hackertube.io

Trends in Deep Learning Hardware: Bill Dally (NVIDIA)

Paul G. Allen School · lecture · Watch on YouTube

Computer architecture & CPUsMachine learning & deep learningUniversity lectures & courses

What it is

Bill Dally of NVIDIA gives a University of Washington lecture on trends in deep learning hardware. It covers numeric formats, specialized instructions, energy per operation and co-design of hardware and software.

Why Hacker News recommends it

HN commenters cite it for NVIDIA's perspective on AI hardware efficiency, including logarithmic number formats and energy use.

  • 3 people recommended it
  • 2 different months
  • 0 strong recommendations
  • first 2023, last 2024

What HN commenters said

  1. “If you see the presentations by Nvidia's Bill Daly [1] it shows they've been evolving hardware and software together.”

    Shows Nvidia evolving hardware and software together

    alecco, 2023-12 · Recommendation · in “Intel CEO: 'The entire industry is motivated to eliminate the CUDA market'”

  2. “Quite an interesting video IMO.”

    Explains 1000x inference gains from numeric types and specialised instructions

    huijzer, 2023-12 · Recommendation · in “ASML and Samsung seal deal on 2nm chips”

  3. “Bill Dally of Nvidia suggests using actual logarithms, as they're more consistent in relative error than floats”

    Idea of using logarithmic numbers for low-precision AI arithmetic

    raphlinus, 2024-01 · Recommendation · in “Why does integer representation of a floating point approximate its logarithm?”

Excerpts from Hacker News comments, linked to the originals. Summaries and labels were written by a language model from those comments.

Topics: bill dally, nvidia, gpu, low-precision arithmetic, energy efficiency

Also recommended by the same people