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 #2576] [CLOSED] Vulkan support: WIP, do not merge #10934

New Issue
Closed
opened 2026-04-12 23:16:05 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-12 23:16:05 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/2576
Author: @ddpasa
Created: 2/18/2024
Status: ❌ Closed

Base: main ← Head: vulkan

📝 Commits (10+)

  • 3629944 Move Mac App to a new dir
  • 2901180 Move hub auth out to new package
  • 16d1b5c Implement new Go based Desktop app
  • ee10ad7 Import "containerd/console" lib to support colorful output in Windows terminal
  • 85282fb Fix lint error on ignored error for win console
  • 30bb64e rerefactor
  • aeedbf4 set exe metadata using resource files
  • f6eb4ed update installer and app.exe metadata
  • 8b9ad33 better windows app and tray icons
  • 2c7e37e update installer title

📊 Changes

97 files changed (+2966 additions, -386 deletions)

View changed files

📝 .gitignore (+2 -1)
📝 app/.gitignore (+1 -92)
📝 app/README.md (+14 -13)
➕ app/assets/app.ico (+0 -0)
➕ app/assets/assets.go (+17 -0)
➕ app/assets/setup.bmp (+0 -0)
➕ app/assets/tray.ico (+0 -0)
➕ app/assets/tray_upgrade.ico (+0 -0)
➕ app/lifecycle/getstarted_nonwindows.go (+9 -0)
➕ app/lifecycle/getstarted_windows.go (+44 -0)
➕ app/lifecycle/lifecycle.go (+92 -0)
➕ app/lifecycle/logging.go (+46 -0)
➕ app/lifecycle/logging_nonwindows.go (+9 -0)
➕ app/lifecycle/logging_windows.go (+19 -0)
➕ app/lifecycle/paths.go (+79 -0)
➕ app/lifecycle/server.go (+139 -0)
➕ app/lifecycle/server_unix.go (+12 -0)
➕ app/lifecycle/server_windows.go (+13 -0)
➕ app/lifecycle/updater.go (+238 -0)
➕ app/lifecycle/updater_nonwindows.go (+12 -0)

...and 77 more files

📄 Description

This is a very preliminary implementation hack of Vulkan support, which llama.cpp recently added.

This is not intended to be merged. This code is far from there. I jsut want to get feedback from ollama devs and some pointers.

I tested this on an Intel Iris Plus G7 GPU on Linux. Phi-2 works fine with 20%-50% speedup compared to CPU with VNNI enabled. It behaves incorrectly for multimodal models such as Bakllava, which I'm still debugging.

I think I need to pull the latest llama.cpp commits to make it work properly, but updating the submodule is throwing bizarre compile time errors.

🔄 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/2576 **Author:** [@ddpasa](https://github.com/ddpasa) **Created:** 2/18/2024 **Status:** ❌ Closed **Base:** `main` ← **Head:** `vulkan` --- ### 📝 Commits (10+) - [`3629944`](https://github.com/ollama/ollama/commit/362994403a578a4d665d19cfa9325fd7de22e9da) Move Mac App to a new dir - [`2901180`](https://github.com/ollama/ollama/commit/290118060c23eb2731cff49f996c3740141a6834) Move hub auth out to new package - [`16d1b5c`](https://github.com/ollama/ollama/commit/16d1b5c3babf21cfa82f673755de57e0ef8d93a8) Implement new Go based Desktop app - [`ee10ad7`](https://github.com/ollama/ollama/commit/ee10ad700c24687f578d8ac46b0ccc6f576911de) Import "containerd/console" lib to support colorful output in Windows terminal - [`85282fb`](https://github.com/ollama/ollama/commit/85282fbb96762f8e6237b8126b732af9950ed0bf) Fix lint error on ignored error for win console - [`30bb64e`](https://github.com/ollama/ollama/commit/30bb64e889d1d6dec631aee3882c3bf804517d41) rerefactor - [`aeedbf4`](https://github.com/ollama/ollama/commit/aeedbf406cbda16e07be14b6b2f50f18b35b6a7a) set exe metadata using resource files - [`f6eb4ed`](https://github.com/ollama/ollama/commit/f6eb4ede9d49c12fb39896046d5ab2af6714ee7d) update installer and app.exe metadata - [`8b9ad33`](https://github.com/ollama/ollama/commit/8b9ad3346806a8bd40746991487e68fa63af3318) better windows app and tray icons - [`2c7e37e`](https://github.com/ollama/ollama/commit/2c7e37e2e66184cda93d5af15358fb4a093ade85) update installer title ### 📊 Changes **97 files changed** (+2966 additions, -386 deletions) <details> <summary>View changed files</summary> 📝 `.gitignore` (+2 -1) 📝 `app/.gitignore` (+1 -92) 📝 `app/README.md` (+14 -13) ➕ `app/assets/app.ico` (+0 -0) ➕ `app/assets/assets.go` (+17 -0) ➕ `app/assets/setup.bmp` (+0 -0) ➕ `app/assets/tray.ico` (+0 -0) ➕ `app/assets/tray_upgrade.ico` (+0 -0) ➕ `app/lifecycle/getstarted_nonwindows.go` (+9 -0) ➕ `app/lifecycle/getstarted_windows.go` (+44 -0) ➕ `app/lifecycle/lifecycle.go` (+92 -0) ➕ `app/lifecycle/logging.go` (+46 -0) ➕ `app/lifecycle/logging_nonwindows.go` (+9 -0) ➕ `app/lifecycle/logging_windows.go` (+19 -0) ➕ `app/lifecycle/paths.go` (+79 -0) ➕ `app/lifecycle/server.go` (+139 -0) ➕ `app/lifecycle/server_unix.go` (+12 -0) ➕ `app/lifecycle/server_windows.go` (+13 -0) ➕ `app/lifecycle/updater.go` (+238 -0) ➕ `app/lifecycle/updater_nonwindows.go` (+12 -0) _...and 77 more files_ </details> ### 📄 Description This is a very preliminary ~~implementation~~ hack of Vulkan support, which llama.cpp recently added. This is not intended to be merged. This code is far from there. I jsut want to get feedback from ollama devs and some pointers. I tested this on an Intel Iris Plus G7 GPU on Linux. Phi-2 works fine with 20%-50% speedup compared to CPU with VNNI enabled. It behaves incorrectly for multimodal models such as Bakllava, which I'm still debugging. I think I need to pull the latest llama.cpp commits to make it work properly, but updating the submodule is throwing bizarre compile time errors. --- <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 2026-04-12 23:16:05 -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#10934
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?