oneapi-src / oneapi-src/unified-runtime

[L0] Backend context will contain all devices even if frontend context contains single device

Open
#2,509 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

level-zero
Dominant language
C++
Stars
57
Forks
120
Avg merge
1d 14h
Merged PRs (30d)
1

Description

Migrate from https://github.com/intel/llvm/issues/10295

The reason that I bring this up again because with PyTorch usage, we can't strictly assign device to specific process and defend wrongfully submission of kernels, for example, to device that different from allocation residency.

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 reading the migrated LLVM issue at https://github.com/intel/llvm/issues/10295 and tracing how frontend and backend contexts are defined. Done means a frontend context limited to one device does not expose all devices through its backend context, preventing kernels from being submitted to a device different from the allocation residency.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
backend
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.