NVIDIA / NVIDIA/AMGX

Memory requirements not calculated correctly

Open
#182 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Cuda
Stars
692
Forks
197
PR merge metrics
No merged PRs in 30d

Description

Issue #181 explains the mechanism behind AMG_Level::A and AMG_Level::Aoriginal. In a nutshell, they either point to the same matrix or A points to one matrix and Aoriginal is NULL.
When calculating the memory requirement for a level https://github.com/NVIDIA/AMGX/blob/32e1f44fa93af7859490a800f137e75b6513420c/base/include/amg_level.h#L194 the size of both A and Aoriginal are taken into account separately, resulting in double counting when Aoriginal != NULL.
We should either not account Aoriginal at all, or add a check to make sure Aoriginal != A (which is current cure is never the case unless A == NULL)

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start in base/include/amg_level.h at the memory requirement calculation around line 194. Read how AMG_Level::A and AMG_Level::Aoriginal are represented and compare their sizes when they refer to the same matrix. Done means the memory requirement no longer double-counts aliased matrices; no test file is mentioned in the issue.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
hpc
Issue type
Bug
Difficulty
2/5
Estimated time
1-3 hours
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
58/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.