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 #4809] Add information on RAM and VRAM requirements of model quantization in library #3035

New Issue
Open
opened 2026-04-12 13:27:21 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 13:27:21 -05:00
Owner
Copy Link

Originally created by @geroldmeisinger on GitHub (Jun 4, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4809

this would help to decide which model can run
a) at all
b) at highest-quality although slowly
c) at highest-quality at GPU-speed
before downloading huge files

I think one of the great plus of Ollama is the curated model library. adding memory information would make it even more convenient.

thank you for this great tool!

Originally created by @geroldmeisinger on GitHub (Jun 4, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4809 this would help to decide which model can run a) at all b) at highest-quality although slowly c) at highest-quality at GPU-speed before downloading huge files I think one of the great plus of Ollama is the curated model library. adding memory information would make it even more convenient. _thank you for this great tool!_
GiteaMirror added the feature request label 2026-04-12 13:27:21 -05:00
GiteaMirror commented 2026-04-12 13:27:22 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Jul 11, 2024):

@geroldmeisinger sorry for the slow response. Ollama actually does graph prediction to see whether or not a model can be loaded into memory, however, the graph is actually different depending on whether you're loading it onto GPU or CPU (or split between the two), so it's somewhat difficult to know ahead of time unless we know the specs of your system.

I love the idea in principle though. It's really frustrating to download a model and then find out it's not going to run.

<!-- gh-comment-id:2221881482 --> @pdevine commented on GitHub (Jul 11, 2024): @geroldmeisinger sorry for the slow response. Ollama actually does graph prediction to see whether or not a model can be loaded into memory, however, the graph is actually different depending on whether you're loading it onto GPU or CPU (or split between the two), so it's _somewhat_ difficult to know ahead of time unless we know the specs of your system. I love the idea in principle though. It's really frustrating to download a model and then find out it's not going to run.
GiteaMirror commented 2026-04-12 13:27:22 -05:00
Author
Owner
Copy Link

@geroldmeisinger commented on GitHub (Jul 11, 2024):

thanks for the answer. is it really that different from system to system though? most GPUs come with either 8GB, 12GB, 16GB, 24GB... VRAM so knowing the +-2GB bracket would be close enough.

<!-- gh-comment-id:2222101652 --> @geroldmeisinger commented on GitHub (Jul 11, 2024): thanks for the answer. is it really that different from system to system though? most GPUs come with either 8GB, 12GB, 16GB, 24GB... VRAM so knowing the +-2GB bracket would be close enough.
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 feature request
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#3035
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?