github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 17:49:24 -05:00
Code Issues 14.6k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #13340] Memory issues since v0.13.1 at embedding on Windows #34571

New Issue
Open
opened 2026-04-22 18:15:57 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-22 18:15:57 -05:00
Owner
Copy Link

Originally created by @negedng on GitHub (Dec 5, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/13340

Originally assigned to: @npardal on GitHub.

What is the issue?

Hi,

I used to embed a markdown file (3000+ chunks) using nomic-embed-text in batches of 32 with max char length 2000. Since the new update, I had to reduce the context length to 1900 chars. Reducing the batch size didn't solve the memory(?) issue.

Not a big problem for me, but I'll copy the logs for future reference

Relevant log output

## On the server:

[GIN] 2025/12/05 - 10:21:32 | 200 |    654.4864ms |       127.0.0.1 | POST     "/api/embed"
panic: caching disabled but unable to fit entire input in a batch

goroutine 50 [running]:
github.com/ollama/ollama/runner/ollamarunner.(*Server).computeBatch(0xc00020eb40, {0x142e, {0x7ff64d4f4c00, 0xc0002cc8c0}, {0x7ff64d501468, 0xc000009530}, {0xc000ea5308, 0x200, 0x25f}, {{0x7ff64d501468, ...}, ...}, ...})
        github.com/ollama/ollama/runner/ollamarunner/runner.go:707 +0x1ac5
github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc00020eb40, {0x7ff64d4e9840, 0xc0003920f0})
        github.com/ollama/ollama/runner/ollamarunner/runner.go:460 +0x30b
created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1
        github.com/ollama/ollama/runner/ollamarunner/runner.go:1411 +0x4c9
[GIN] 2025/12/05 - 10:21:32 | 500 |    358.1718ms |       127.0.0.1 | POST     "/api/embed"

## In the client:
do embedding request: Post "http://127.0.0.1:65153/embedding": read tcp 127.0.0.1:55845->127.0.0.1:65153: wsarecv: An existing connection was forcibly closed by the remote host.

OS

Windows

GPU

Nvidia

CPU

Intel

Ollama version

0.13.1

Originally created by @negedng on GitHub (Dec 5, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/13340 Originally assigned to: @npardal on GitHub. ### What is the issue? Hi, I used to embed a markdown file (3000+ chunks) using `nomic-embed-text` in batches of 32 with max char length 2000. Since the new update, I had to reduce the context length to 1900 chars. Reducing the batch size didn't solve the memory(?) issue. Not a big problem for me, but I'll copy the logs for future reference ### Relevant log output ```shell ## On the server: [GIN] 2025/12/05 - 10:21:32 | 200 | 654.4864ms | 127.0.0.1 | POST "/api/embed" panic: caching disabled but unable to fit entire input in a batch goroutine 50 [running]: github.com/ollama/ollama/runner/ollamarunner.(*Server).computeBatch(0xc00020eb40, {0x142e, {0x7ff64d4f4c00, 0xc0002cc8c0}, {0x7ff64d501468, 0xc000009530}, {0xc000ea5308, 0x200, 0x25f}, {{0x7ff64d501468, ...}, ...}, ...}) github.com/ollama/ollama/runner/ollamarunner/runner.go:707 +0x1ac5 github.com/ollama/ollama/runner/ollamarunner.(*Server).run(0xc00020eb40, {0x7ff64d4e9840, 0xc0003920f0}) github.com/ollama/ollama/runner/ollamarunner/runner.go:460 +0x30b created by github.com/ollama/ollama/runner/ollamarunner.Execute in goroutine 1 github.com/ollama/ollama/runner/ollamarunner/runner.go:1411 +0x4c9 [GIN] 2025/12/05 - 10:21:32 | 500 | 358.1718ms | 127.0.0.1 | POST "/api/embed" ## In the client: do embedding request: Post "http://127.0.0.1:65153/embedding": read tcp 127.0.0.1:55845->127.0.0.1:65153: wsarecv: An existing connection was forcibly closed by the remote host. ``` ### OS Windows ### GPU Nvidia ### CPU Intel ### Ollama version 0.13.1
GiteaMirror added the bug label 2026-04-22 18:15:57 -05:00
GiteaMirror commented 2026-04-22 18:15:59 -05:00
Author
Owner
Copy Link

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

This is caused by the batch size being too small - rather than reducing the batch size, you should set it to be the same size as the context length (or max input).

<!-- gh-comment-id:3618219897 --> @jessegross commented on GitHub (Dec 5, 2025): This is caused by the batch size being too small - rather than reducing the batch size, you should set it to be the same size as the context length (or max input).
GiteaMirror commented 2026-04-22 18:15:59 -05:00
Author
Owner
Copy Link

@negedng commented on GitHub (Dec 6, 2025):

I don't think so, reducing the batch size to 1 did not solve the issue. The problem was with the length of the input

<!-- gh-comment-id:3619826474 --> @negedng commented on GitHub (Dec 6, 2025): I don't think so, reducing the batch size to 1 did not solve the issue. The problem was with the length of the input
GiteaMirror commented 2026-04-22 18:16:00 -05:00
Author
Owner
Copy Link

@jessegross commented on GitHub (Dec 8, 2025):

Yes, you should increase the batch size, not decrease it.

<!-- gh-comment-id:3628611245 --> @jessegross commented on GitHub (Dec 8, 2025): Yes, you should _increase_ the batch size, not decrease it.
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#34571
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?