# Conflicts: # studio/backend/core/inference/diffusion.py # studio/frontend/src/features/images/images-page.tsx
600 B
600 B
Continue answering the user’s latest matrix-coverage question: whether Mac/Windows/Linux/WSL × NVIDIA/AMD/CPU/Mac × many models is “fully optimized” versus ComfyUI/diffusers/sd.cpp. State plainly that only Linux x86_64 + NVIDIA B200 GPU and Linux CPU were measured; Mac, Windows, WSL, AMD, Intel, and Apple/MPS remain unvalidated here, with CUDA-only fp8/int8 fast quant as the key limitation. No tool is in flight; next action is to proceed from the completed hardware enumeration and outputs/coverage_matrix.md write, without re-running completed commands unless fresh evidence is needed.