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.

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

ビルドの自動化 #85

Closed
Hiroshiba opened this issue Sep 11, 2021 · 1 comment
Closed

ビルドの自動化 #85

Hiroshiba opened this issue Sep 11, 2021 · 1 comment

Comments

@Hiroshiba
Copy link
Member

内容

VOICEVOXのコアライブラリが公開されました。 https://github.com/Hiroshiba/voicevox_core
これを用いれば、ビルドを自動化できるはずです。
(ちなみにエンジンのビルドは数十分かかります)

このタスクは非常に長くなると思われます。
ちょっとでも取り掛かった方がいたら、ぜひコメントで知見ややり方を共有して頂けると・・・!

@Hiroshiba
Copy link
Member Author

nuitkaビルドはすごく時間かかるけど、numpyとかのビルド結果をキャッシュすればかなり短くできるかも・・・?

Hiroshiba pushed a commit that referenced this issue Sep 21, 2021
* windows workflow: prepare cache

* windows workflow

* copy libtorch/lib

* fix core cache path

* fix site-packages path

* quote path

* skip unnecessary os check

* rm cudnn url from cpu matrix

* use bin dir to search cuda dlls

* install voicevox core python package

* fix version spec

* makelib

* cmd /C instead cmd

* call directly makelib.bat

* call bat with relative path

* install ccache

* rm voicevox core url dump file

* fix ccache_url ref

* copy core dll

* nuitka assume yes for download

* generic site-packages dir path

* extract cuda dll only (reduce disk usage)

* fix cuda dll path

* fix mkdir cuda dll dir

* setup-python outputs.python-version

* use windows-2019

* replace bashslash to slash in cp

* set nuitka cache dir env

* nuitka cache dir as env in step

* fix cuda path escaping

* use cuda slash path to rm

* use shell variable for cuda slash path

* fix command substitution

* fix quote

* fix command substitution

* use [] instead - in yaml

* fix step order (save temporal capacity)

* libtorch extract only dlls

* cudnn extract only dlls

* show disk space before nuitka build for debug

* fix shell name

* merge dlls with upload-artifact (save storage space)

* add disk space log

* disk space log as step

* fix cache step id (-dll-)

* use symlink to pack artifact

* fix env to create symlink

* ref matirx.device

* log symlinks

* fix windows nuitka cache key; rm target

* unquote asterisk

* create symlink instead of copy (Git Bash)

* use step env for MSYS env var

* overwrite symlink

* remove CUDA/cuDNN download (included in LibTorch)

* remove CUDA/cuDNN download commands (included in LibTorch)

* use only include for matrix

* fix libtorch step name

* reorder steps; libtorch former

* run build.yml on master push

* download openjtalk dic (with retry)

* fail workflow if openjtalk dic download fail

* follow core 0.6.0 (windows)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant