github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #5559] ollama 0.2.0 version is slower than 0.1.48 #3475

New Issue
Closed
opened 2026-04-12 14:09:54 -05:00 by GiteaMirror · 5 comments
GiteaMirror commented 2026-04-12 14:09:54 -05:00
Owner
Copy Link

Originally created by @ceykmc on GitHub (Jul 9, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/5559

Originally assigned to: @jmorganca on GitHub.

What is the issue?

I upgrade ollama from 0.1.48 to 0.2.0 in ubuntu20.04.
With all configures unchanged, using deepseek-coder-v2:16b, RTX4090D
in 0.1.48, inference speed is 0.4s
but in 0.2.0, inference speed is 4s.
I don't know why.

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.2.0

Originally created by @ceykmc on GitHub (Jul 9, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/5559 Originally assigned to: @jmorganca on GitHub. ### What is the issue? I upgrade ollama from 0.1.48 to 0.2.0 in ubuntu20.04. With all configures unchanged, using deepseek-coder-v2:16b, RTX4090D in 0.1.48, inference speed is 0.4s but in 0.2.0, inference speed is 4s. I don't know why. ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.2.0
GiteaMirror added the bug label 2026-04-12 14:09:54 -05:00
GiteaMirror commented 2026-04-12 14:09:56 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Jul 9, 2024):

Hi @ceykmc do you happen to have the logs available? journalctl -u ollama

<!-- gh-comment-id:2216486564 --> @jmorganca commented on GitHub (Jul 9, 2024): Hi @ceykmc do you happen to have the logs available? `journalctl -u ollama`
GiteaMirror commented 2026-04-12 14:09:57 -05:00
Author
Owner
Copy Link

@ceykmc commented on GitHub (Jul 9, 2024):

when use 0.2.0
if set OLLAMA_NUM_PARALLEL=1 which is default value, everything is fine.
when change OLLAMA_NUM_PARALLEL large than 1, e.g. OLLAMA_NUM_PARALLEL=4, every inference is slow, just as first load time.

in my situation, when set OLLAMA_NUM_PARALLEL=1, the first inference time is 4s (which is fine, because the model need to be load), then the inference time is 0.4s.
when set OLLAMA_NUM_PARALLEL=4, the first inference time is 4s , the rest inference time is still 4s.
maybe if OLLAMA_NUM_PARALLEL large than 1, the model will be reloaded every time?

<!-- gh-comment-id:2216757026 --> @ceykmc commented on GitHub (Jul 9, 2024): when use 0.2.0 if set OLLAMA_NUM_PARALLEL=1 which is default value, everything is fine. when change OLLAMA_NUM_PARALLEL large than 1, e.g. OLLAMA_NUM_PARALLEL=4, every inference is slow, just as first load time. in my situation, when set OLLAMA_NUM_PARALLEL=1, the first inference time is 4s (which is fine, because the model need to be load), then the inference time is 0.4s. when set OLLAMA_NUM_PARALLEL=4, the first inference time is 4s , the rest inference time is still 4s. maybe if OLLAMA_NUM_PARALLEL large than 1, the model will be reloaded every time?
GiteaMirror commented 2026-04-12 14:09:58 -05:00
Author
Owner
Copy Link

@binarrii commented on GitHub (Jul 9, 2024):

I encountered the same problem, just like @ceykmc mentioned.

<!-- gh-comment-id:2216815972 --> @binarrii commented on GitHub (Jul 9, 2024): I encountered the same problem, just like @ceykmc mentioned.
GiteaMirror commented 2026-04-12 14:09:58 -05:00
Author
Owner
Copy Link

@binarrii commented on GitHub (Jul 9, 2024):

I encountered the same problem, just like @ceykmc mentioned.

It seems that the model needs to be reloaded every time a request is made.

<!-- gh-comment-id:2216820170 --> @binarrii commented on GitHub (Jul 9, 2024): > I encountered the same problem, just like @ceykmc mentioned. It seems that the model needs to be reloaded every time a request is made.
GiteaMirror commented 2026-04-12 14:09:58 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Jul 9, 2024):

Hi folks, this should be fixed in https://github.com/ollama/ollama/releases/tag/v0.2.1 – sorry about the issue. Let me know if you're still seeing slowness in 0.2.1

<!-- gh-comment-id:2216845608 --> @jmorganca commented on GitHub (Jul 9, 2024): Hi folks, this should be fixed in https://github.com/ollama/ollama/releases/tag/v0.2.1 – sorry about the issue. Let me know if you're still seeing slowness in 0.2.1
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#3475
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?