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 #13190] Vulkan an, nur am loopen #55233

New Issue
Open
opened 2026-04-29 08:34:18 -05:00 by GiteaMirror · 8 comments
GiteaMirror commented 2026-04-29 08:34:18 -05:00
Owner
Copy Link

Originally created by @Profex86 on GitHub (Nov 21, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/13190

Originally assigned to: @dhiltgen on GitHub.

What is the issue?

Das Bild sagt eigentlich alles aus. Ollama Version ist die Aktuelle.

Image

Relevant log output

OS

Windows

GPU

AMD

CPU

AMD

Ollama version

0.13.0

Originally created by @Profex86 on GitHub (Nov 21, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/13190 Originally assigned to: @dhiltgen on GitHub. ### What is the issue? Das Bild sagt eigentlich alles aus. Ollama Version ist die Aktuelle. <img width="2559" height="1439" alt="Image" src="https://github.com/user-attachments/assets/33b4be46-8faf-439d-9a07-380500830f64" /> ### Relevant log output ```shell ``` ### OS Windows ### GPU AMD ### CPU AMD ### Ollama version 0.13.0
GiteaMirror added the vulkanbug labels 2026-04-29 08:34:18 -05:00
GiteaMirror commented 2026-04-29 08:34:21 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Nov 21, 2025):

Im Log steht mehr.

<!-- gh-comment-id:3562474265 --> @rick-github commented on GitHub (Nov 21, 2025): Im [Log](https://docs.ollama.com/troubleshooting) steht mehr.
GiteaMirror commented 2026-04-29 08:34:22 -05:00
Author
Owner
Copy Link

@Profex86 commented on GitHub (Nov 21, 2025):

ollama_backend.log

Vulkaninfo.log

Image
<!-- gh-comment-id:3562514691 --> @Profex86 commented on GitHub (Nov 21, 2025): [ollama_backend.log](https://github.com/user-attachments/files/23674209/ollama_backend.log) [Vulkaninfo.log](https://github.com/user-attachments/files/23674295/Vulkaninfo.log) <img width="1969" height="1393" alt="Image" src="https://github.com/user-attachments/assets/4327bd0a-b171-4f64-b000-2fb78e3f345a" />
GiteaMirror commented 2026-04-29 08:34:25 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Nov 21, 2025):

https://github.com/ollama/ollama/issues/12600#issuecomment-3556205865

<!-- gh-comment-id:3562539734 --> @rick-github commented on GitHub (Nov 21, 2025): https://github.com/ollama/ollama/issues/12600#issuecomment-3556205865
GiteaMirror commented 2026-04-29 08:34:30 -05:00
Author
Owner
Copy Link

@Profex86 commented on GitHub (Nov 21, 2025):

#12600 (comment)

On Nobara (Linux) the exact same Ollama version runs perfectly – I can fully utilize the total 128 GB of VRAM without any issues.
But this not my problem, my models answered symbols or loop words on windows.
On Windows, it’s not about VRAM at all: no model runs, whether it’s 1‑bit or 120‑bit. Something is fundamentally broken, almost as if my prompt isn’t being correctly converted into tokens.

<!-- gh-comment-id:3562580761 --> @Profex86 commented on GitHub (Nov 21, 2025): > [#12600 (comment)](https://github.com/ollama/ollama/issues/12600#issuecomment-3556205865) On Nobara (Linux) the exact same Ollama version runs perfectly – I can fully utilize the total 128 GB of VRAM without any issues. But this not my problem, my models answered symbols or loop words on windows. On Windows, it’s not about VRAM at all: no model runs, whether it’s 1‑bit or 120‑bit. Something is fundamentally broken, almost as if my prompt isn’t being correctly converted into tokens.
GiteaMirror commented 2026-04-29 08:34:35 -05:00
Author
Owner
Copy Link

@Profex86 commented on GitHub (Nov 21, 2025):

Temporary one time Fix (Windows, AMD MI50): I was able to get Ollama running by replacing the Vulkan backend files with the ones shipped in LM Studio.

Copied ggml-vulkan.dll and vulkan-1.dll from LM Studio’s install folder into Ollama’s program directory.

After that, models answered correctly. but, restart ollama, same thing, answer is broken

<!-- gh-comment-id:3562659190 --> @Profex86 commented on GitHub (Nov 21, 2025): Temporary one time Fix (Windows, AMD MI50): I was able to get Ollama running by replacing the Vulkan backend files with the ones shipped in LM Studio. Copied ggml-vulkan.dll and vulkan-1.dll from LM Studio’s install folder into Ollama’s program directory. After that, models answered correctly. but, restart ollama, same thing, answer is broken
GiteaMirror commented 2026-04-29 08:34:37 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Nov 21, 2025):

PR #12992 contains 40 updates to Vulkan from upstream, so it's possible the drift vs. LM Studio might be resolved once that's merged.

<!-- gh-comment-id:3564982008 --> @dhiltgen commented on GitHub (Nov 21, 2025): PR #12992 contains 40 updates to Vulkan from upstream, so it's possible the drift vs. LM Studio might be resolved once that's merged.
GiteaMirror commented 2026-04-29 08:34:38 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Dec 5, 2025):

Please give the 0.13.2 RC a try and see if you see better results
https://github.com/ollama/ollama/releases

<!-- gh-comment-id:3614877037 --> @dhiltgen commented on GitHub (Dec 5, 2025): Please give the 0.13.2 RC a try and see if you see better results https://github.com/ollama/ollama/releases
GiteaMirror commented 2026-04-29 08:34:43 -05:00
Author
Owner
Copy Link

@Profex86 commented on GitHub (Dec 29, 2025):

Es hat auf Anhieb perfekt funktioniert! Momentan optimiere ich, wie man unter Windows die bestmögliche Performance herausholen kann. Meine Erkenntnisse werde ich hier bald teilen. Liebe Grüße

<!-- gh-comment-id:3696991529 --> @Profex86 commented on GitHub (Dec 29, 2025): Es hat auf Anhieb perfekt funktioniert! Momentan optimiere ich, wie man unter Windows die bestmögliche Performance herausholen kann. Meine Erkenntnisse werde ich hier bald teilen. Liebe Grüße
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 vulkan
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#55233
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?