Daniel Hiltgen d88c527be3 Build multiple CPU variants and pick the best

This reduces the built-in linux version to not use any vector extensions
which enables the resulting builds to run under Rosetta on MacOS in
Docker.  Then at runtime it checks for the actual CPU vector
extensions and loads the best CPU library available

2024-01-11 08:42:47 -08:00

3.4 KiB

Raw Blame History

Development

Install cmake or (optionally, required tools for GPUs)
run go generate ./...
run go build .

Install required tools:

cmake version 3.24 or higher
go version 1.20 or higher
gcc version 11.4.0 or higher

brew install go cmake gcc

Optionally enable debugging and more verbose logging:

export CGO_CFLAGS="-g"

Get the required libraries and build the native LLM code:

go generate ./...

Then build ollama:

go build .

Now you can run ollama:

./ollama

Linux

Linux CUDA (NVIDIA)

Your operating system distribution may already have packages for NVIDIA CUDA. Distro packages are often preferable, but instructions are distro-specific. Please consult distro-specific docs for dependencies if available!

Install cmake and golang as well as NVIDIA CUDA development and runtime packages. Then generate dependencies:

go generate ./...

Then build the binary:

go build .

Linux ROCm (AMD)

Your operating system distribution may already have packages for AMD ROCm and CLBlast. Distro packages are often preferable, but instructions are distro-specific. Please consult distro-specific docs for dependencies if available!

Install CLBlast and ROCm developement packages first, as well as cmake and golang. Adjust the paths below (correct for Arch) as appropriate for your distributions install locations and generate dependencies:

CLBlast_DIR=/usr/lib/cmake/CLBlast ROCM_PATH=/opt/rocm go generate ./...

Then build the binary:

go build .

ROCm requires elevated privileges to access the GPU at runtime. On most distros you can add your user account to the render group, or run as root.

Advanced CPU Settings

By default, running go generate ./... will compile a few different variations of the LLM library based on common CPU families and vector math capabilities, including a lowest-common-denominator which should run on almost any 64 bit CPU somewhat slowly. At runtime, Ollama will auto-detect the optimal variation to load. If you would like to build a CPU-based build customized for your processor, you can set OLLAMA_CUSTOM_CPU_DEFS to the llama.cpp flags you would like to use. For example, to compile an optimized binary for an Intel i9-9880H, you might use:

OLLAMA_CUSTOM_CPU_DEFS="-DLLAMA_AVX=on -DLLAMA_AVX2=on -DLLAMA_F16C=on -DLLAMA_FMA=on" go generate ./...
go build .

Containerized Linux Build

If you have Docker available, you can build linux binaries with ./scripts/build_linux.sh which has the CUDA and ROCm dependencies included. The resulting binary is placed in ./dist

Windows

Note: The windows build for Ollama is still under development.