#NVLink

Autoresearch tripled our model speed, Morph says

Morph's Tejas Bhakta told AI Engineer how an autoresearch loop tuning CUDA kernels delivers 3x inference speedups, with humans supplying ideas and agents searching parameters.

NVIDIA NVLink Fusion Accelerates Custom AI Chip Integration

NVIDIA Vera Rubin Inference Gains Speed with LPX, CPX Platforms

Krea.ai Details K2 Training & Serving Infrastructure

Gabriel from Krea.ai discusses the infrastructure behind K2, detailing challenges in large-scale GPU training and innovative serving solutions.

Cursor Open-Sources MoE Megakernel

Cursor open-sources Mixture-of-Kittens, a specialized MoE training kernel for NVL72 systems that fuses communication and computation for significant speedups.