github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[PR #12277] install: break up windows and linux bundles #39637

New Issue
Open
opened 2026-04-23 00:37:15 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-23 00:37:15 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/12277
Author: @dhiltgen
Created: 9/12/2025
Status: 🔄 Open

Base: main ← Head: fine_grain_bundles

📝 Commits (2)

  • 63ee668 install: fine grain packaging
  • ccb7e56 install: update linux install script for new bundle layout

📊 Changes

5 files changed (+227 additions, -109 deletions)

View changed files

📝 docs/linux.md (+61 -20)
📝 docs/windows.md (+45 -15)
📝 scripts/build_linux.sh (+12 -4)
📝 scripts/build_windows.ps1 (+26 -20)
📝 scripts/install.sh (+83 -50)

📄 Description

Fixes #12268

Marking draft until I run further tests.

With the addition of CUDA v13, our base Windows and Linux bundles have grown in size. This change breaks down our binary artifacts into more pieces so the individual sizes are more manageable and provides the ability for users to download only what they need. Since the base bundle no longer contains any GPU libraries, I elected to rename it to make it more obvious the bundle lacked GPU content. The result is users with CPU only systems have a much smaller bundle size. Users with GPUs now can download only the parts they actually need.

This does add complexity to the manual install flows, so I expanded the Windows and Linux docs to be more complete.

I have not modified the container images, but that might be something to consider. If we do, perhaps the "latest" tag would map to a "full" image that has everything for simplicity, and then -cpu, -cuda-v12, -cuda-v13, and -rocm tags for stripped down images.

Before merging, the install script change should be split out into a separate PR to sequence with whatever release this lands in.

Windows

  • 14M ollama-windows-amd64-cpu.zip
  • 1019M ollama-windows-amd64-cuda-v12.zip
  • 618M ollama-windows-amd64-cuda-v13.zip
  • 373M ollama-windows-amd64-rocm.zip

Linux

  • 13M ollama-linux-amd64-cpu.tgz
  • 1.0G ollama-linux-amd64-cuda-v12.tgz
  • 635M ollama-linux-amd64-cuda-v13.tgz
  • 1.2G ollama-linux-amd64-rocm.tgz
  • 11M ollama-linux-arm64-cpu.tgz
  • 1.0G ollama-linux-arm64-cuda-v12.tgz
  • 748M ollama-linux-arm64-cuda-v13.tgz
  • 429M ollama-linux-arm64-jetpack5.tgz
  • 343M ollama-linux-arm64-jetpack6.tgz

🔄 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/12277 **Author:** [@dhiltgen](https://github.com/dhiltgen) **Created:** 9/12/2025 **Status:** 🔄 Open **Base:** `main` ← **Head:** `fine_grain_bundles` --- ### 📝 Commits (2) - [`63ee668`](https://github.com/ollama/ollama/commit/63ee668687bd0fabea6a99a6990844c0a6635f34) install: fine grain packaging - [`ccb7e56`](https://github.com/ollama/ollama/commit/ccb7e5678d8b7a508020fb784eee0aaab3c6a39c) install: update linux install script for new bundle layout ### 📊 Changes **5 files changed** (+227 additions, -109 deletions) <details> <summary>View changed files</summary> 📝 `docs/linux.md` (+61 -20) 📝 `docs/windows.md` (+45 -15) 📝 `scripts/build_linux.sh` (+12 -4) 📝 `scripts/build_windows.ps1` (+26 -20) 📝 `scripts/install.sh` (+83 -50) </details> ### 📄 Description Fixes #12268 Marking draft until I run further tests. With the addition of CUDA v13, our base Windows and Linux bundles have grown in size. This change breaks down our binary artifacts into more pieces so the individual sizes are more manageable and provides the ability for users to download only what they need. Since the base bundle no longer contains any GPU libraries, I elected to rename it to make it more obvious the bundle lacked GPU content. The result is users with CPU only systems have a much smaller bundle size. Users with GPUs now can download only the parts they actually need. This does add complexity to the manual install flows, so I expanded the Windows and Linux docs to be more complete. I have not modified the container images, but that might be something to consider. If we do, perhaps the "latest" tag would map to a "full" image that has everything for simplicity, and then `-cpu`, `-cuda-v12`, `-cuda-v13`, and `-rocm` tags for stripped down images. Before merging, the install script change should be split out into a separate PR to sequence with whatever release this lands in. ## Windows - 14M ollama-windows-amd64-cpu.zip - 1019M ollama-windows-amd64-cuda-v12.zip - 618M ollama-windows-amd64-cuda-v13.zip - 373M ollama-windows-amd64-rocm.zip ## Linux - 13M ollama-linux-amd64-cpu.tgz - 1.0G ollama-linux-amd64-cuda-v12.tgz - 635M ollama-linux-amd64-cuda-v13.tgz - 1.2G ollama-linux-amd64-rocm.tgz - 11M ollama-linux-arm64-cpu.tgz - 1.0G ollama-linux-arm64-cuda-v12.tgz - 748M ollama-linux-arm64-cuda-v13.tgz - 429M ollama-linux-arm64-jetpack5.tgz - 343M ollama-linux-arm64-jetpack6.tgz --- <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 2026-04-23 00:37:15 -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#39637
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?