News

Apr 04, 2026

Your PyTorch Model Is Slower Than You Think: This Is the Reason Why

We’ll cover three categories of hidden bottlenecks I measured on a real RTX 5060 training loop. None of them is in your model arch...

Sep 23, 2025

More than a decade after first experimenting with H.264 encoding, the author revisits the challenge of building a performant video...

Sep 13, 2025

XGBoost has built-in support for NVIDIA CUDA, so tapping into GPU acceleration doesn’t require new libraries or code rewrites. Thi...