[GH-ISSUE #1200] CPU instead of GPU for Q5_1 models #47123

Closed
opened 2026-04-28 03:19:40 -05:00 by GiteaMirror · 1 comment
Owner

Originally created by @ivanfioravanti on GitHub (Nov 19, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1200

Originally assigned to: @mxyng on GitHub.

Using alfred q5_1 model uses CPU instead of GPU, latest is using GPU properly

Tested on Apple M2 Ultra (cores: 8E+16P+76GPU) 192GB RAM using asitop

Here commands and attached logs:
ollama run alfred:40b-1023-q5_1 "give me a list of document with {city:city name, country:country name} at least 3 use json format"

ollama run latest "give me a list of document with {city:city name, country:country name} at least 3 use json format"

serverCPU.log
serverGPU.log

Originally created by @ivanfioravanti on GitHub (Nov 19, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1200 Originally assigned to: @mxyng on GitHub. Using alfred q5_1 model uses CPU instead of GPU, latest is using GPU properly Tested on Apple M2 Ultra (cores: 8E+16P+76GPU) 192GB RAM using asitop Here commands and attached logs: `ollama run alfred:40b-1023-q5_1 "give me a list of document with {city:city name, country:country name} at least 3 use json format"` `ollama run latest "give me a list of document with {city:city name, country:country name} at least 3 use json format"` [serverCPU.log](https://github.com/jmorganca/ollama/files/13404438/serverCPU.log) [serverGPU.log](https://github.com/jmorganca/ollama/files/13404439/serverGPU.log)
GiteaMirror added the bug label 2026-04-28 03:19:40 -05:00
Author
Owner

@ivanfioravanti commented on GitHub (Nov 20, 2023):

Thanks @mxyng super fast fix!

<!-- gh-comment-id:1819878335 --> @ivanfioravanti commented on GitHub (Nov 20, 2023): Thanks @mxyng super fast fix!
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#47123