RAAIS 2017 - Simon Knowles, Co-Founder and CTO of Graphcore
The Research and Applied AI Summit - RAAIS3 HN recommenders · 2017–2018
7K views · talk · dated · Uploaded 2017-07-13
Simon Knowles, co-founder and CTO of Graphcore, talks about "intelligence processors", chips designed specifically for machine learning. The talk was given at the RAAIS 2017 summit and covers Graphcore's architecture.
HN points to it as a technical talk from a chip designer on why AI hardware should put memory close to compute and use async data transfers.
2:25:52The spelled-out intro to neural networks and backpropagation: building micrograd
Andrej Karpathy
18:40But what is a neural network? | Deep learning chapter 1
3Blue1Brown
6:538-bit computer update
Ben Eater
1:21:22Light Years Ahead | The 1969 Apollo Guidance Computer
TNMoC
27:25“Hello, world” from scratch on a 6502 — Part 1
Ben Eater
1:30:28AlphaGo - The Movie | Full award-winning documentary
Google DeepMind
27:14Transformers, the tech behind LLMs | Deep Learning Chapter 5
3Blue1Brown
47:52Finding the Missing Memristor - R. Stanley Williams
The Qualcomm Institute
3 comments from Hacker News
Interesting approach, but very far off. I prefer GraphCore as a promising new lead in AI hardware. Moving computation close in with memory, and making data transfers async will speed up DL. [VIDEO]
Indeed, and that's why there are a couple of startups working on new chips and why Google has the TPU. Here's a nice technical talk from Graphcore's CTO about that [VIDEO]
I went to a talk from the CTO of Graphcore ( [LINK] ) on Monday. They are designing chips targeted at machine learning. As I understood it, their architecture comprises - lots of "tiles" - small processing cores with collocated memory (essentially DSPs) - very high bandwidth (90TB/s!) switching fabric to move data between tiles - "Bulk Synchronous Parallel" operation, meaning that the tiles do their work and then the switching fabric moves the data, and then we repeat. The key challenge he pointed to was power - both in terms of getting energy in (modern CPUs/GPUs take similar current to your car starter motor!) and also getting the heat out. Logic gates take a lot more power than RAM, so he argued that collocating small chunks of RAM right next to your processing core was much better from a power perspective (meaning you could then pack yet more into your chip) as well as obviously being better from a performance perspective. [VIDEO] isn't quite the presentation I saw, but it has quite a lot of overlap. Hope that helps!