github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-22 13:42:25 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #6741] Llama 3.1 70b 128k context not fitting 96Gb #30008

New Issue
Closed
opened 2026-04-22 09:24:38 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-22 09:24:38 -05:00
Owner
Copy Link

Originally created by @dmatora on GitHub (Sep 11, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/6741

What is the issue?

Not only it doesn't fit 96Gb (offloading only 10 layers out of 81), but processing actual ~128k request crashes with CUDA error: out of memory on 160Gb (will all layers offloaded)

As mentioned here https://github.com/ollama/ollama/issues/6279#issuecomment-2342546437_
this is obviously a bug

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.3.10

Originally created by @dmatora on GitHub (Sep 11, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/6741 ### What is the issue? Not only it doesn't fit 96Gb (offloading only 10 layers out of 81), but processing actual ~128k request crashes with `CUDA error: out of memory` on 160Gb (will all layers offloaded) As mentioned here https://github.com/ollama/ollama/issues/6279#issuecomment-2342546437_ this is obviously a bug ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.3.10
GiteaMirror added the nvidiamemorybug labels 2026-04-22 09:24:39 -05:00
GiteaMirror commented 2026-04-22 09:24:40 -05:00
Author
Owner
Copy Link

@dmatora commented on GitHub (Sep 11, 2024):

ollama show --modelfile llama3.1:70b > Modelfile3.1-70b
echo 'PARAMETER num_ctx 131072' >> Modelfile3.1-70b
ollama create llama3.1:70b-128k -f Modelfile3.1-70b

optionally echo 'PARAMETER num_gpu 81' >> Modelfile3.1-70b to force 81 layers offloading on 4x4090

<!-- gh-comment-id:2344282365 --> @dmatora commented on GitHub (Sep 11, 2024): ``` ollama show --modelfile llama3.1:70b > Modelfile3.1-70b echo 'PARAMETER num_ctx 131072' >> Modelfile3.1-70b ollama create llama3.1:70b-128k -f Modelfile3.1-70b ``` optionally `echo 'PARAMETER num_gpu 81' >> Modelfile3.1-70b` to force 81 layers offloading on 4x4090
GiteaMirror commented 2026-04-22 09:24:41 -05:00
Author
Owner
Copy Link

@dmatora commented on GitHub (Mar 14, 2025):

Issue was ollama didn't support context quantisation which is solved now

<!-- gh-comment-id:2723656177 --> @dmatora commented on GitHub (Mar 14, 2025): Issue was ollama didn't support context quantisation which is solved 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 memory nvidia
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#30008
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?