Claude
46b9066bba
GPU tier table extended: ultra ≥16 GB → SDXL (unchanged) high 8-16 GB → SDXL (unchanged) medium 4-8 GB → SD 2.x (unchanged) legacy 2-4 GB → SD 1.5 (~1.7 GB fp16) ← new: GTX 970/1060/RX 580 etc. minimal <2 GB → SD 1.5 + sequential CPU offload ← new: very old/integrated GPUs gpu_detect.py: - Detects CUDA compute capability (CC); fp16 disabled for CC < 6.0 (pre-Pascal) - GpuInfo gains compute_capability and warnings fields - _make_warnings() emits human-readable warnings for low VRAM and old CC - model tier fallback updated from 'low' to 'legacy' local_diffusion.py: - minimal/legacy tiers use enable_sequential_cpu_offload() + enable_attention_slicing(1) - target resolution per tier: ultra/high=1024, medium=768, legacy/minimal=512 - .to(device) skipped when sequential CPU offload is active gpu_status.py: - Response now includes compute_capability and warnings docker-compose.gpu.yml: - Full nvidia-container-toolkit install instructions in header comment - nvidia-docker2 (legacy) fallback documented as comment block inline - AMD ROCm swap-in instructions added - GPU tier table documented in header scripts/gpu_setup.py: - Prints compute capability, fp16 status, tier, and model selection at startup - Prints per-tier warnings (old CC, low VRAM) https://claude.ai/code/session_01WVDg7amsy1TTtxvpku7bcM
77 lines
2.4 KiB
Python
77 lines
2.4 KiB
Python
"""
|
|
GPU status and model management endpoints.
|
|
All under /api/gpu prefix.
|
|
"""
|
|
from fastapi import APIRouter
|
|
from pydantic import BaseModel
|
|
from typing import Optional, List
|
|
import asyncio
|
|
|
|
router = APIRouter(prefix="/api/gpu", tags=["gpu"])
|
|
|
|
|
|
@router.get("/status")
|
|
async def gpu_status():
|
|
"""
|
|
Return GPU capabilities, VRAM, tier, and per-model download/ready state.
|
|
Frontend polls this to show GPU badge and tool availability.
|
|
"""
|
|
from app.services.gpu_detect import get_cached_gpu_info, get_model_ids
|
|
from app.services.local_diffusion import get_all_model_states
|
|
|
|
info = get_cached_gpu_info()
|
|
model_ids = get_model_ids(info.tier)
|
|
|
|
return {
|
|
"backend": info.backend,
|
|
"device_name": info.device_name,
|
|
"vram_gb": info.vram_gb,
|
|
"compute_capability": info.compute_capability,
|
|
"tier": info.tier,
|
|
"fp16": info.fp16,
|
|
"warnings": info.warnings,
|
|
"capabilities": info.capabilities,
|
|
"models": {
|
|
op: {"model_id": mid, "available": mid is not None}
|
|
for op, mid in model_ids.items()
|
|
},
|
|
"pipeline_states": get_all_model_states(),
|
|
}
|
|
|
|
|
|
class PrefetchRequest(BaseModel):
|
|
operations: Optional[List[str]] = None
|
|
|
|
|
|
@router.post("/prefetch")
|
|
async def prefetch_models(req: PrefetchRequest = PrefetchRequest()):
|
|
"""
|
|
Kick off background model downloads for the requested operations.
|
|
Returns immediately; poll /api/gpu/prefetch-status for progress.
|
|
Default: prefetch inpaint, txt2img, img2img.
|
|
"""
|
|
ops = req.operations or ["inpaint", "txt2img", "img2img"]
|
|
valid = {"inpaint", "txt2img", "img2img", "outpaint", "upscale"}
|
|
ops = [op for op in ops if op in valid]
|
|
|
|
from app.services.local_diffusion import get_local_diffusion_provider
|
|
provider = get_local_diffusion_provider()
|
|
|
|
async def _prefetch():
|
|
for op in ops:
|
|
try:
|
|
await provider._get_pipeline(op)
|
|
print(f"[gpu] Prefetch complete: {op}")
|
|
except Exception as exc:
|
|
print(f"[gpu] Prefetch failed for {op}: {exc}")
|
|
|
|
asyncio.create_task(_prefetch())
|
|
return {"status": "prefetch_started", "operations": ops}
|
|
|
|
|
|
@router.get("/prefetch-status")
|
|
async def prefetch_status():
|
|
"""Poll model download / load progress."""
|
|
from app.services.local_diffusion import get_all_model_states
|
|
return {"models": get_all_model_states()}
|