[GH-ISSUE #4722] Slower performance on Arm64 with Phi3 and Lexi-Llama on 1.39 #2977

Open
opened 2026-04-12 13:21:29 -05:00 by GiteaMirror · 1 comment
Owner

Originally created by @ghost on GitHub (May 30, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4722

What is the issue?

I am getting 5x slower speeds with the new binary/version of ollama-linux-arm64 1.39 compared to the 1.38 release, and/or earlier.

Not sure what's going on. Am using Orange Pi 5 Plus.

Like I said - 1.38 works flawlessly, even on my Raspberry Pi 5.

OS

Linux

GPU

Other

CPU

Other

Ollama version

1.39

Originally created by @ghost on GitHub (May 30, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4722 ### What is the issue? I am getting 5x slower speeds with the new binary/version of ollama-linux-arm64 1.39 compared to the 1.38 release, and/or earlier. Not sure what's going on. Am using Orange Pi 5 Plus. Like I said - 1.38 works flawlessly, even on my Raspberry Pi 5. ### OS Linux ### GPU Other ### CPU Other ### Ollama version 1.39
GiteaMirror added the performancebug labels 2026-04-12 13:21:29 -05:00
Author
Owner

@glenamac commented on GitHub (May 31, 2024):

I'm experiencing issue slow startup with arm64 but once the model is loaded it runs fine. Is that your experience also? Do you have logging being saved anywhere? What does the line that reads
msg="llama runner started
say?

<!-- gh-comment-id:2142764965 --> @glenamac commented on GitHub (May 31, 2024): I'm experiencing [issue](https://github.com/ollama/ollama/issues/4705) slow startup with arm64 but once the model is loaded it runs fine. Is that your experience also? Do you have logging being saved anywhere? What does the line that reads `msg="llama runner started` say?
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#2977