Labyricorn/voicebox
1
0
Fork 0
mirror of https://github.com/jamiepine/voicebox.git synced 2026-09-15 12:50:42 -07:00
Code Issues Packages Projects Releases Wiki Activity

feat: split CUDA backend into independently versioned server + libs archives

Browse Source 
Switch CUDA builds from PyInstaller --onefile to --onedir and split the
output into two separately versioned archives:

1. Server core (~200-400MB) — versioned with the app, redownloaded on
   every app update
2. CUDA libs (~2GB) — versioned independently (cu126-v1), only
   redownloaded when the CUDA toolkit or torch version changes

This eliminates the ~2.4GB full redownload on every version bump.
After initial setup, most app updates only need ~200-400MB.

Closes #297
This commit is contained in:
James Pine
2026-03-17 04:04:17 -07:00
parent 2c1ee94891
commit 564d787927
6 changed files with 511 additions and 142 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/split_binary.py
+7 -1
View File
@@ -1,6 +1,12 @@
"""
Split a large binary into chunks for GitHub Releases (<2 GB each).
DEPRECATED: For CUDA builds, use scripts/package_cuda.py instead.
This script was used when the CUDA binary was built with --onefile and
needed to be split into parts for the 2GB GitHub Release asset limit.
With the switch to --onedir + dual archives (server core + CUDA libs),
package_cuda.py handles the packaging.
Usage:
python scripts/split_binary.py backend/dist/voicebox-server-cuda.exe
python scripts/split_binary.py backend/dist/voicebox-server-cuda.exe --chunk-size 1900000000
@@ -34,7 +40,7 @@ def split(input_path: Path, chunk_size: int, output_dir: Path):
part_index = len(parts)
part_name = f"{input_path.stem}.part{part_index:02d}{input_path.suffix}"
part_path = output_dir / part_name
part_path.write_bytes(data[i:i + chunk_size])
part_path.write_bytes(data[i : i + chunk_size])
parts.append(part_name)
# Write manifest (ordered list of part filenames)