github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-21 13:11:39 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #13920] ROCm fails to initialize on AMD Radeon RX 9070 XT #9109

New Issue
Open
opened 2026-04-12 21:57:42 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-12 21:57:42 -05:00
Owner
Copy Link

Originally created by @GrouwBer on GitHub (Jan 26, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/13920

What is the issue?

server.log

Ollama detects the AMD Radeon RX 9070 XT but the logs show the device being filtered out : "filtering device which didn't fully initialize"

Relevant log output

time=2026-01-26T16:13:57.699-03:00 level=DEBUG source=runner.go:153 msg="filtering device which didn't fully initialize" id=0 libdir=C:\Users\aldoc\AppData\Local\Programs\Ollama\lib\ollama\rocm pci_id=0000:03:00.0 library=ROCm
time=2026-01-26T16:13:57.699-03:00 level=DEBUG source=runner.go:40 msg="GPU bootstrap discovery took" duration=1.0199949s
time=2026-01-26T16:13:57.699-03:00 level=INFO source=types.go:60 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="95.6 GiB" available="81.3 GiB"
time=2026-01-26T16:13:57.699-03:00 level=INFO source=routes.go:1725 msg="entering low vram mode" "total vram"="0 B" threshold="20.0 GiB"

OS

Windows

GPU

AMD

CPU

AMD

Ollama version

0.15.1

Originally created by @GrouwBer on GitHub (Jan 26, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/13920 ### What is the issue? [server.log](https://github.com/user-attachments/files/24866825/server.log) Ollama detects the AMD Radeon RX 9070 XT but the logs show the device being filtered out : "filtering device which didn't fully initialize" ### Relevant log output ```shell time=2026-01-26T16:13:57.699-03:00 level=DEBUG source=runner.go:153 msg="filtering device which didn't fully initialize" id=0 libdir=C:\Users\aldoc\AppData\Local\Programs\Ollama\lib\ollama\rocm pci_id=0000:03:00.0 library=ROCm time=2026-01-26T16:13:57.699-03:00 level=DEBUG source=runner.go:40 msg="GPU bootstrap discovery took" duration=1.0199949s time=2026-01-26T16:13:57.699-03:00 level=INFO source=types.go:60 msg="inference compute" id=cpu library=cpu compute="" name=cpu description=cpu libdirs=ollama driver="" pci_id="" type="" total="95.6 GiB" available="81.3 GiB" time=2026-01-26T16:13:57.699-03:00 level=INFO source=routes.go:1725 msg="entering low vram mode" "total vram"="0 B" threshold="20.0 GiB" ``` ### OS Windows ### GPU AMD ### CPU AMD ### Ollama version 0.15.1
GiteaMirror added the bug label 2026-04-12 21:57:42 -05:00
GiteaMirror commented 2026-04-12 21:57:44 -05:00
Author
Owner
Copy Link

@brtptrs commented on GitHub (Jan 27, 2026):

I am testing ollama on a Strix Halo system running Ubuntu 26.04 dev
Until a few days back, it ran amazingly well.

Now ollama starts but any action results in this error:
Error: 500 Internal Server Error: do load request: Post "http://127.0.0.1:42829/load": EOF

~$ sudo journalctl -u ollama -f
Shows this error:

Jan 27 08:56:29 xyz1 ollama[10339]: time=2026-01-27T08:56:29.353+01:00 level=INFO source=sched.go:479 msg="Load failed" model=/usr/share/ollama/.ollama/models/blobs/sha256-bfb40fc6bb9c3b2ed529b480e04f824c005ea8f86733d4ebbf0c204de484891e error="do load request: Post "http://127.0.0.1:32911/load": EOF"

Jan 27 08:56:29 xyz1 ollama[10339]: time=2026-01-27T08:56:29.353+01:00 level=ERROR source=server.go:302 msg="llama runner terminated" error="exit status 2"

Jan 27 08:58:47 xyz1 ollama[10339]: ROCm error: out of memory

Jan 27 08:58:47 xyz1 ollama[10339]: //ml/backend/ggml/ggml/src/ggml-cuda/ggml-cuda.cu:94: ROCm error

Jan 27 08:58:48 xyz1 ollama[10339]: time=2026-01-27T08:58:48.470+01:00 level=ERROR source=server.go:302 msg="llama runner terminated" error="exit status 2"

Jan 27 08:58:48 xyz1 ollama[10339]: time=2026-01-27T08:58:48.470+01:00 level=INFO source=sched.go:479 msg="Load failed" model=/usr/share/ollama/.ollama/models/blobs/sha256-bfb40fc6bb9c3b2ed5bbf0c204de484891e error="do load request: Post "http://127.0.0.1:37053/load": EOF"

<!-- gh-comment-id:3804007394 --> @brtptrs commented on GitHub (Jan 27, 2026): I am testing ollama on a Strix Halo system running Ubuntu 26.04 dev Until a few days back, it ran amazingly well. Now ollama starts but any action results in this error: Error: 500 Internal Server Error: do load request: Post "http://127.0.0.1:42829/load": EOF ~$ sudo journalctl -u ollama -f Shows this error: Jan 27 08:56:29 xyz1 ollama[10339]: time=2026-01-27T08:56:29.353+01:00 level=INFO source=sched.go:479 msg="Load failed" model=/usr/share/ollama/.ollama/models/blobs/sha256-bfb40fc6bb9c3b2ed529b480e04f824c005ea8f86733d4ebbf0c204de484891e error="do load request: Post \"http://127.0.0.1:32911/load\": EOF" Jan 27 08:56:29 xyz1 ollama[10339]: time=2026-01-27T08:56:29.353+01:00 level=ERROR source=server.go:302 msg="llama runner terminated" error="exit status 2" Jan 27 08:58:47 xyz1 ollama[10339]: ROCm error: out of memory Jan 27 08:58:47 xyz1 ollama[10339]: //ml/backend/ggml/ggml/src/ggml-cuda/ggml-cuda.cu:94: ROCm error Jan 27 08:58:48 xyz1 ollama[10339]: time=2026-01-27T08:58:48.470+01:00 level=ERROR source=server.go:302 msg="llama runner terminated" error="exit status 2" Jan 27 08:58:48 xyz1 ollama[10339]: time=2026-01-27T08:58:48.470+01:00 level=INFO source=sched.go:479 msg="Load failed" model=/usr/share/ollama/.ollama/models/blobs/sha256-bfb40fc6bb9c3b2ed5bbf0c204de484891e error="do load request: Post \"http://127.0.0.1:37053/load\": EOF"
GiteaMirror commented 2026-04-12 21:57:45 -05:00
Author
Owner
Copy Link

@codeliger commented on GitHub (Jan 28, 2026):

For some reason EndeavourOS (arch linux) now installs nvidia packages by default, I used yay -Qs nvidia to find all nvidia packages then yay -Rc {package} to remove the package and all dependencies. I think the linux-firmware-nvidia was the problem. I removed it and installed linux-firmware-amdgpu and restarted the computer, then reinstalled ollama and it now detects my 9070 XT properly.

<!-- gh-comment-id:3811915209 --> @codeliger commented on GitHub (Jan 28, 2026): For some reason EndeavourOS (arch linux) now installs nvidia packages by default, I used `yay -Qs nvidia` to find all nvidia packages then `yay -Rc {package}` to remove the package and all dependencies. I think the `linux-firmware-nvidia` was the problem. I removed it and installed `linux-firmware-amdgpu` and restarted the computer, then reinstalled ollama and it now detects my 9070 XT properly.
GiteaMirror commented 2026-04-12 21:57:46 -05:00
Author
Owner
Copy Link

@brtptrs commented on GitHub (Jan 29, 2026):

I was running kernel 6.18 and had to use the --no-dkms flag when running amdgpu-install

Temporary solution (for me):

  • Start ubuntu with kernel 6.17
  • Remove all 6.18 packages
  • run amdgpu-install without the --no-dkms flag

Now ollama version is 0.15.2 is running again!

Mini Desktop Ryzen AI Max+ 395 96GB/128GB+

<!-- gh-comment-id:3817128040 --> @brtptrs commented on GitHub (Jan 29, 2026): I was running kernel 6.18 and had to use the --no-dkms flag when running amdgpu-install Temporary solution (for me): - Start ubuntu with kernel 6.17 - Remove all 6.18 packages - run amdgpu-install without the --no-dkms flag Now ollama version is 0.15.2 is running again! Mini Desktop Ryzen AI Max+ 395 96GB/128GB+
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
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#9109
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?