Tag
6 articles
Learn to use NVIDIA's cuDNN Frontend Graph API to optimize scaled dot-product attention computations, enabling kernel fusion, autotuning, and plan reuse for enhanced deep learning performance.
This article explains CUDA Agent, a reinforcement learning system that uses large language models to generate optimized GPU kernels, outperforming traditional compilers in execution speed and efficiency.
French startup Kog is challenging the notion that GPUs are poorly suited for agentic AI workflows by developing innovative techniques to maximize GPU inference efficiency.
This article explains how large language models can be optimized to run locally on 24GB GPUs using quantization and architectural efficiency. It explores the technical strategies behind models like Qwen3.6, Mistral Small, and DeepSeek-R1-Distill.
Learn how fused kernels and automatic mixed precision (AMP) techniques, such as those in NVIDIA Apex and PyTorch's torch.amp, can dramatically accelerate transformer training by optimizing computational efficiency and reducing memory overhead.
This article explains how to implement NVIDIA's Transformer Engine with mixed-precision, FP8 support, benchmarking, and fallback execution for optimizing transformer model performance.
