Articles by tag: gpu-optimization

NVIDIA CUDA cores accelerated by AI by 38% without human involvement

Multi-agent Cursor system autonomously optimized 235 CUDA cores for Blackwell B200. Results, methodology, and significance for HPC developers.

GPU acceleration of the Game of Life: CUDA Triton benchmarks

Comparison of PyTorch, CUDA, Triton for the cellular automaton "Game of Life" on A40. Results 22.5 ms, kernel code. Study optimizations for memory-bound tasks. Read the analysis.

vLLM Production Stack: basic launch in K8s

Deployment of vLLM in Kubernetes with Production Stack. GPU/CPU settings, tool-calling, multimodal models. YAML configurations and memory optimizations for middle/senior devops.

MLOps on free GPUs: processing PDF archives

How to process 9 GB of PDF archives on T4 GPU with 16 GB RAM. ZeroGPU, PyMuPDF, variance filtering, memory management. Practical guide.