Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

How AlphaChip transformed computer chip design

deepmind.google

Post reply on HN
Print view View original

111–120 of 215 posts

Re: How AlphaChip transformed computer chip design

#111

Post by isotypic »

negativeonehalf wrote: post #71
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…

FD: I have been following this whole thing for a while, and know personally a number of the people involved. The AlphaChip authors address criticism in their addendum, and in a prior statement from the co-lead authors: https://www.nature.com/articles/s41586-024-08032-5 , https://www.annagoldie.com/home/statement - The 2023 ISPD paper didn't pre-train at all. This means no learning from experience, for a learning-base…

Why does pretraining or not matter in the ISPD 2023 paper? The circuit_training repo, as noted in the rebuttal of the rebuttal by the ISPD 2023 paper authors, claims training from scratch is "comparable or better" than fine-tuning the pre-trained model. So no matter your opinion on the importance of the pretraining step, this result isn't replicable, at which point the ball is in Google's court to release code/checkpoints to show otherwise.

Re: How AlphaChip transformed computer chip design

#112

Post by foobarqux »

clickwiseorange wrote: post #94
negativeonehalf wrote: post #86

Earlier quoted context omitted.

For AlphaChip, pre-training is just training. You train, and save the weights in between. This has always been supported by the Google's open-source repository. I've read Kahng's FAQ, and he fails to address this, which is unsurprising, because there's simply no excuse for cutting out pre-training for a learning-based method. In his setup, every time AlphaChip sees a new chip, he re-randomizes the weights and makes i…

It's not that one needs an excuse. The Google CT repo said clearly you don't need to pretrain. "supported" usually includes at least an illustration, some scripts to get it going - no such thing there before Kahng's paper. Pre-trained was not recommended and was not supported. Everything optimized in Nature RL is an approximation. HPWL is where you start, and RL uses it in the objective function too. As shown in "Str…

I think clicking the time stamp field in the comment will allow you to get a reply box.

Re: How AlphaChip transformed computer chip design

#113

Post by therealcamino »

hinkley wrote: post #14

TSMC made a point of calling out that their latest generation of software for automating chip design has features that allow you to select logic designs for TDP over raw speed. I think that’s our answer to keep Dennard scaling alive in spirit if not in body. Speed of light is still going to matter, so physical proximity of communicating components will always matter, but I wonder how many wins this will represent ver…

EDA software has long allowed trading off power, delay, and area during optimization . But TSMC doesn't produce those tools, as far as I'm aware.

Re: How AlphaChip transformed computer chip design

#114

Post by gabegobblegoldi »

thesz wrote: post #100

Eurisco [1], if I remember correctly, was once used to perform placement-and-route task and was pretty good at it. [1] https://en.wikipedia.org/wiki/Eurisko What's more, Eurisco was then used in designing Traveler TCS' game fleet of battle spaceships. And Eurisco used symmetry-based placement learned from VLSI design in the design of the spaceships' fleet. Can AlphaChip's heuistics be used anywhere else?

Doesn’t look like it. In fact the original paper claimed that their RL method could be used for all sorts of combinatorial optimization problems. Yet they chose an obscure problem in chip design and showed their results on proprietary data instead of standard public benchmarks.

Instead they could have demonstrated their amazing method on any number of standard NP hard optimization problems e.g. traveling salesman, bin packing, ILP, etc. where we can generate tons of examples and verify easily whether it produces better results than other solvers or not.

This is why many in the chip design and optimization community felt that the paper was suspicious. Even with this addendum they adamantly refuse to share any results that can be independently verified.

Re: How AlphaChip transformed computer chip design

#115

Post by clickwiseorange »

negativeonehalf wrote: post #99
clickwiseorange wrote: post #94

Earlier quoted context omitted.

It's not that one needs an excuse. The Google CT repo said clearly you don't need to pretrain. "supported" usually includes at least an illustration, some scripts to get it going - no such thing there before Kahng's paper. Pre-trained was not recommended and was not supported. Everything optimized in Nature RL is an approximation. HPWL is where you start, and RL uses it in the objective function too. As shown in "Str…

The Nature paper describes the importance of pre-training repeatedly. The ability to learn from experience is the whole point of the method. Pre-training is just training and saving the weights -- this is ML 101. I'm glad you agree that HPWL is a proxy metric. Optimizing HPWL is a fun applied math puzzle, but it's not chip design. I am unaware of a single instance of someone using SA to generate real-world, usable ma…

1. The Nature paper said one thing, the code did something else, as we've discovered. The RL method does some training as it goes. So, pre-training is not the same as training. Hence "pre". Another problem with pretraining in Google work is data contamination - we can't compare test and training data. The Google folks admitted to training and testing on different versions of the same design. That's bad. Rejection-level bad.

2. HPWL is indeed a nice simple objective. So nice that Jeff Dean's recent talks use it. It is chip design. All commercial circuit placers without exception optimize it and report it. All EDA publications report it. Google's RL optimized HPWL + density + congestion

3. This shows you aren't familiar with EDA. Simulated Annealing was the king of placement from mid 1980s to mid 1990s. Most chips were placed by SA. But you don't have to go far - as I recall, the Nature paper says they used SA to postprocess macro placements.

SA can indeed find mediocre solutions quickly, but keeps on improving them, just like RL. Perhaps, you aren't familiar with SA. I am. There are provable results showing SA finds optimal solution if given enough time. Not for RL.

Re: How AlphaChip transformed computer chip design

#116

Post by Zamiel_Snawley »

ninetyninenine wrote: post #70

What occupation is there that is purely intellectual that has no chance of an AI ever progressing to a point where it can take it over?

I think only sentimentality can prevent take over by a sufficiently competent AI.

I don’t want art that wasn’t made by a human, no matter how visually stunning or indistinguishable it is.

Re: How AlphaChip transformed computer chip design

#117

Post by ninetyninenine »

Zamiel_Snawley wrote: post #116
ninetyninenine wrote: post #70

What occupation is there that is purely intellectual that has no chance of an AI ever progressing to a point where it can take it over?

I think only sentimentality can prevent take over by a sufficiently competent AI. I don’t want art that wasn’t made by a human, no matter how visually stunning or indistinguishable it is.

Discounting fraud... what if the AI produces something genuinely better. Genuinely moving you to tears? What then?

Imagine your favorite movie, the most moving book. You read it, it changed you, then you found out it was an AI that generated it in a mere 10 seconds.

Artificial sentimentality is useless in the face of reality. That human endeavor is simply data points along an multi-dimensional best fit curve.

Re: How AlphaChip transformed computer chip design

#118

Post by deelowe »

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.

I read the paper. Superhuman is a metric they defined in the paper which has to do with how long it takes a human to do certain tasks.

Re: How AlphaChip transformed computer chip design

#119

Post by AshamedCaptain »

negativeonehalf wrote: post #99
clickwiseorange wrote: post #94

Earlier quoted context omitted.

It's not that one needs an excuse. The Google CT repo said clearly you don't need to pretrain. "supported" usually includes at least an illustration, some scripts to get it going - no such thing there before Kahng's paper. Pre-trained was not recommended and was not supported. Everything optimized in Nature RL is an approximation. HPWL is where you start, and RL uses it in the objective function too. As shown in "Str…

The Nature paper describes the importance of pre-training repeatedly. The ability to learn from experience is the whole point of the method. Pre-training is just training and saving the weights -- this is ML 101. I'm glad you agree that HPWL is a proxy metric. Optimizing HPWL is a fun applied math puzzle, but it's not chip design. I am unaware of a single instance of someone using SA to generate real-world, usable ma…

SA and HPWL are most definitely used as of today for the chips that power the GPUs used for "ML 101". But frankly this has the same value as saying "some sort algorithm is used somewhere" -- they're well entrenched basics of the field. To claim that SA produces "bad congestion" is like claiming that using steel pans produces bad cooking -- needs a shitton of context and qualification since you cannot generalize this way.

Re: How AlphaChip transformed computer chip design

#120

Post by negativeonehalf »

isotypic wrote: post #111
negativeonehalf wrote: post #71

Earlier quoted context omitted.

FD: I have been following this whole thing for a while, and know personally a number of the people involved. The AlphaChip authors address criticism in their addendum, and in a prior statement from the co-lead authors: https://www.nature.com/articles/s41586-024-08032-5 , https://www.annagoldie.com/home/statement - The 2023 ISPD paper didn't pre-train at all. This means no learning from experience, for a learning-base…

Why does pretraining or not matter in the ISPD 2023 paper? The circuit_training repo, as noted in the rebuttal of the rebuttal by the ISPD 2023 paper authors, claims training from scratch is "comparable or better" than fine-tuning the pre-trained model. So no matter your opinion on the importance of the pretraining step, this result isn't replicable, at which point the ball is in Google's court to release code/checkp…

The quick-start guide in the repo that said you don't have to pre-train for the sample test case, meaning that you can validate your setup without pre-training. That does not mean you don't need to pre-train! Again, the paper talks at length about the importance of pre-training.
Post reply on HN