github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #1368] Different behavior between running on the host versus running on GPUs. #26480

New Issue
Closed
opened 2026-04-22 02:46:36 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-22 02:46:36 -05:00
Owner
Copy Link

Originally created by @phalexo on GitHub (Dec 4, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1368

Originally assigned to: @dhiltgen on GitHub.

When running on the GPUs (one or more) the output is either one character or one, seemingly, unrelated word and then lines of '#"
It does it for a while. Sometimes it gets an exception in cuBLAS error 15 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:7586

But the same model(s) run on my host and produce output.

Originally created by @phalexo on GitHub (Dec 4, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1368 Originally assigned to: @dhiltgen on GitHub. When running on the GPUs (one or more) the output is either one character or one, seemingly, unrelated word and then lines of '#" It does it for a while. Sometimes it gets an exception in cuBLAS error 15 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:7586 But the same model(s) run on my host and produce output.
GiteaMirror commented 2026-04-22 02:46:37 -05:00
Author
Owner
Copy Link

@phalexo commented on GitHub (Dec 4, 2023):

I have rebuilt ollama from the cloned source, and I still have the same issue. Junk output when running on GPUs and an error/exception when I input a second query.

a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
anca#####################################################################################################################################################################g{"timestamp":1701707692,"level":"INFO","function":"log_server_request","line":2478,"message":"request","remote_addr":"127.0.0.1","remote_port":48378,"status":200,"method":"POST","path":"/completion","params":{}}
{"timestamp":1701707692,"level":"INFO","function":"log_server_request","line":2478,"message":"request","remote_addr":"127.0.0.1","remote_port":42320,"status":200,"method":"POST","path":"/tokenize","params":{}}
[GIN] 2023/12/04 - 11:34:52 | 200 | 21.627947963s | 127.0.0.1 | POST "/api/generate"

Hello World.
{"timestamp":1701707715,"level":"INFO","function":"log_server_request","line":2478,"message":"request","remote_addr":"127.0.0.1","remote_port":34652,"status":200,"method":"HEAD","path":"/","params":{}}
{"timestamp":1701707715,"level":"INFO","function":"log_server_request","line":2478,"message":"request","remote_addr":"127.0.0.1","remote_port":34652,"status":200,"method":"POST","path":"/detokenize","params":{}}

cuBLAS error 15 at /home/developer/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:7586
current device: 0
⠸ 2023/12/04 11:35:16 llama.go:436: exit status 1
2023/12/04 11:35:16 llama.go:510: llama runner stopped successfully
[GIN] 2023/12/04 - 11:35:16 | 200 | 494.763097ms | 127.0.0.1 | POST "/api/generate"
Error: llama runner exited, you may not have enough available memory to run this model

<!-- gh-comment-id:1839031647 --> @phalexo commented on GitHub (Dec 4, 2023): I have rebuilt ollama from the cloned source, and I still have the same issue. Junk output when running on GPUs and an error/exception when I input a second query. a a a a a a a a a a a a a a a anca#####################################################################################################################################################################g{"timestamp":1701707692,"level":"INFO","function":"log_server_request","line":2478,"message":"request","remote_addr":"127.0.0.1","remote_port":48378,"status":200,"method":"POST","path":"/completion","params":{}} {"timestamp":1701707692,"level":"INFO","function":"log_server_request","line":2478,"message":"request","remote_addr":"127.0.0.1","remote_port":42320,"status":200,"method":"POST","path":"/tokenize","params":{}} [GIN] 2023/12/04 - 11:34:52 | 200 | 21.627947963s | 127.0.0.1 | POST "/api/generate" >>> >>> <s> Hello World. </s> {"timestamp":1701707715,"level":"INFO","function":"log_server_request","line":2478,"message":"request","remote_addr":"127.0.0.1","remote_port":34652,"status":200,"method":"HEAD","path":"/","params":{}} {"timestamp":1701707715,"level":"INFO","function":"log_server_request","line":2478,"message":"request","remote_addr":"127.0.0.1","remote_port":34652,"status":200,"method":"POST","path":"/detokenize","params":{}} cuBLAS error 15 at /home/developer/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:7586 current device: 0 ⠸ 2023/12/04 11:35:16 llama.go:436: exit status 1 2023/12/04 11:35:16 llama.go:510: llama runner stopped successfully [GIN] 2023/12/04 - 11:35:16 | 200 | 494.763097ms | 127.0.0.1 | POST "/api/generate" Error: llama runner exited, you may not have enough available memory to run this model
GiteaMirror commented 2026-04-22 02:46:38 -05:00
Author
Owner
Copy Link

@BruceMacD commented on GitHub (Dec 4, 2023):

This is probably related to a multi-gpu bug that has been on-going. Linking for future reference: #969

<!-- gh-comment-id:1839675821 --> @BruceMacD commented on GitHub (Dec 4, 2023): This is probably related to a multi-gpu bug that has been on-going. Linking for future reference: #969
GiteaMirror commented 2026-04-22 02:46:38 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Jan 27, 2024):

Quite a few GPU related fixes have gone in over the past few weeks. Please give this another try with the latest release 0.1.22 and let us know if you're still hitting the problem.

<!-- gh-comment-id:1912911100 --> @dhiltgen commented on GitHub (Jan 27, 2024): Quite a few GPU related fixes have gone in over the past few weeks. Please give this another try with the latest release 0.1.22 and let us know if you're still hitting the problem.
GiteaMirror commented 2026-04-22 02:46:39 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Feb 1, 2024):

If you're still having problems with 0.1.22 or newer, please re-open.

<!-- gh-comment-id:1922457583 --> @dhiltgen commented on GitHub (Feb 1, 2024): If you're still having problems with 0.1.22 or newer, please re-open.
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#26480
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?