Jim Keller - Moore's Law in the age of AI Chips

Channel:Matroid · Length:31:05 · Published:2020-04-20 · Views:33363

Watch on YouTube

Summary

Jim Keller's talk at the 2020 Scaled Machine Learning conference on Moore's Law in the age of AI chips. He compares scalar, vector and GPU-style parallel approaches to computation.

Why HN recommends it

Cited in hardware threads as a Keller talk on the different ways of doing parallel processing.

Jim Keller had an interesting talk recently [1] about ways of doing parallel processing — HN, 2020

Machine learning & deep learning · Computer architecture & CPUs · Tech pioneers & interviews

Subtitles (excerpt)

they actually didn't tell me I was talking about AI chips until I was news to me that's gonna be fun and I didn't tell them the title so I got told yesterday they printed it up as TBD and I thought that was a really good title and then it's morning I found out it was to be announced so I'm catching up here I got I'll do the best I can but everybody knows what TBD means right to be determined right and I've been talking about Moore's law and read the Tesla's with me she gave a talk on technical technology optimism because a lots of people think for for complicated reasons that there's a complexity wall Moore's laws dead we can't scale and I gave a talk at Berkeley and the professor's all asked me can you convince people that there's future and computer science and transistors so they'll stop wasting all their time on AI software and because well if somebody says by the way Moore's law is dead and computer architecture innovation is over and you're a college student you're gonna go into that like it's not gonna happen but it's really a funny space so when we tell you a little story and then at some point I hope to make fun of everybody so everybody ready for that I did know I was the slot after the slide of lunch but I found out on the first one so I'm the

On Hacker News

Jim Keller had an interesting talk recently [1] about ways of doing parallel processing to better us the billions of transistors we have - assuming the task is parallelizable. There's the scalar core (i.e the basic CPU) which is easy to program realtively. Then a scalar core with vector instructions - difficult to program efficiently. Then there are arrays of scalar cores, i.e. GPUs, so relativel… — tails4e, 2020-07
The pattern of holes technique in that link (MURA) is also used in UV lithography for IC fabrication (example: [VIDEO] ). — metrognome, 2020-12
Jim Keller giving a talk on Moore's Law at Berkeley, [LINK] And another [VIDEO] And another [LINK] DEC, PA Semi, Apple, AMD, Tesla, Intel. He's been everywhere. [LINK] The internal announcement memo: [LINK] — CalChris, 2020-06
What makes this more bizarre than Keller’s typical short stints at previous companies is that he has done a ton of media in the last year. He’s probably given more time to journalists/interviewers in 2020 than in the previous 3 decades of his career combined. A Fortune piece from May of this year gave some insights into his plans (as well as provided a nice overview of his career) ( [LINK] ). > K… — cepth, 2020-06

Related videos