CPU/RAM Offload CPU+GPU inference
Open
Nobody has claimed this yet.
area/backends
enhancement
- Dominant language
- Go
- Stars
- 49.2k
- Forks
- 4.5k
- Avg merge
- 1d 3m
- Merged PRs (30d)
- 239
Description
Is your feature request related to a problem? Please describe.
Problem - no way to use all available memory VRAM + RAM
Describe the solution you'd like
Hi, as far as i know you either have GPU or CPU support for models, is there a plan to add mixed inference? It will be great for MoE models for example.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
The issue does not identify any files, tests, or entry points. Start by locating the model inference paths and determining how CPU RAM and GPU VRAM are currently selected; done means mixed CPU+GPU inference can use both memory pools, including for MoE models.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- go
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100