github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #5923] Slow Model Loading Speed on macOS System #3697

New Issue
Open
opened 2026-04-12 14:30:45 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-12 14:30:45 -05:00
Owner
Copy Link

Originally created by @ghost on GitHub (Jul 24, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/5923

What is the issue?

I am experiencing slow model loading speeds when using Ollama on my macOS system. Here are the specifications of my setup:

macOS Version: 14.5
Processor: M3 Max
Memory: 128GB
Storage: 2TB (with performance on par with the 8TB version)
Ollama version: 0.2.8

Despite having sufficient hardware capabilities, the model loading speed typically hovers around 700MB/s. During the loading process, I do not observe any component (CPU, Disk, Memory, GPU) being fully utilized or experiencing high usage. Could you please help me understand the reason for this bottleneck and suggest any potential solutions or optimizations?

Thank you for your assistance.

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.2.8

Originally created by @ghost on GitHub (Jul 24, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/5923 ### What is the issue? I am experiencing slow model loading speeds when using Ollama on my macOS system. Here are the specifications of my setup: macOS Version: 14.5 Processor: M3 Max Memory: 128GB Storage: 2TB (with performance on par with the 8TB version) Ollama version: 0.2.8 Despite having sufficient hardware capabilities, the model loading speed typically hovers around 700MB/s. During the loading process, I do not observe any component (CPU, Disk, Memory, GPU) being fully utilized or experiencing high usage. Could you please help me understand the reason for this bottleneck and suggest any potential solutions or optimizations? Thank you for your assistance. ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.2.8
GiteaMirror added the bug label 2026-04-12 14:30:45 -05:00
GiteaMirror commented 2026-04-12 14:30:45 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Jul 24, 2024):

Hi @Yuhuadi,

I have a MacStation with 192 GB of RAM. When I run llama3:70b in the terminal and check the Activity Monitor app, the average data transfer rate is 920 MB/s.

The speed of 920 MB/s is quite good for an internal SSD on macOS. It indicates that your system is efficiently handling the data transfer requirements for running llama3:70b, which should contribute to smooth performance.

<!-- gh-comment-id:2248990343 --> @igorschlum commented on GitHub (Jul 24, 2024): Hi @Yuhuadi, I have a MacStation with 192 GB of RAM. When I run llama3:70b in the terminal and check the Activity Monitor app, the average data transfer rate is 920 MB/s. The speed of 920 MB/s is quite good for an internal SSD on macOS. It indicates that your system is efficiently handling the data transfer requirements for running llama3:70b, which should contribute to smooth performance.
GiteaMirror commented 2026-04-12 14:30:46 -05:00
Author
Owner
Copy Link

@ghost commented on GitHub (Jul 28, 2024):

Hi @Yuhuadi,

I have a MacStation with 192 GB of RAM. When I run llama3:70b in the terminal and check the Activity Monitor app, the average data transfer rate is 920 MB/s.

The speed of 920 MB/s is quite good for an internal SSD on macOS. It indicates that your system is efficiently handling the data transfer requirements for running llama3:70b, which should contribute to smooth performance.

As far as I know, the bandwidth of Mac's hard drives far exceeds this number. Even when connected externally via USB-C, speeds can reach several GB/s.

<!-- gh-comment-id:2254345485 --> @ghost commented on GitHub (Jul 28, 2024): > Hi @Yuhuadi, > > I have a MacStation with 192 GB of RAM. When I run llama3:70b in the terminal and check the Activity Monitor app, the average data transfer rate is 920 MB/s. > > The speed of 920 MB/s is quite good for an internal SSD on macOS. It indicates that your system is efficiently handling the data transfer requirements for running llama3:70b, which should contribute to smooth performance. As far as I know, the bandwidth of Mac's hard drives far exceeds this number. Even when connected externally via USB-C, speeds can reach several GB/s.
GiteaMirror commented 2026-04-12 14:30:46 -05:00
Author
Owner
Copy Link

@igorschlum commented on GitHub (Jul 28, 2024):

@Yuhuadi Ollama not only read the file from the HardDrive, the data have to be transfered to the GPU and do several tasks. Could be great if it's faster, but limitation can be on llama.cpp
It shloud be interesting to test other app on mac that can run LLM and see if they load LLM faster that Ollama and then improve Ollama.

<!-- gh-comment-id:2254461551 --> @igorschlum commented on GitHub (Jul 28, 2024): @Yuhuadi Ollama not only read the file from the HardDrive, the data have to be transfered to the GPU and do several tasks. Could be great if it's faster, but limitation can be on llama.cpp It shloud be interesting to test other app on mac that can run LLM and see if they load LLM faster that Ollama and then improve Ollama.
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#3697
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?