Researchers baked 3x inference speedups directly into LLM weights — without speculative decoding

Researchers from the University of Maryland, Lawrence Livermore, Columbia and TogetherAI have developed a training technique that triples LLM inference speed without auxiliary models or infrastructure ...
CNX Software

Rockchip RK3588 and RK3576 H.264 and H.265 video decoders gain mainline Linux support

Rockchip RK3588 and RK3576 get mainline Linux support for H.264 (AVC) and H.265 (HEVC) hardware video coders thanks to work from Collabora.
The Lancet

Decoding the language of sleep with artificial intelligence

Sleep is one of medicine's underused data streams. Clinically, disturbed sleep has often been treated as a symptom of a disorder, but sleep is also a physiological state in which brain, cardiac, ...
IEEE

A Reconfigurable LDPC Decoder for NAND Flash Supporting Codeword Lengths Up to 4kB

Abstract: This work presents a reconfigurable quasi-cyclic low-density parity-check (QC-LDPC) decoder supporting NAND flash memories with codeword lengths up to 4 kB. The proposed design provides ...
IEEE

LDPC Decoding via Processing-in-Memory for Mobile and Wireless Communications

Abstract: Low-density parity-check (LDPC) codes are widely used in modern communication systems, including 5G networks, due to their near-capacity error correction ...