Joined May 2012
Fixed the 100% ComfyUI Float8 crash on base 16GB Mac Mini M4! We uses node isolation to route text encoders to CPU while keeping tensors on GPU. Because of Apple UMA, it achieves a zero-copy bypass with only +120ms latency. Watch & grab free JSONs: youtu.be
415
Devraj Gupta retweeted
Please subscribe to support me in YouTube channel @aiwithjsdev
1
13
Devraj Gupta retweeted
Can the base $599 Mac mini M4 (16GB) REALLY handle local AI? 🤔 I put ComfyUI Desktop to the ultimate stress test: running Flux, LTX Video, Hunyuan & Local LLMs. See if 16GB unified memory is actually enough, or if it crashes! 👇 ▶️ youtu.be/1jmyfu2vguQ #MacMiniM4 #ComfyUI
2
1
2
224
Devraj Gupta retweeted
Can a $599 16GB Mac Mini M4 actually run Wan 2.1 or MiniMax H3 locally without crashing? Let's put it to the test and see how it handles ComfyUI workflows! 💻🔥 Check out the full video here: youtube.com/shorts/Sxflta-vE…
1
200
Devraj Gupta retweeted
ComfyUI crashing on your M4 Mac with MPS errors? The MPS backend chokes on the text encoder format. Fix: Route just that text node to the CPU. Unified memory handles the handoff instantly back to the GPU. Watch the fix: youtube.com/shorts/DcNAZ4xiD…
1
36