Engineers can use similar frameworks to port their existing CUDA optimizations to new chips.
They do not need to rediscover the same low-level improvements from scratch.
Source: http://bair.berkeley.edu/blog/2026/07/29/cuda-to-mlx-k-search/
Loading
Research1 min read
Researchers built K-Search to translate existing CUDA kernels into optimized MLX code for Apple chips. The system achieved near-expert performance levels without manual rewriting.
Why it matters: Transferring decades of NVIDIA kernel knowledge to new hardware saves engineers time and improves model speed on local devices.
By OpenSmartRoute editorial · written through the router by llm-onprem
From Berkeley AI Research - “From CUDA to MLX: How K-Search Brings Decades of Kernel Expertise to Apple Silicon”
Engineers can use similar frameworks to port their existing CUDA optimizations to new chips.
They do not need to rediscover the same low-level improvements from scratch.
Source: http://bair.berkeley.edu/blog/2026/07/29/cuda-to-mlx-k-search/
Keep reading
Research1 min read
Google Research released over 30 AI-generated interactive simulations for STEM subjects. Teachers can create custom guided simulations using generative user interfaces.
AI1 min read
Hacktron AI used a new version of Claude to breach OpenAI's security.
editorial-writerllm-onpremollama-qwen3-5-4b/qwen3.5:4beditorial-writerDrafted automatically from the source and reviewed automatically; it may contain errors. Refer to the source for the authoritative text. How routing works
AI labs use more AI to track agent errors. Some companies prefer human logs instead.