mirror of
https://github.com/jamiepine/voicebox.git
synced 2026-09-26 13:45:16 -07:00
feat: CUDA backend swap via binary download and restart
Add the ability to download a CUDA-enabled backend binary (~2.4 GB) and swap it in via a backend-only restart, solving the #1 user pain point (19 open 'GPU not detected' issues caused by GitHub's 2 GB asset limit). Backend: - cuda_download.py: download from R2 (primary) or GitHub split-parts (fallback), SHA-256 verification, atomic writes, progress via SSE - 4 new endpoints: GET/POST/DELETE /backend/cuda-*, GET cuda-progress - server.py: --version flag, auto-detect variant from binary name - build_binary.py: --cuda flag for CUDA PyInstaller builds - split_binary.py: split large binaries into <2GB GitHub Release assets - CI workflow for building CUDA binary Tauri: - restart_server command (stop -> wait -> start) - start_server prefers CUDA binary from {data_dir}/backends/ if present - Version mismatch check: runs --version before launching CUDA binary Frontend: - GpuAcceleration component: download, progress, restart, switch, delete - API client + types for CUDA status and management - Platform lifecycle: restartServer() on Tauri/Web - Aggressive 1s health polling during restart for fast reconnection
This commit is contained in:
James Pine
+36
-8
@@ -1,8 +1,13 @@
|
||||
"""
|
||||
PyInstaller build script for creating standalone Python server binary.
|
||||
|
||||
Usage:
|
||||
python build_binary.py # Build default (CPU) server binary
|
||||
python build_binary.py --cuda # Build CUDA-enabled server binary
|
||||
"""
|
||||
|
||||
import PyInstaller.__main__
|
||||
import argparse
|
||||
import os
|
||||
import platform
|
||||
from pathlib import Path
|
||||
@@ -13,15 +18,22 @@ def is_apple_silicon():
|
||||
return platform.system() == "Darwin" and platform.machine() == "arm64"
|
||||
|
||||
|
||||
def build_server():
|
||||
"""Build Python server as standalone binary."""
|
||||
def build_server(cuda=False):
|
||||
"""Build Python server as standalone binary.
|
||||
|
||||
Args:
|
||||
cuda: If True, build with CUDA support and name the binary
|
||||
voicebox-server-cuda instead of voicebox-server.
|
||||
"""
|
||||
backend_dir = Path(__file__).parent
|
||||
|
||||
binary_name = 'voicebox-server-cuda' if cuda else 'voicebox-server'
|
||||
|
||||
# PyInstaller arguments
|
||||
args = [
|
||||
'server.py', # Use server.py as entry point instead of main.py
|
||||
'--onefile',
|
||||
'--name', 'voicebox-server',
|
||||
'--name', binary_name,
|
||||
]
|
||||
|
||||
# Add local qwen_tts path if specified (for editable installs)
|
||||
@@ -49,6 +61,7 @@ def build_server():
|
||||
'--hidden-import', 'backend.utils.progress',
|
||||
'--hidden-import', 'backend.utils.hf_progress',
|
||||
'--hidden-import', 'backend.utils.validation',
|
||||
'--hidden-import', 'backend.cuda_download',
|
||||
'--hidden-import', 'torch',
|
||||
'--hidden-import', 'transformers',
|
||||
'--hidden-import', 'fastapi',
|
||||
@@ -70,8 +83,16 @@ def build_server():
|
||||
'--collect-submodules', 'jaraco',
|
||||
])
|
||||
|
||||
# Add MLX-specific imports if building on Apple Silicon
|
||||
if is_apple_silicon():
|
||||
# Add CUDA-specific hidden imports
|
||||
if cuda:
|
||||
print("Building with CUDA support")
|
||||
args.extend([
|
||||
'--hidden-import', 'torch.cuda',
|
||||
'--hidden-import', 'torch.backends.cudnn',
|
||||
])
|
||||
|
||||
# Add MLX-specific imports if building on Apple Silicon (never for CUDA builds)
|
||||
if is_apple_silicon() and not cuda:
|
||||
print("Building for Apple Silicon - including MLX dependencies")
|
||||
args.extend([
|
||||
'--hidden-import', 'backend.backends.mlx_backend',
|
||||
@@ -91,7 +112,7 @@ def build_server():
|
||||
'--collect-all', 'mlx',
|
||||
'--collect-all', 'mlx_audio',
|
||||
])
|
||||
else:
|
||||
elif not cuda:
|
||||
print("Building for non-Apple Silicon platform - PyTorch only")
|
||||
|
||||
args.extend([
|
||||
@@ -105,8 +126,15 @@ def build_server():
|
||||
# Run PyInstaller
|
||||
PyInstaller.__main__.run(args)
|
||||
|
||||
print(f"Binary built in {backend_dir / 'dist' / 'voicebox-server'}")
|
||||
print(f"Binary built in {backend_dir / 'dist' / binary_name}")
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
build_server()
|
||||
parser = argparse.ArgumentParser(description="Build voicebox-server binary")
|
||||
parser.add_argument(
|
||||
'--cuda',
|
||||
action='store_true',
|
||||
help="Build CUDA-enabled binary (voicebox-server-cuda)",
|
||||
)
|
||||
cli_args = parser.parse_args()
|
||||
build_server(cuda=cli_args.cuda)
|
||||
|
||||
@@ -0,0 +1,200 @@
|
||||
"""
|
||||
CUDA backend binary download, assembly, and verification.
|
||||
|
||||
Downloads split parts of the CUDA-enabled voicebox-server binary from
|
||||
GitHub Releases, reassembles them, verifies integrity via SHA-256,
|
||||
and places the binary in the app's data directory for use on next
|
||||
backend restart.
|
||||
"""
|
||||
|
||||
import hashlib
|
||||
import logging
|
||||
import os
|
||||
import sys
|
||||
from pathlib import Path
|
||||
from typing import Optional
|
||||
|
||||
from .config import get_data_dir
|
||||
from .utils.progress import get_progress_manager
|
||||
from . import __version__
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
GITHUB_RELEASES_URL = "https://github.com/jamiepine/voicebox/releases/download"
|
||||
|
||||
PROGRESS_KEY = "cuda-backend"
|
||||
|
||||
|
||||
def get_backends_dir() -> Path:
|
||||
"""Directory where downloaded backend binaries are stored."""
|
||||
d = get_data_dir() / "backends"
|
||||
d.mkdir(parents=True, exist_ok=True)
|
||||
return d
|
||||
|
||||
|
||||
def get_cuda_binary_name() -> str:
|
||||
"""Platform-specific CUDA binary filename."""
|
||||
if sys.platform == "win32":
|
||||
return "voicebox-server-cuda.exe"
|
||||
return "voicebox-server-cuda"
|
||||
|
||||
|
||||
def get_cuda_binary_path() -> Optional[Path]:
|
||||
"""Return path to CUDA binary if it exists."""
|
||||
p = get_backends_dir() / get_cuda_binary_name()
|
||||
if p.exists():
|
||||
return p
|
||||
return None
|
||||
|
||||
|
||||
def is_cuda_active() -> bool:
|
||||
"""Check if the current process is the CUDA binary.
|
||||
|
||||
The CUDA binary sets this env var on startup (see server.py).
|
||||
"""
|
||||
return os.environ.get("VOICEBOX_BACKEND_VARIANT") == "cuda"
|
||||
|
||||
|
||||
def get_cuda_status() -> dict:
|
||||
"""Get current CUDA backend status for the API."""
|
||||
progress_manager = get_progress_manager()
|
||||
cuda_path = get_cuda_binary_path()
|
||||
progress = progress_manager.get_progress(PROGRESS_KEY)
|
||||
|
||||
return {
|
||||
"available": cuda_path is not None,
|
||||
"active": is_cuda_active(),
|
||||
"binary_path": str(cuda_path) if cuda_path else None,
|
||||
"downloading": progress is not None and progress.get("status") == "downloading",
|
||||
"download_progress": progress,
|
||||
}
|
||||
|
||||
|
||||
async def download_cuda_binary(version: Optional[str] = None):
|
||||
"""Download the CUDA backend binary from GitHub Releases.
|
||||
|
||||
Downloads split parts listed in a manifest file, concatenates them,
|
||||
and verifies the SHA-256 checksum for integrity. Atomic write
|
||||
(temp file -> rename).
|
||||
|
||||
Args:
|
||||
version: Version tag (e.g. "v0.2.0"). Defaults to current app version.
|
||||
"""
|
||||
import httpx
|
||||
|
||||
if version is None:
|
||||
version = f"v{__version__}"
|
||||
|
||||
progress = get_progress_manager()
|
||||
binary_name = get_cuda_binary_name()
|
||||
dest_dir = get_backends_dir()
|
||||
final_path = dest_dir / binary_name
|
||||
temp_path = dest_dir / f"{binary_name}.download"
|
||||
|
||||
# Clean up any leftover partial download
|
||||
if temp_path.exists():
|
||||
temp_path.unlink()
|
||||
|
||||
logger.info(f"Starting CUDA backend download for {version}")
|
||||
progress.update_progress(
|
||||
PROGRESS_KEY, current=0, total=0,
|
||||
filename="Fetching manifest...", status="downloading",
|
||||
)
|
||||
|
||||
base_url = f"{GITHUB_RELEASES_URL}/{version}"
|
||||
stem = Path(binary_name).stem # voicebox-server-cuda
|
||||
|
||||
try:
|
||||
async with httpx.AsyncClient(follow_redirects=True, timeout=30.0) as client:
|
||||
# Fetch the manifest (list of split part filenames)
|
||||
manifest_url = f"{base_url}/{stem}.manifest"
|
||||
manifest_resp = await client.get(manifest_url)
|
||||
manifest_resp.raise_for_status()
|
||||
parts = [p.strip() for p in manifest_resp.text.strip().splitlines() if p.strip()]
|
||||
|
||||
if not parts:
|
||||
raise ValueError("Empty manifest — no split parts found")
|
||||
|
||||
logger.info(f"Found {len(parts)} split parts to download")
|
||||
|
||||
# Fetch expected checksum (optional — for integrity verification)
|
||||
expected_sha = None
|
||||
try:
|
||||
sha_url = f"{base_url}/{stem}.sha256"
|
||||
sha_resp = await client.get(sha_url)
|
||||
if sha_resp.status_code == 200:
|
||||
# Format: "sha256hex filename\n"
|
||||
expected_sha = sha_resp.text.strip().split()[0]
|
||||
logger.info(f"Expected SHA-256: {expected_sha[:16]}...")
|
||||
except Exception:
|
||||
logger.warning("Could not fetch checksum file — skipping verification")
|
||||
|
||||
# Download and concatenate parts
|
||||
total_downloaded = 0
|
||||
with open(temp_path, "wb") as f:
|
||||
for i, part_name in enumerate(parts):
|
||||
part_url = f"{base_url}/{part_name}"
|
||||
logger.info(f"Downloading part {i + 1}/{len(parts)}: {part_name}")
|
||||
|
||||
async with client.stream("GET", part_url) as response:
|
||||
response.raise_for_status()
|
||||
async for chunk in response.aiter_bytes(chunk_size=1024 * 1024):
|
||||
f.write(chunk)
|
||||
total_downloaded += len(chunk)
|
||||
progress.update_progress(
|
||||
PROGRESS_KEY, current=total_downloaded, total=0,
|
||||
filename=f"Part {i + 1}/{len(parts)}",
|
||||
status="downloading",
|
||||
)
|
||||
|
||||
# Verify integrity if checksum was available
|
||||
if expected_sha:
|
||||
progress.update_progress(
|
||||
PROGRESS_KEY, current=total_downloaded, total=total_downloaded,
|
||||
filename="Verifying integrity...", status="downloading",
|
||||
)
|
||||
sha256 = hashlib.sha256()
|
||||
with open(temp_path, "rb") as f:
|
||||
while True:
|
||||
chunk = f.read(1024 * 1024)
|
||||
if not chunk:
|
||||
break
|
||||
sha256.update(chunk)
|
||||
|
||||
actual = sha256.hexdigest()
|
||||
if actual != expected_sha:
|
||||
raise ValueError(
|
||||
f"Integrity check failed: expected {expected_sha[:16]}..., "
|
||||
f"got {actual[:16]}..."
|
||||
)
|
||||
logger.info(f"Integrity verified: {actual[:16]}...")
|
||||
|
||||
# Atomic move into place
|
||||
if final_path.exists():
|
||||
final_path.unlink()
|
||||
temp_path.rename(final_path)
|
||||
|
||||
# Make executable on Unix
|
||||
if sys.platform != "win32":
|
||||
final_path.chmod(0o755)
|
||||
|
||||
logger.info(f"CUDA backend downloaded to {final_path}")
|
||||
progress.mark_complete(PROGRESS_KEY)
|
||||
|
||||
except Exception as e:
|
||||
# Clean up on failure
|
||||
if temp_path.exists():
|
||||
temp_path.unlink()
|
||||
logger.error(f"CUDA backend download failed: {e}")
|
||||
progress.mark_error(PROGRESS_KEY, str(e))
|
||||
raise
|
||||
|
||||
|
||||
async def delete_cuda_binary() -> bool:
|
||||
"""Delete the downloaded CUDA binary. Returns True if deleted."""
|
||||
path = get_cuda_binary_path()
|
||||
if path and path.exists():
|
||||
path.unlink()
|
||||
logger.info(f"Deleted CUDA binary: {path}")
|
||||
return True
|
||||
return False
|
||||
@@ -206,6 +206,7 @@ async def health():
|
||||
gpu_type=gpu_type,
|
||||
vram_used_mb=vram_used,
|
||||
backend_type=backend_type,
|
||||
backend_variant=os.environ.get("VOICEBOX_BACKEND_VARIANT", "cpu"),
|
||||
)
|
||||
|
||||
|
||||
@@ -1797,6 +1798,75 @@ async def get_active_tasks():
|
||||
)
|
||||
|
||||
|
||||
# ============================================
|
||||
# CUDA BACKEND MANAGEMENT
|
||||
# ============================================
|
||||
|
||||
@app.get("/backend/cuda-status")
|
||||
async def get_cuda_status():
|
||||
"""Get CUDA backend download/availability status."""
|
||||
from . import cuda_download
|
||||
return cuda_download.get_cuda_status()
|
||||
|
||||
|
||||
@app.post("/backend/download-cuda")
|
||||
async def download_cuda_backend():
|
||||
"""Download the CUDA backend binary. Returns immediately; track progress via SSE."""
|
||||
from . import cuda_download
|
||||
|
||||
# Check if already downloaded
|
||||
if cuda_download.get_cuda_binary_path() is not None:
|
||||
raise HTTPException(status_code=409, detail="CUDA backend already downloaded")
|
||||
|
||||
async def _download():
|
||||
try:
|
||||
await cuda_download.download_cuda_binary()
|
||||
except Exception as e:
|
||||
import logging
|
||||
logging.getLogger(__name__).error(f"CUDA download failed: {e}")
|
||||
|
||||
asyncio.create_task(_download())
|
||||
return {"message": "CUDA backend download started", "progress_key": "cuda-backend"}
|
||||
|
||||
|
||||
@app.delete("/backend/cuda")
|
||||
async def delete_cuda_backend():
|
||||
"""Delete the downloaded CUDA backend binary."""
|
||||
from . import cuda_download
|
||||
|
||||
if cuda_download.is_cuda_active():
|
||||
raise HTTPException(
|
||||
status_code=409,
|
||||
detail="Cannot delete CUDA backend while it is active. Switch to CPU first.",
|
||||
)
|
||||
|
||||
deleted = await cuda_download.delete_cuda_binary()
|
||||
if not deleted:
|
||||
raise HTTPException(status_code=404, detail="No CUDA backend found to delete")
|
||||
|
||||
return {"message": "CUDA backend deleted"}
|
||||
|
||||
|
||||
@app.get("/backend/cuda-progress")
|
||||
async def get_cuda_download_progress():
|
||||
"""Get CUDA backend download progress via Server-Sent Events."""
|
||||
progress_manager = get_progress_manager()
|
||||
|
||||
async def event_generator():
|
||||
async for event in progress_manager.subscribe("cuda-backend"):
|
||||
yield event
|
||||
|
||||
return StreamingResponse(
|
||||
event_generator(),
|
||||
media_type="text/event-stream",
|
||||
headers={
|
||||
"Cache-Control": "no-cache",
|
||||
"Connection": "keep-alive",
|
||||
"X-Accel-Buffering": "no",
|
||||
},
|
||||
)
|
||||
|
||||
|
||||
# ============================================
|
||||
# STARTUP & SHUTDOWN
|
||||
# ============================================
|
||||
|
||||
@@ -127,6 +127,7 @@ class HealthResponse(BaseModel):
|
||||
gpu_type: Optional[str] = None # GPU type (CUDA, MPS, or None)
|
||||
vram_used_mb: Optional[float] = None
|
||||
backend_type: Optional[str] = None # Backend type (mlx or pytorch)
|
||||
backend_variant: Optional[str] = None # Binary variant (cpu or cuda)
|
||||
|
||||
|
||||
class ModelStatus(BaseModel):
|
||||
|
||||
@@ -20,6 +20,9 @@ soundfile>=0.12.0
|
||||
numpy>=1.24.0
|
||||
numba>=0.60.0,<0.61.0
|
||||
|
||||
# HTTP client (for CUDA backend download)
|
||||
httpx>=0.27.0
|
||||
|
||||
# Utilities
|
||||
python-multipart>=0.0.6
|
||||
Pillow>=10.0.0
|
||||
|
||||
@@ -64,7 +64,29 @@ if __name__ == "__main__":
|
||||
default=None,
|
||||
help="Data directory for database, profiles, and generated audio",
|
||||
)
|
||||
parser.add_argument(
|
||||
"--version",
|
||||
action="store_true",
|
||||
help="Print version and exit",
|
||||
)
|
||||
args = parser.parse_args()
|
||||
|
||||
if args.version:
|
||||
from backend import __version__
|
||||
print(f"voicebox-server {__version__}")
|
||||
sys.exit(0)
|
||||
|
||||
# Detect backend variant from binary name
|
||||
# voicebox-server-cuda → sets VOICEBOX_BACKEND_VARIANT=cuda
|
||||
import os
|
||||
binary_name = os.path.basename(sys.executable).lower()
|
||||
if "cuda" in binary_name:
|
||||
os.environ["VOICEBOX_BACKEND_VARIANT"] = "cuda"
|
||||
logger.info("Backend variant: CUDA")
|
||||
else:
|
||||
os.environ.setdefault("VOICEBOX_BACKEND_VARIANT", "cpu")
|
||||
logger.info(f"Backend variant: {os.environ['VOICEBOX_BACKEND_VARIANT']}")
|
||||
|
||||
logger.info(f"Parsed arguments: host={args.host}, port={args.port}, data_dir={args.data_dir}")
|
||||
|
||||
# Set data directory if provided
|
||||
|
||||
Reference in New Issue
Block a user