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 #6570] [CLOSED] llama: opt-in at build time #17425

New Issue
Closed
opened 2026-04-16 06:02:21 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-16 06:02:21 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/6570
Author: @dhiltgen
Created: 8/30/2024
Status: ❌ Closed

Base: jmorganca/llama ← Head: go_server

📝 Commits (10+)

📊 Changes

299 files changed (+165277 additions, -898 deletions)

View changed files

📝 .dockerignore (+2 -0)
📝 .gitattributes (+2 -0)
📝 .github/workflows/release.yaml (+180 -29)
📝 .github/workflows/test.yaml (+1 -42)
📝 .gitignore (+3 -1)
📝 Dockerfile (+91 -60)
➕ Dockerfile.new (+221 -0)
➕ build/darwin/amd64/placeholder (+1 -0)
➕ build/darwin/arm64/placeholder (+1 -0)
➕ build/embed_darwin_amd64.go (+8 -0)
➕ build/embed_darwin_arm64.go (+8 -0)
➕ build/embed_linux.go (+6 -0)
➕ build/embed_unused.go (+8 -0)
➕ build/linux/amd64/placeholder (+1 -0)
➕ build/linux/arm64/placeholder (+1 -0)
📝 envconfig/config.go (+0 -48)
➖ gpu/assets.go (+0 -148)
📝 gpu/gpu.go (+3 -4)
📝 integration/concurrency_test.go (+1 -1)
📝 integration/utils_test.go (+1 -1)

...and 80 more files

📄 Description

This PR layers on #6547 for the new Go server.

Unfortunately the sizes are too large to try to make the opt-in strategy work at runtime (the linux tgz would significantly exceed the 2G github artifact size limit) so this makes the opt-in strategy work at build time.

Notable refinements:

  • the ggml library is now moved out as a payload in the tar file to reduce the binary size, and the names are adjusted to avoid clashing between cuda v11, v12, and rocm.
  • The static cgo wiring for the main app is shifted over to the new llama package and the old go generate wiring for the static build is removed as no longer needed.
  • An initial foundation for requirement information is added to the runner so eventually we can pick compatible runners more easily
  • Use the CPU vector flags when compiling the GPU runners

I'm still working through verifying all the build stages, so I'll mark it draft for now until I confirm they're all correct.

🔄 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/6570 **Author:** [@dhiltgen](https://github.com/dhiltgen) **Created:** 8/30/2024 **Status:** ❌ Closed **Base:** `jmorganca/llama` ← **Head:** `go_server` --- ### 📝 Commits (10+) - [`adafbf9`](https://github.com/ollama/ollama/commit/adafbf908143dfa23c5ea28df889665e8969e54b) add sync of llama.cpp - [`c815358`](https://github.com/ollama/ollama/commit/c815358a8114e373c05e04bfbf62427a57555910) Initial `llama` Go module - [`8523233`](https://github.com/ollama/ollama/commit/8523233d55c3ba7b6605273407ca577ad45744a9) fix `.gitattributes` - [`211a8b0`](https://github.com/ollama/ollama/commit/211a8b02907c2b97a17694d473d22097f31c511b) Add missing hipcc flags - [`00a8a37`](https://github.com/ollama/ollama/commit/00a8a37bda419d341fca6659a5a77a6e8b25dbe5) Update README.md - [`9b11f27`](https://github.com/ollama/ollama/commit/9b11f272de71342ab1a7148b891f3495ba9ee04f) Update README.md - [`87868a9`](https://github.com/ollama/ollama/commit/87868a94188a5458d57f478363e876ba8b030db5) Update README.md - [`6b3e946`](https://github.com/ollama/ollama/commit/6b3e946f694c036033884064240306af302d0f0f) Update README.md - [`c08bb0b`](https://github.com/ollama/ollama/commit/c08bb0b1077e1c06a5f321fe2733596cc76f662a) rename `server` to `runner` - [`ad64db1`](https://github.com/ollama/ollama/commit/ad64db1a6385d1cf2b49e95633ca1a5c04683ffa) wip... ### 📊 Changes **299 files changed** (+165277 additions, -898 deletions) <details> <summary>View changed files</summary> 📝 `.dockerignore` (+2 -0) 📝 `.gitattributes` (+2 -0) 📝 `.github/workflows/release.yaml` (+180 -29) 📝 `.github/workflows/test.yaml` (+1 -42) 📝 `.gitignore` (+3 -1) 📝 `Dockerfile` (+91 -60) ➕ `Dockerfile.new` (+221 -0) ➕ `build/darwin/amd64/placeholder` (+1 -0) ➕ `build/darwin/arm64/placeholder` (+1 -0) ➕ `build/embed_darwin_amd64.go` (+8 -0) ➕ `build/embed_darwin_arm64.go` (+8 -0) ➕ `build/embed_linux.go` (+6 -0) ➕ `build/embed_unused.go` (+8 -0) ➕ `build/linux/amd64/placeholder` (+1 -0) ➕ `build/linux/arm64/placeholder` (+1 -0) 📝 `envconfig/config.go` (+0 -48) ➖ `gpu/assets.go` (+0 -148) 📝 `gpu/gpu.go` (+3 -4) 📝 `integration/concurrency_test.go` (+1 -1) 📝 `integration/utils_test.go` (+1 -1) _...and 80 more files_ </details> ### 📄 Description This PR layers on #6547 for the new Go server. Unfortunately the sizes are too large to try to make the opt-in strategy work at runtime (the linux tgz would significantly exceed the 2G github artifact size limit) so this makes the opt-in strategy work at build time. Notable refinements: - the ggml library is now moved out as a payload in the tar file to reduce the binary size, and the names are adjusted to avoid clashing between cuda v11, v12, and rocm. - The static cgo wiring for the main app is shifted over to the new llama package and the old `go generate` wiring for the static build is removed as no longer needed. - An initial foundation for requirement information is added to the runner so eventually we can pick compatible runners more easily - Use the CPU vector flags when compiling the GPU runners I'm still working through verifying all the build stages, so I'll mark it draft for now until I confirm they're all correct. --- <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 2026-04-16 06:02:21 -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#17425
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?