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

Minor memory-related fixes and improvements: pretty-print bytes, #221

Merged
merged 6 commits into from
Apr 1, 2024

Conversation

vasdommes
Copy link
Collaborator

  • Pretty-print bytes, e.g. maxSharedMemory = 25.0 GB (26843545600 bytes)
  • Fix RAM estimates printed in pmp2sdp
  • Fix unnecessary matrix copying in compute_block_residues().
    In theory, this should reduce peak RAM usage at this step, but in practice it (almost) doesn't.
  • Some code cleanup

…oth GB and bytes, for better readability.

Also remove extra newlines coming from El::Print

Add pretty_print_bytes() function to sdpb_util/ostream
e.g:
maxSharedMemory              = 25.0 GB (26843545600 bytes)
It was used to print shared window sizes, but now we print them in BigInt_Shared_Memory_Syrk_Context constructor.
- Remove outdated comment about copies of Q
- Fix BigFloat size calculation: use actual size instead of serialized. Actual size is 12B larger.
- Accounted for more PSD blocks: primal_residues, dX, dY, R, Z
- pretty print bytes

R,Z - from compute_search_direction(), TODO we shall account for them only if peak usage is there.
…ix allocation + deallocate_unused_half()) + code cleanup + reformat
block_views should contain views of input block matrices, but extra copy was made by mistake.
@vasdommes vasdommes added this to the 3.0.0 milestone Apr 1, 2024
@vasdommes vasdommes self-assigned this Apr 1, 2024
@vasdommes vasdommes merged commit 24f0117 into master Apr 1, 2024
2 checks passed
@vasdommes vasdommes deleted the minor-memory-fixes branch April 1, 2024 22:41
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant