github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #3818] llama3:70b-instruct response time. #2361

New Issue
Closed
opened 2026-04-12 12:40:46 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 12:40:46 -05:00
Owner
Copy Link

Originally created by @Atlanta11 on GitHub (Apr 22, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/3818

Hello,what else can I do to make the AI respond faster because currently everything is working but a bit on the slow side with an Nvidia GeForce RTX 4090 and i9-14900k with 64 GB of RAM. I was able to download the model ollama run llama3:70b-instruct fairly quickly at a speed of 30 MB per second.
Here is my server.log.
server.log

Schermafbeelding 2024-04-22 133915

OS

Windows

GPU

Nvidia

CPU

Intel

Ollama version

llama3:70b-instruct

Originally created by @Atlanta11 on GitHub (Apr 22, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/3818 Hello,what else can I do to make the AI respond faster because currently everything is working but a bit on the slow side with an Nvidia GeForce RTX 4090 and i9-14900k with 64 GB of RAM. I was able to download the model ollama run llama3:70b-instruct fairly quickly at a speed of 30 MB per second. Here is my server.log. [server.log](https://github.com/ollama/ollama/files/15062206/server.log) ![Schermafbeelding 2024-04-22 133915](https://github.com/ollama/ollama/assets/9640147/195d26be-7909-49f9-bd5f-277ecb229576) ### OS Windows ### GPU Nvidia ### CPU Intel ### Ollama version llama3:70b-instruct
GiteaMirror added the bug label 2026-04-12 12:40:46 -05:00
GiteaMirror commented 2026-04-12 12:40:47 -05:00
Author
Owner
Copy Link

@markhill343 commented on GitHub (Apr 23, 2024):

The 70B should be around ~50GB. The GPU only works with the 24GB it loads into the VRAM. The remaing part is stored in the system ram and only your CPU can process that. So you basically need another 4090.

<!-- gh-comment-id:2071693729 --> @markhill343 commented on GitHub (Apr 23, 2024): The 70B should be around ~50GB. The GPU only works with the 24GB it loads into the VRAM. The remaing part is stored in the system ram and only your CPU can process that. So you basically need another 4090.
GiteaMirror commented 2026-04-12 12:40:47 -05:00
Author
Owner
Copy Link

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

Yes quite a bit of the model is running on CPU in this case so it's expected to be slow, sorry

<!-- gh-comment-id:2103558919 --> @jmorganca commented on GitHub (May 9, 2024): Yes quite a bit of the model is running on CPU in this case so it's expected to be slow, sorry
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#2361
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?