How AlphaChip transformed computer chip design • Hacker News Board

Re: How AlphaChip transformed computer chip design

#41

Post by vighneshiyer » Fri, Sep 27, 2024, 6:05 PM UTC

This work from Google (original Nature paper: https://www.nature.com/articles/s41586-021-03544-w) has been credibly criticized by several researchers in the EDA CAD discipline. These papers are of interest:

- A rebuttal by a researcher within Google who wrote this at the same time as the "AlphaChip" work was going on ("Stronger Baselines for Evaluating Deep Reinforcement Learning in Chip Placement"): http://47.190.89.225/pub/education/MLcontra.pdf

- The 2023 ISPD paper from a group at UCSD ("Assessment of Reinforcement Learning for Macro Placement"): https://vlsicad.ucsd.edu/Publications/Conferences/396/c396.p...

- A paper from Igor Markov which critically evaluates the "AlphaChip" algorithm ("The False Dawn: Reevaluating Google's Reinforcement Learning for Chip Macro Placement"): https://arxiv.org/pdf/2306.09633

In short, the Google authors did not fairly evaluate their RL macro placement algorithm against other SOTA algorithms: rather they claim to perform better than a human at macro placement, which is far short of what mixed-placement algorithms are capable of today. The RL technique also requires significantly more compute than other algorithms and ultimately is learning a surrogate function for placement iteration rather than learning any novel representation of the placement problem itself.

In full disclosure, I am quite skeptical of their work and wrote a detailed post on my website: https://vighneshiyer.com/misc/ml-for-placement/

Re: How AlphaChip transformed computer chip design

#42

Post by mikewarot » Fri, Sep 27, 2024, 6:06 PM UTC

ilaksh wrote: post #26

mikewarot wrote: post #16

I understand the achievement, but can't square it with my belief that uniform systolic arrays will prove to be the best general purpose compute engine for neural networks. Those are almost trivial to route, by nature.

Isn't this already the case for large portions of GPUs? Like, many of the blocks would be systolic arrays? I think the next step is arrays of memory-based compute.

Imagine a bit level systolic array. Just a sea of LUTs, with latches to allow the magic of graph coloring to remove all timing concerns by clocking everything in 2 phases.

GPUs still treat memory as separate from compute, they just have wider bottlenecks than CPUs.

Re: How AlphaChip transformed computer chip design

#43

Post by lordswork » Fri, Sep 27, 2024, 6:08 PM UTC

Some interesting context on this work: 2 researchers were bullied to the point of leaving Google for Anthropic by a senior researcher (who has now been terminated himself): https://www.wired.com/story/google-brain-ai-researcher-fired...

They must feel vindicated by their work turning out to be so fruitful now.

Re: How AlphaChip transformed computer chip design

#44

Post by lordswork » Fri, Sep 27, 2024, 6:09 PM UTC

epistasis wrote: post #6

yeahwhatever10 wrote: post #5

Why do they keep saying "superhuman"? Algorithms are used for these tasks, humans aren't laying out trillions of transistors by hand.

Google is good at many things, but perhaps their strongest skill is media positioning.

The media hates Google.

Re: How AlphaChip transformed computer chip design

#45

Post by onjectic » Fri, Sep 27, 2024, 6:11 PM UTC

cobrabyte wrote: post #38

I'd love a tool like this for PCB design/layout

First thing my mind went to as well, I’m sure this is already being worked on, I think it would be more impactful than even this.

Re: How AlphaChip transformed computer chip design

#46

Post by s-macke » Fri, Sep 27, 2024, 6:13 PM UTC

vighneshiyer wrote: post #41

This work from Google (original Nature paper: https://www.nature.com/articles/s41586-021-03544-w ) has been credibly criticized by several researchers in the EDA CAD discipline. These papers are of interest: - A rebuttal by a researcher within Google who wrote this at the same time as the "AlphaChip" work was going on ("Stronger Baselines for Evaluating Deep Reinforcement Learning in Chip Placement"): http://47.190.8…

When I first read about AlphaChip yesterday, my first question was how it compares to other optimization algorithms such as genetic algorithms or simulated annealing. Thank you for confirming that my questions are valid.

Re: How AlphaChip transformed computer chip design

#47

Post by gdiamos » Fri, Sep 27, 2024, 6:19 PM UTC

vighneshiyer wrote: post #41

This work from Google (original Nature paper: https://www.nature.com/articles/s41586-021-03544-w ) has been credibly criticized by several researchers in the EDA CAD discipline. These papers are of interest: - A rebuttal by a researcher within Google who wrote this at the same time as the "AlphaChip" work was going on ("Stronger Baselines for Evaluating Deep Reinforcement Learning in Chip Placement"): http://47.190.8…

Criticism is an important part of the scientific process.

Whichever approach ends up winning is improved by careful evaluation and replication of results

Re: How AlphaChip transformed computer chip design

#48

Post by jeffbee » Fri, Sep 27, 2024, 6:25 PM UTC

vighneshiyer wrote: post #41

This work from Google (original Nature paper: https://www.nature.com/articles/s41586-021-03544-w ) has been credibly criticized by several researchers in the EDA CAD discipline. These papers are of interest: - A rebuttal by a researcher within Google who wrote this at the same time as the "AlphaChip" work was going on ("Stronger Baselines for Evaluating Deep Reinforcement Learning in Chip Placement"): http://47.190.8…

It seems like this is multiple parties pursuing distinct arguments. Is Google saying that this technique is applicable in the way that the rebuttals are saying it is not? When I read the paper and the update I did not feel as though Google claimed that it is general, that you can just rip it off and run it and get a win. They trained it to make TPUs, then they used it to make TPUs. The fact that it doesn't optimize whatever "ibm14" is seems beside the point.

Re: How AlphaChip transformed computer chip design

#49

Post by pinewurst » Fri, Sep 27, 2024, 6:32 PM UTC

jonas21 wrote: post #18

epistasis wrote: post #6

Earlier quoted context omitted.

Google is good at many things, but perhaps their strongest skill is media positioning.

I feel like they're particularly bad at this, especially compared to other large companies.

Familiarity breeds contempt. They've been pushing the Google==Superhuman thing since the Internet Boom with declining efficacy.

Re: How AlphaChip transformed computer chip design

#50

Post by therein » Fri, Sep 27, 2024, 6:35 PM UTC

int0x29 wrote: post #20

xanderlewis wrote: post #12

Earlier quoted context omitted.

Nice. Do you offer API access for a monthly fee?

I'll need 7 5 gigawatt datacenters in the middle of major urban areas or we might lose the Bubble Sort race with the Chinese.

Have we decided when are we deprecating it? I'm already cultivating another team in a remote location to work on a competing product that we will include into Google Cloud a month before deprecating this one.