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 #9220] cannot be used with preferred buffer type ROCm_Host, using CPU instead #52521

New Issue
Closed
opened 2026-04-28 23:36:23 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-28 23:36:23 -05:00
Owner
Copy Link

Originally created by @ca80000 on GitHub (Feb 19, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9220

llm_load_tensors: tensor 'token_embd.weight' (q4_K) (and 0 others) cannot be used with preferred buffer type ROCm_Host, using CPU instead

environment ： windows
GPU : 7900XTX
ROCm version : 6.1.2
model : huihui_ai/deepseek-r1-abliterated:32b

Help, How do I fix this

Originally created by @ca80000 on GitHub (Feb 19, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9220 llm_load_tensors: tensor 'token_embd.weight' (q4_K) (and 0 others) cannot be used with preferred buffer type ROCm_Host, using CPU instead environment ： windows GPU : 7900XTX ROCm version : 6.1.2 model : huihui_ai/deepseek-r1-abliterated:32b Help, How do I fix this
GiteaMirror commented 2026-04-28 23:36:24 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 19, 2025):

It's just a warning, you can ignore it. Server logs would confirm, but you are probably spilling from VRAM to system RAM.

<!-- gh-comment-id:2668342291 --> @rick-github commented on GitHub (Feb 19, 2025): It's just a warning, you can ignore it. [Server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues) would confirm, but you are probably spilling from VRAM to system RAM.
GiteaMirror commented 2026-04-28 23:36:24 -05:00
Author
Owner
Copy Link

@inkeliz commented on GitHub (Nov 1, 2025):

I'm hit the same "bug". The performance a disaster now. Even with "OLLAMA_NUM_PARALLEL=1" and "OLLAMA_CONTEXT_LENGTH=256" with a small 8b Q4 model it still showing:

llm_load_tensors: tensor 'token_embd.weight' (q4_K) (and 0 others) cannot be used with preferred buffer type ROCm_Host, using CPU instead

I'm using a similar card, a RX 7900XT, which has 20GB. It shows:

load_tensors: tensor 'token_embd.weight' (q4_K) (and 0 others) cannot be used with preferred buffer type ROCm_Host, using CPU instead
load_tensors: offloading 32 repeating layers to GPU
load_tensors: offloading output layer to GPU
load_tensors: offloaded 33/33 layers to GPU
load_tensors:        ROCm0 model buffer size =  4403.49 MiB
load_tensors:   CPU_Mapped model buffer size =   281.81 MiB
<!-- gh-comment-id:3476931681 --> @inkeliz commented on GitHub (Nov 1, 2025): I'm hit the same "bug". The performance a disaster now. Even with "OLLAMA_NUM_PARALLEL=1" and "OLLAMA_CONTEXT_LENGTH=256" with a small 8b Q4 model it still showing: ``` llm_load_tensors: tensor 'token_embd.weight' (q4_K) (and 0 others) cannot be used with preferred buffer type ROCm_Host, using CPU instead ``` I'm using a similar card, a RX 7900XT, which has 20GB. It shows: ``` load_tensors: tensor 'token_embd.weight' (q4_K) (and 0 others) cannot be used with preferred buffer type ROCm_Host, using CPU instead load_tensors: offloading 32 repeating layers to GPU load_tensors: offloading output layer to GPU load_tensors: offloaded 33/33 layers to GPU load_tensors: ROCm0 model buffer size = 4403.49 MiB load_tensors: CPU_Mapped model buffer size = 281.81 MiB ```
GiteaMirror commented 2026-04-28 23:36:24 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Nov 2, 2025):

Server log.

<!-- gh-comment-id:3476977141 --> @rick-github commented on GitHub (Nov 2, 2025): [Server log](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.mdx).
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#52521
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?