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

[PR #10882] [MERGED] kvcache: Skip computing causal mask for worst case graph reservation #44641

New Issue
Closed
opened 2026-04-25 00:15:02 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-25 00:15:02 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/10882
Author: @jessegross
Created: 5/27/2025
Status: ✅ Merged
Merged: 5/27/2025
Merged by: @jessegross

Base: main ← Head: jessegross/cache_start

📝 Commits (1)

  • 9cb1752 kvcache: Skip computing causal mask for worst case graph reservation

📊 Changes

1 file changed (+12 additions, -1 deletions)

View changed files

📝 kvcache/causal.go (+12 -1)

📄 Description

Computing an attention mask for a large context and max batch is expensive - over 100ms. Models like Gemma3 that have multiple types of caches and custom attention masks need to do this 4 times, so this adds approximately 500ms to startup time when using 128k context

When we are reserving the worst case graph, we don't need the mask, only its shape, so we can skip this.

🔄 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/10882 **Author:** [@jessegross](https://github.com/jessegross) **Created:** 5/27/2025 **Status:** ✅ Merged **Merged:** 5/27/2025 **Merged by:** [@jessegross](https://github.com/jessegross) **Base:** `main` ← **Head:** `jessegross/cache_start` --- ### 📝 Commits (1) - [`9cb1752`](https://github.com/ollama/ollama/commit/9cb1752bb61b57c8e5c28db2d7e8aeb40fe2b455) kvcache: Skip computing causal mask for worst case graph reservation ### 📊 Changes **1 file changed** (+12 additions, -1 deletions) <details> <summary>View changed files</summary> 📝 `kvcache/causal.go` (+12 -1) </details> ### 📄 Description Computing an attention mask for a large context and max batch is expensive - over 100ms. Models like Gemma3 that have multiple types of caches and custom attention masks need to do this 4 times, so this adds approximately 500ms to startup time when using 128k context When we are reserving the worst case graph, we don't need the mask, only its shape, so we can skip this. --- <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 2026-04-25 00:15:02 -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#44641
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?