Hacker News new | past | comments | ask | show | jobs | submit login

This seems to simply be an implementation of the algorithm on the GPU. I'm not saying it isn't worth doing that. But a breakthrough? I wouldn't call it that.

This won't allow faster chips. It's only a time saver.




This will indeed allow faster chips, because reverse lithography used to be so slow that it was only used for the most critical parts of a chip. Now that it is more tractable, RL can be applied to many more parts of the chip. That will allow designers to better optimize the layout in those other parts, which leads to faster chips.


Parallizing all important algorithms are necessary. For you it’s just a time saver, for NVIDIA it’s business and one more thing to add to it’s computational quasi monopoly.


I didn't say it's not necessary. I didn't say it's not good business to do it. I even explicitly said it's worth doing which you have ignored for some reason.

I just disagree that this is a breakthrough.


Faster chips are only time savers




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: