github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #1402] Optimum-NVIDIA - Unlock blazingly fast LLM inference in just 1 line of code #47256

New Issue
Closed
opened 2026-04-28 03:28:36 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-28 03:28:36 -05:00
Owner
Copy Link

Originally created by @iplayfast on GitHub (Dec 6, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1402

This article looks like it would be worth while to implement a check to see if nvidia is installed and add that line of code. Looks like it's quantizing an 8 bit float.

https://huggingface.co/blog/optimum-nvidia

Originally created by @iplayfast on GitHub (Dec 6, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1402 This article looks like it would be worth while to implement a check to see if nvidia is installed and add that line of code. Looks like it's quantizing an 8 bit float. https://huggingface.co/blog/optimum-nvidia
GiteaMirror commented 2026-04-28 03:28:37 -05:00
Author
Owner
Copy Link

@easp commented on GitHub (Dec 7, 2023):

As always, it's important to understand the conditions under which the claimed speedup was achieved. In this case:

  • High end GPU
  • fp8 model weights
  • Batch size of 4
  • Using tansformers python package

Ollama doesn't use the transformers python package, so this isn't going to be a single line of code to implement.
It uses a batch size of 1, already uses quantized model weights, and uses other optimizations (through llama.cpp). The quantization alone probably accounts for a significant part of of the 2.6x speedup they reported on the 4090 GPU.

<!-- gh-comment-id:1846145834 --> @easp commented on GitHub (Dec 7, 2023): As always, it's important to understand the conditions under which the claimed speedup was achieved. In this case: - High end GPU - fp8 model weights - Batch size of 4 - Using tansformers python package Ollama doesn't use the transformers python package, so this isn't going to be a single line of code to implement. It uses a batch size of 1, already uses quantized model weights, and uses other optimizations (through llama.cpp). The quantization alone probably accounts for a significant part of of the 2.6x speedup they reported on the 4090 GPU.
GiteaMirror commented 2026-04-28 03:28:37 -05:00
Author
Owner
Copy Link

@iplayfast commented on GitHub (Dec 7, 2023):

I suspected as much, but didn't want to let an opportunity to enhance ollama pass by.

<!-- gh-comment-id:1846149620 --> @iplayfast commented on GitHub (Dec 7, 2023): I suspected as much, but didn't want to let an opportunity to enhance ollama pass by.
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
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#47256
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?