Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add CUDA hardware acceleration for textgen #70

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Add CUDA hardware acceleration for textgen #70

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
45 changes: 23 additions & 22 deletions packages/llama-cpp-python/default.nix
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,14 +1,16 @@
{ buildPythonPackage, fetchFromGitHub, lib, stdenv, darwin, cmake, ninja, pathspec, poetry-core, pyproject-metadata, scikit-build-core, setuptools, diskcache, numpy, typing-extensions }:
let
inherit (stdenv) isDarwin;
osSpecific = with darwin.apple_sdk.frameworks; if isDarwin then [ Accelerate CoreGraphics CoreVideo ] else [ ];
llama-cpp-pin = fetchFromGitHub {
owner = "ggerganov";
repo = "llama.cpp";
rev = "a98b1633d5a94d0aa84c7c16e1f8df5ac21fc850";
hash = "sha256-HNwyPJXsUL41zLA+90Yu7kCpihW0HBOeW2jDs8sw7qs=";
};
in
{ buildPythonPackage
, fetchFromGitHub
, lib
, cudaPackages
, cmake
, ninja
, pathspec
, pyproject-metadata
, scikit-build-core
, diskcache
, numpy
, typing-extensions
}:
buildPythonPackage rec {
pname = "llama-cpp-python";
version = "0.2.7";
Expand All @@ -18,26 +20,25 @@ buildPythonPackage rec {
owner = "abetlen";
repo = pname;
rev = "refs/tags/v${version}";
hash = "sha256-jL2jVTKwmTx6pSnoN5n4NtQ3hs3weXiQTKFQdjL172U=";
hash = "sha256-2uPWH8ik/YznJTNBCopz58YjDJ7i1l9hgp8t0Nwjm5Q=";
fetchSubmodules = true;
};

preConfigure = ''
cp -r ${llama-cpp-pin}/. ./vendor/llama.cpp
chmod -R +w ./vendor/llama.cpp
'';
preBuild = ''
cd ..
'';
buildInputs = osSpecific;
dontUseCmakeConfigure = true;
SKBUILD_CMAKE_ARGS = lib.strings.concatStringsSep ";" [
"-DLLAMA_CUBLAS=on"
];

buildInputs = [
cudaPackages.cudatoolkit
];

nativeBuildInputs = [
cmake
ninja
pathspec
poetry-core
pyproject-metadata
scikit-build-core
setuptools
];

propagatedBuildInputs = [
Expand Down