Oxide and Friends cover image

Heterogeneous Computing with Raja Koduri

Oxide and Friends

00:00

Exploring Memory Intensive Workloads and GPU Efficiency

Explore the significance of memory in LLM workloads, comparing GPU architectures like Nvidia, AMD, and Intel using transport analogies. The discussion highlights the challenges of achieving performance across hardware variations due to implicit biases in data and compute packing.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app