gpu-to-rack testcases
  • Nvloom_cli now prints its version, git commit it was built from, and specified buffer size
  • Nvloom_cli now prints units when reporting results
  • Native compilation for sm_103 on CUDA 12.9 toolkits

Changed

  • Expanded README.md
  • Rack-to-rack are now both unidir and bidir, and bidir rack-to-rack are symmetry-optimized.

Fixed

  • Bug where requesting allocations over 4 GiB would fail with CUDA_OUT_OF_MEMORY
Assets 2
Loading