GitLab

Fix a rebase glitch from the old C++ runner build model

This adds support for the Jetson JetPack variants into the Go runner

* Remove llama.cpp submodule and shift new build to top

* CI: install msys and clang gcc on win

Needed for deepseek to work properly on windows

* Re-introduce the llama package

This PR brings back the llama package, making it possible to call llama.cpp and
ggml APIs from Go directly via CGo. This has a few advantages:

- C APIs can be called directly from Go without needing to use the previous
  "server" REST API
- On macOS and for CPU builds on Linux and Windows, Ollama can be built without
  a go generate ./... step, making it easy to get up and running to hack on
  parts of Ollama that don't require fast inference
- Faster build times for AVX,AVX2,CUDA and ROCM (a full build of all runners
  takes <5 min on a fast CPU)
- No git submodule making it easier to clone and build from source

This is a big PR, but much of it is vendor code except for:

- llama.go CGo bindings
- example/: a simple example of running inference
- runner/: a subprocess server designed to replace the llm/ext_server package
- Makefile an as minimal as possible Makefile to build the runner package for
  different...

* Optimize container images for startup

This change adjusts how to handle runner payloads to support
container builds where we keep them extracted in the filesystem.
This makes it easier to optimize the cpu/cuda vs cpu/rocm images for
size, and should result in faster startup times for container images.

* Refactor payload logic and add buildx support for faster builds

* Move payloads around

* Review comments

* Converge to buildx based helper scripts

* Use docker buildx action for release

* Quiet down dockers new lint warnings

Docker has recently added lint warnings to build.  This cleans up those warnings.

* Fix go lint regression

Signed-off-by: Xiaodong Ye <yeahdongcn@gmail.com>

We're over budget for github's maximum release artifact size with rocm + 2 cuda
versions.  This splits rocm back out as a discrete artifact, but keeps the layout so it can
be extracted into the same location as the main bundle.

Based on compute capability and driver version, pick
v12 or v11 cuda variants.

This adds new variants for arm64 specific to Jetson platforms

This should help speed things up a little

This adjusts linux to follow a similar model to windows with a discrete archive
(zip/tgz) to cary the primary executable, and dependent libraries. Runners are
still carried as payloads inside the main binary

Darwin retain the payload model where the go binary is fully self contained.

The centos 7 arm mirrors have disappeared due to the EOL 2 days
ago, and the vault sed workaround which works for x86 doesn't work for arm.

This should resolve a number of memory leak and stability defects by allowing
us to isolate llama.cpp in a separate process and shutdown when idle, and
gracefully restart if it has problems.  This also serves as a first step to be
able to run multiple copies to support multiple models concurrently.

This reverts commit 5dacc1eb.

We had started using rocky linux 8, but they've updated to GCC 10.3,
which breaks NVCC.  10.2 is compatible (or 10.4, but that's not
available from rocky linux 8 repos yet)

Flesh out our github actions CI so we can build official releaes.

This refines where we extract the LLM libraries to by adding a new
OLLAMA_HOME env var, that defaults to `~/.ollama` The logic was already
idempotenent, so this should speed up startups after the first time a
new release is deployed.  It also cleans up after itself.

We now build only a single ROCm version (latest major) on both windows
and linux.  Given the large size of ROCms tensor files, we split the
dependency out.  It's bundled into the installer on windows, and a
separate download on windows.  The linux install script is now smart and
detects the presence of AMD GPUs and looks to see if rocm v6 is already
present, and if not, then downloads our dependency tar file.

For Linux discovery, we now use sysfs and check each GPU against what
ROCm supports so we can degrade to CPU gracefully instead of having
llama.cpp+rocm assert/crash on us.  For Windows, we now use go's windows
dynamic library loading logic to access the amdhip64.dll APIs to query
the GPU information.

Without this env var, podman's GPU logic doesn't map the GPU through

This adds ROCm support back as a discrete image.

The size increase for rocm support in the standard image is problematic
We'll revisit multiple tags for rocm support in a follow up PR.

The linux build now support parallel CPU builds to speed things up.
This also exposes AMD GPU targets as an optional setting for advaced
users who want to alter our default set.

This renames Dockerfile.build to Dockerfile, and adds some new stages
to support 2 modes of building - the build_linux.sh script uses
intermediate stages to extract the artifacts for ./dist, and the default
build generates a container image usable by both cuda and rocm cards.
This required transitioniing the x86 base to the rocm image to avoid
layer bloat.