github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 05:05:37 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[PR #15301] [MERGED] ggml: skip cublasGemmBatchedEx during graph reservation #61804

New Issue
Closed
opened 2026-04-29 16:49:06 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-29 16:49:06 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/15301
Author: @jessegross
Created: 4/3/2026
Status: ✅ Merged
Merged: 4/3/2026
Merged by: @jessegross

Base: main ← Head: jessegross/gemma4-reserve

📝 Commits (1)

  • 0cf2922 ggml: skip cublasGemmBatchedEx during graph reservation

📊 Changes

2 files changed (+45 additions, -3 deletions)

View changed files

📝 llama/patches/0020-ggml-No-alloc-mode.patch (+24 -3)
📝 ml/backend/ggml/ggml/src/ggml-cuda/common.cuh (+21 -0)

📄 Description

cublasGemmBatchedEx fails during graph capture when pool allocations return fake pointers. This is triggered when NUM_PARALLEL is greater than 1 for models like gemma4 that use batched matmuls. Skip it during reservation since the memory tracking is already handled by the pool allocations.

Fixes #15249

🔄 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/15301 **Author:** [@jessegross](https://github.com/jessegross) **Created:** 4/3/2026 **Status:** ✅ Merged **Merged:** 4/3/2026 **Merged by:** [@jessegross](https://github.com/jessegross) **Base:** `main` ← **Head:** `jessegross/gemma4-reserve` --- ### 📝 Commits (1) - [`0cf2922`](https://github.com/ollama/ollama/commit/0cf2922e9308b0b3e07f9c74b38b59c7367b331d) ggml: skip cublasGemmBatchedEx during graph reservation ### 📊 Changes **2 files changed** (+45 additions, -3 deletions) <details> <summary>View changed files</summary> 📝 `llama/patches/0020-ggml-No-alloc-mode.patch` (+24 -3) 📝 `ml/backend/ggml/ggml/src/ggml-cuda/common.cuh` (+21 -0) </details> ### 📄 Description cublasGemmBatchedEx fails during graph capture when pool allocations return fake pointers. This is triggered when NUM_PARALLEL is greater than 1 for models like gemma4 that use batched matmuls. Skip it during reservation since the memory tracking is already handled by the pool allocations. Fixes #15249 --- <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 2026-04-29 16:49:06 -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#61804
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?