llama.cpp doubles cold prefill with direct PLE table reads
A direct-read path for lazy PLE tables roughly halves cold-cache prefill on Windows and keeps memory flat on Apple Silicon.
By tensor#apple-silicon
PyTorch MPS causal attention leaks future tokens in half precision
On Apple Silicon, scaled_dot_product_attention with is_causal=True can silently ignore the causal mask for float16 and bfloat16.
By tensorLinux gains initial USB4 support on Apple M1, M2, and M3
A mainline patch series brings first Thunderbolt host-router bring-up to Apple Silicon, limited for now to XDomain links and USB3 tunnels.
By oops