ROCm / ROCm/AMDMIGraphX

Performance metrics available in C++

Open
#1,397 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C++
Stars
333
Forks
150
Avg merge
4d 19h
Merged PRs (30d)
54

Description

It is desirable to be able to return the execution time per layer in a structure, populated by "run".

The structure should ideally reference each node in the graph, or a group of nodes, demarked by some markers.

A possible use of this feature would be in the pruner: The pruner can ask migraphx what the attained performance is, without overcommitting compute/training resources in the particular pruned dimensions, and only do so if the performance improves.

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 by tracing the C++ implementation of "run" and the pruner integration mentioned in the issue. Determine how nodes or node groups could be associated with execution timing and what structure the caller should receive. Done means "run" returns per-layer or marked-group performance metrics that the pruner can inspect.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
performance
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.