Live data from Hacker News

Jim Keller moves to AI chip startup

reuters.com

1–10 of 94 posts

Re: Jim Keller moves to AI chip startup

#4
post #2

This page has some details of Tenstorrent's current chips [1]. Looks like a manycore simd design focused on tensor ops from a brief look. Apparently they also have some sort or compression scheme to boost memory bandwidth. 1: https://www.tenstorrent.com/technology/

They have a Hot Chips presentation on YouTube with a lot more detail: https://www.youtube.com/watch?v=HLjumOyWj0g

I don't notice anything in particular that stands out vs. the many other AI chips people are making, at first glance. But I'm far from an expert. There are several other technical videos on their YouTube channel as well: https://www.youtube.com/channel/UC7041p6DlAh0r4_Fnlk10pQ

Re: Jim Keller moves to AI chip startup

#5
post #4
post #2

This page has some details of Tenstorrent's current chips [1]. Looks like a manycore simd design focused on tensor ops from a brief look. Apparently they also have some sort or compression scheme to boost memory bandwidth. 1: https://www.tenstorrent.com/technology/

They have a Hot Chips presentation on YouTube with a lot more detail: https://www.youtube.com/watch?v=HLjumOyWj0g I don't notice anything in particular that stands out vs. the many other AI chips people are making, at first glance. But I'm far from an expert. There are several other technical videos on their YouTube channel as well: https://www.youtube.com/channel/UC7041p6DlAh0r4_Fnlk10pQ

Thanks for these! My assumption is that hiring Keller means they plan to gain advantage by world class execution rather than some crazy architectural leap of faith.

Re: Jim Keller moves to AI chip startup

#7
post #2

This page has some details of Tenstorrent's current chips [1]. Looks like a manycore simd design focused on tensor ops from a brief look. Apparently they also have some sort or compression scheme to boost memory bandwidth. 1: https://www.tenstorrent.com/technology/

Their marketing material states: "Facilitating machines to go beyond pattern recognition and into cause-and-effect learning".

I wonder what they are referring to. Are they accelerating what SHAP's GradientExplainer [1] does? (namely: crafting inputs at a specific layer, propagating forward to see the influence on class prediction, and sort of backpropagating to pixels) Or is it about something more related to Judea Pearl's work on causality?

[1] https://github.com/slundberg/shap#deep-learning-example-with...

Re: Jim Keller moves to AI chip startup

#10
post #5
post #4

Earlier quoted context omitted.

They have a Hot Chips presentation on YouTube with a lot more detail: https://www.youtube.com/watch?v=HLjumOyWj0g I don't notice anything in particular that stands out vs. the many other AI chips people are making, at first glance. But I'm far from an expert. There are several other technical videos on their YouTube channel as well: https://www.youtube.com/channel/UC7041p6DlAh0r4_Fnlk10pQ

Thanks for these! My assumption is that hiring Keller means they plan to gain advantage by world class execution rather than some crazy architectural leap of faith.

Watching the video I think turning each tensor into packets is quite clever as you get some ability to send them around a network and organise layers/data manipulation/transforming the layers/compression all as part of the stack.

I’m pretty surprised no-one has actually exposed the actor model for parallelising neural networks, it seems it would work quite well and allow you to have a layer per node (or actually many split configurations). Maybe data locality would be an issue with actor based approaches. They seem to be solving this at a lower level but with less knowledge of the actual parallelism in software.

Post reply on HN