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 14.5k Packages Projects Releases 122 Wiki Activity

[PR #10226] [MERGED] ggml: Fix memory leak on input tensors #23725

New Issue
Closed
opened 2026-04-19 17:10:22 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-19 17:10:22 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/10226
Author: @jessegross
Created: 4/10/2025
Status: ✅ Merged
Merged: 4/11/2025
Merged by: @jessegross

Base: main ← Head: jessegross/leak

📝 Commits (3)

  • aab6020 ggml: Use pointer receivers for Context
  • a448106 ggml: Don't allocate CPU buffers as CUDA Host buffers
  • 62a5458 ggml: Fix memory leak on input tensors

📊 Changes

1 file changed (+35 additions, -26 deletions)

View changed files

📝 ml/backend/ggml/ggml.go (+35 -26)

📄 Description

For every forward pass through the model, we need to allocate input tensors: tokens, images, positions, outputs and masks. These get allocated in system memory.

However, when we close the context that the tensors were allocated through, the metadata gets freed but the actual backend memory does not. This results in a significant memory leak.

This makes it so that all the memory allocated through a context gets freed when it is closed.

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/ollama/ollama/pull/10226 **Author:** [@jessegross](https://github.com/jessegross) **Created:** 4/10/2025 **Status:** ✅ Merged **Merged:** 4/11/2025 **Merged by:** [@jessegross](https://github.com/jessegross) **Base:** `main` ← **Head:** `jessegross/leak` --- ### 📝 Commits (3) - [`aab6020`](https://github.com/ollama/ollama/commit/aab60209bfc9434da19ddbd03236e487cf789d76) ggml: Use pointer receivers for Context - [`a448106`](https://github.com/ollama/ollama/commit/a448106b57e617c0d6ebc9cc23203ead6a6058b5) ggml: Don't allocate CPU buffers as CUDA Host buffers - [`62a5458`](https://github.com/ollama/ollama/commit/62a5458582a9e835dc6d6c83869cfa99b2f7fd21) ggml: Fix memory leak on input tensors ### 📊 Changes **1 file changed** (+35 additions, -26 deletions) <details> <summary>View changed files</summary> 📝 `ml/backend/ggml/ggml.go` (+35 -26) </details> ### 📄 Description For every forward pass through the model, we need to allocate input tensors: tokens, images, positions, outputs and masks. These get allocated in system memory. However, when we close the context that the tensors were allocated through, the metadata gets freed but the actual backend memory does not. This results in a significant memory leak. This makes it so that all the memory allocated through a context gets freed when it is closed. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the pull-request label 2026-04-19 17:10:22 -05:00
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 pull-request
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#23725
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?