github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #2897] Windows preview CUDA 5.2 support #1772

New Issue
Closed
opened 2026-04-12 11:47:25 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-12 11:47:25 -05:00
Owner
Copy Link

Originally created by @lyczak on GitHub (Mar 3, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/2897

Originally assigned to: @dhiltgen on GitHub.

Hello folks,

I've been trying to get started with the Windows preview version of ollama. However, I'm currently encountering an issue where my GTX 970 is not detected by the software. I've tried updating drivers and updating Windows to no avail. Assuming this is related to old CUDA version (CUDA 5.2) as mentioned in #1865 then it should've been fixed by #2116 but I don't know if this fix has been tested on the Windows preview version of ollama. Poking around in that PR, it seems like the commit which adds support for CUDA 5.0, 7.5, and 8.0 touches gen_common.sh, gen_linux.sh, and gen_windows.ps1 under llm/generate whereas the commit targeting CUDA 5.2 only touches gen_linux.sh. Could this be the source of the issue?

Assuming this was the problem, I was hoping to try using WSL but unfortunately I'm running Windows Server 2019 and can't install WSL2. I may set up a dual boot with Ubuntu later today to see if my GPU is recognized there. I'm more than happy to help with additional testing although I don't have time to setup the toolchain and build things myself right this moment. Thanks for helping to maintain this project!

Relevant lines from my server.log are as follows:

time=2024-03-02T11:54:38.873-05:00 level=INFO source=gpu.go:94 msg="Detecting GPU type"
time=2024-03-02T11:54:39.015-05:00 level=INFO source=gpu.go:265 msg="Searching for GPU management library nvml.dll"
time=2024-03-02T11:54:45.186-05:00 level=INFO source=gpu.go:311 msg="Discovered GPU libraries: []"
time=2024-03-02T11:54:45.186-05:00 level=INFO source=gpu.go:265 msg="Searching for GPU management library rocm_smi64.dll"
time=2024-03-02T11:54:45.190-05:00 level=INFO source=gpu.go:311 msg="Discovered GPU libraries: []"
time=2024-03-02T11:54:45.190-05:00 level=INFO source=cpu_common.go:11 msg="CPU has AVX2"
time=2024-03-02T11:54:45.190-05:00 level=INFO source=cpu_common.go:11 msg="CPU has AVX2"
time=2024-03-02T11:54:45.191-05:00 level=INFO source=llm.go:77 msg="GPU not available, falling back to CPU"
Originally created by @lyczak on GitHub (Mar 3, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/2897 Originally assigned to: @dhiltgen on GitHub. Hello folks, I've been trying to get started with the Windows preview version of ollama. However, I'm currently encountering an issue where my GTX 970 is not detected by the software. I've tried updating drivers and updating Windows to no avail. Assuming this is related to old CUDA version (CUDA 5.2) as mentioned in #1865 then it should've been fixed by #2116 but I don't know if this fix has been tested on the Windows preview version of ollama. Poking around in that PR, it seems like the commit which [adds support for CUDA 5.0, 7.5, and 8.0 ](https://github.com/ollama/ollama/pull/2116/commits/a447a083f2169e2a3c975cb5951d8b0b0dcddb04) touches `gen_common.sh`, `gen_linux.sh`, and `gen_windows.ps1` under `llm/generate` whereas the commit [targeting CUDA 5.2](https://github.com/ollama/ollama/pull/2116/commits/681a91499010be819dd45a1390e668b0817e7338) only touches `gen_linux.sh`. Could this be the source of the issue? Assuming this was the problem, I was hoping to try using WSL but unfortunately I'm running Windows Server 2019 and can't install WSL2. I may set up a dual boot with Ubuntu later today to see if my GPU is recognized there. I'm more than happy to help with additional testing although I don't have time to setup the toolchain and build things myself right this moment. Thanks for helping to maintain this project! Relevant lines from my server.log are as follows: ``` time=2024-03-02T11:54:38.873-05:00 level=INFO source=gpu.go:94 msg="Detecting GPU type" time=2024-03-02T11:54:39.015-05:00 level=INFO source=gpu.go:265 msg="Searching for GPU management library nvml.dll" time=2024-03-02T11:54:45.186-05:00 level=INFO source=gpu.go:311 msg="Discovered GPU libraries: []" time=2024-03-02T11:54:45.186-05:00 level=INFO source=gpu.go:265 msg="Searching for GPU management library rocm_smi64.dll" time=2024-03-02T11:54:45.190-05:00 level=INFO source=gpu.go:311 msg="Discovered GPU libraries: []" time=2024-03-02T11:54:45.190-05:00 level=INFO source=cpu_common.go:11 msg="CPU has AVX2" time=2024-03-02T11:54:45.190-05:00 level=INFO source=cpu_common.go:11 msg="CPU has AVX2" time=2024-03-02T11:54:45.191-05:00 level=INFO source=llm.go:77 msg="GPU not available, falling back to CPU" ```
GiteaMirror added the bugnvidia labels 2026-04-12 11:47:25 -05:00
GiteaMirror commented 2026-04-12 11:47:28 -05:00
Author
Owner
Copy Link

@lyczak commented on GitHub (Mar 3, 2024):

Update on this: I tested ollama on Ubuntu bare metal running on the same hardware and was able to make use of the GTX 970 without issue.

<!-- gh-comment-id:1975277724 --> @lyczak commented on GitHub (Mar 3, 2024): Update on this: I tested ollama on Ubuntu bare metal running on the same hardware and was able to make use of the GTX 970 without issue.
GiteaMirror commented 2026-04-12 11:47:29 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Mar 6, 2024):

The driver install should have installed nvml.dll into the system. The fact that we're not finding it is a bit odd.

Can you try running the server with OLLAMA_DEBUG="1" so we can get a little more information on what paths it's trying and maybe poke around a bit on your system to see if you can find where that library exists?

<!-- gh-comment-id:1981343706 --> @dhiltgen commented on GitHub (Mar 6, 2024): The driver install should have installed `nvml.dll` into the system. The fact that we're not finding it is a bit odd. Can you try running the server with `OLLAMA_DEBUG="1"` so we can get a little more information on what paths it's trying and maybe poke around a bit on your system to see if you can find where that library exists?
GiteaMirror commented 2026-04-12 11:47:30 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Mar 21, 2024):

If you're still having troubles, please grab a debug log so we can investigate.

<!-- gh-comment-id:2012037942 --> @dhiltgen commented on GitHub (Mar 21, 2024): If you're still having troubles, please grab a debug log so we can investigate.
GiteaMirror commented 2026-04-12 11:47:31 -05:00
Author
Owner
Copy Link

@lyczak commented on GitHub (Mar 21, 2024):

Hi Daniel,

Sorry I haven't gotten back to y'all on this issue. I've been swamped with other work and don't currently have access to the machine I was using to test this. You're welcome to keep this issue closed for now and once I have the time and access to the machine, I'll run some more tests and grab some debugging information for you. Thanks and take care!

<!-- gh-comment-id:2013144838 --> @lyczak commented on GitHub (Mar 21, 2024): Hi Daniel, Sorry I haven't gotten back to y'all on this issue. I've been swamped with other work and don't currently have access to the machine I was using to test this. You're welcome to keep this issue closed for now and once I have the time and access to the machine, I'll run some more tests and grab some debugging information for you. Thanks and take care!
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 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#1772
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?