github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #1692] Mac OS Sonoma crashes completely when loading LLM #953

New Issue
Closed
opened 2026-04-12 10:39:12 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-04-12 10:39:12 -05:00
Owner
Copy Link

Originally created by @sanctimon on GitHub (Dec 24, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1692

I have pulled the model (dolphin-mixtral:latest) and when I attempt to run, the entire machine freezes. A few minutes later it restarts.

Specs: MacBook Pro M1 Pro. 16GB RAM.

With Activity Monitor on, it seems to be filling up the RAM quite quickly before the crash.

Originally created by @sanctimon on GitHub (Dec 24, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1692 I have pulled the model (dolphin-mixtral:latest) and when I attempt to run, the entire machine freezes. A few minutes later it restarts. Specs: MacBook Pro M1 Pro. 16GB RAM. With Activity Monitor on, it seems to be filling up the RAM quite quickly before the crash.
GiteaMirror added the bug label 2026-04-12 10:39:12 -05:00
GiteaMirror commented 2026-04-12 10:39:13 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Dec 24, 2023):

Hi @sanctimon
When you enter
Ollama --version
What version number shows up?
If it's 0.0.0 thad is because you installed Ollama with brew install Ollama
And brew script is not consistent and install an older version.
Download the app from ollama.ai home page and retry.

<!-- gh-comment-id:1868470217 --> @igorschlum commented on GitHub (Dec 24, 2023): Hi @sanctimon When you enter Ollama --version What version number shows up? If it's 0.0.0 thad is because you installed Ollama with brew install Ollama And brew script is not consistent and install an older version. Download the app from ollama.ai home page and retry.
GiteaMirror commented 2026-04-12 10:39:13 -05:00
Author
Owner
Copy Link

@BruceMacD commented on GitHub (Dec 24, 2023):

Hi @sanctimon, Ollama shouldn't crash your computer (this part is a bug), but in this case you don't have enough RAM to run dolphin-mixtral, mixtral based models require a good amount of memory. In this case probably like ~30GB at a guess.

<!-- gh-comment-id:1868528708 --> @BruceMacD commented on GitHub (Dec 24, 2023): Hi @sanctimon, Ollama shouldn't crash your computer (this part is a bug), but in this case you don't have enough RAM to run `dolphin-mixtral`, mixtral based models require a good amount of memory. In this case probably like ~30GB at a guess.
GiteaMirror commented 2026-04-12 10:39:13 -05:00
Author
Owner
Copy Link

@sanctimon commented on GitHub (Dec 24, 2023):

I see, thank you both.

  1. I agree that Ollama should not cause the equivalent of BSOD in Macs; it should simply throw an error that memory to run a model is insufficient.
  2. Is there no way to use cache or swap for these cases, particularly when the SSD in these new MacBook Pros is so frightfully fast?
<!-- gh-comment-id:1868606885 --> @sanctimon commented on GitHub (Dec 24, 2023): I see, thank you both. 1. I agree that Ollama should not cause the equivalent of BSOD in Macs; it should simply throw an error that memory to run a model is insufficient. 2. Is there no way to use cache or swap for these cases, particularly when the SSD in these new MacBook Pros is so frightfully fast?
GiteaMirror commented 2026-04-12 10:39:14 -05:00
Author
Owner
Copy Link

@BruceMacD commented on GitHub (Dec 29, 2023):

@sanctimon Good question, running a model should currently do some memory mapping which will use your storage (SSD). However for token inference the whole model needs to pass through the CPU/GPU, given that each tensor in the model is involved in the inference process for every token.

<!-- gh-comment-id:1872100490 --> @BruceMacD commented on GitHub (Dec 29, 2023): @sanctimon Good question, running a model should currently do some memory mapping which will use your storage (SSD). However for token inference the whole model needs to pass through the CPU/GPU, given that each tensor in the model is involved in the inference process for every token.
GiteaMirror commented 2026-04-12 10:39:14 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Jan 9, 2024):

@sanctimon did you tried with 0.1.18? Memory handling is a little different and it could tell you if you don't have enough memory.

<!-- gh-comment-id:1883309431 --> @igorschlum commented on GitHub (Jan 9, 2024): @sanctimon did you tried with 0.1.18? Memory handling is a little different and it could tell you if you don't have enough memory.
GiteaMirror commented 2026-04-12 10:39:14 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Jan 25, 2024):

@sanctimon Sorry that you saw this. There have been a bunch of fixes here for improving memory management, and I think it should be working now. Can you try again w/ 0.1.20 (or 0.1.21 when it comes out)?

I'll go ahead and close the issue for now.

<!-- gh-comment-id:1911122905 --> @pdevine commented on GitHub (Jan 25, 2024): @sanctimon Sorry that you saw this. There have been a bunch of fixes here for improving memory management, and I *think* it should be working now. Can you try again w/ 0.1.20 (or 0.1.21 when it comes out)? I'll go ahead and close the issue for now.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#953
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?