NVIDIA / NVIDIA/AMGX

Thoughts on supporting lower precision floats

Open
#331 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

Given the recent work on lower precision floating point being done by the PyTorch community, (See here: https://pytorch.org/blog/training-using-float8-fsdp2/?utm_content=317436495&utm_medium=social&utm_source=linkedin&hss_channel=lcp-78618366) has there been any thoughts on if AMGX will follow suite and support floats with less precision than 32 and 64?

As someone less familiar with the internals of AMGX, and more with the community-driven interfaces, where might one start in the code to make this addition?

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

Begin with AMGX's community-facing interfaces and current 32- and 64-bit float support, as the issue provides no file or test entry point. Done would require a concrete decision and implementation scope for lower-precision float support; no acceptance test is specified.

Written by the indexing model from the issue text.

Assessment

Tech stack
pytorch
Domain
hpc
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.