Live data from Hacker News

Training Stable Diffusion from Scratch Costs <$160k

mosaicml.com

51–52 of 52 posts

Re: Training Stable Diffusion from Scratch Costs <$160k

#51

Interesting to think about where the cost will go in a few years. I remember in college intro to CS class back in 1998, where I heard the story of building the first computer that could perform at 1 TFLOPS[1]. It cost $46 million and took up 1600 square feet. Now a $600 Mac Mini will do double that. [1] https://en.wikipedia.org/wiki/ASCI_Red

It is not going to go down much anymore, because the end of Moore's law has been reached as physical limitations become a factor. You cannot scale chips close to 1 atom wide transistors.

Moore's law isn't dead. Only Dennard's law. See slide 13 here[0]. Moore's law stated that the number of transistors per area will double every n months. That's still happening. Besides, neither Moore's law nor Dennard scaling are even the most critical scaling law to be concerned about...

...that's probably Koomey's law[1], which looks well on track to hold for the rest of our careers. But eventually as computing approaches the Landauer limit[2] it must asymptotically level off as well. Probably starting around year 2050. Then we'll need to actually start "doing more with less" and minimizing the number of computations done for specific tasks. That will begin a very very productive time for custom silicon that is very task-specialized and low-level algorithmic optimization.

[0] Shows that Moore's law (green line) is expected to start leveling off soon, but it has not yet slowed down. It also shows Koomey's law (orange line) holding indefinitely. Fun fact, if Koomey's law holds, we'll have exaflop power in 0: (Slide 13) https://www.sec.gov/Archives/edgar/data/937966/0001193125212...

1: "The constant rate of doubling of the number of computations per joule of energy dissipated" https://en.wikipedia.org/wiki/Koomey%27s_law

2: "The thermodynamic limit for the minimum amount of energy theoretically necessary to perform an irreversible single-bit operation." https://en.wikipedia.org/wiki/Landauer%27s_principle

Re: Training Stable Diffusion from Scratch Costs <$160k

#52

Earlier quoted context omitted.

It is not going to go down much anymore, because the end of Moore's law has been reached as physical limitations become a factor. You cannot scale chips close to 1 atom wide transistors.

Moore's law isn't dead. Only Dennard's law. See slide 13 here[0]. Moore's law stated that the number of transistors per area will double every n months. That's still happening. Besides, neither Moore's law nor Dennard scaling are even the most critical scaling law to be concerned about... ...that's probably Koomey's law[1], which looks well on track to hold for the rest of our careers. But eventually as computing app…

Also even MHz increases have had a bit of a comeback lately, with the fastest mid-2000's Pentium 4's reaching 3.8-4.2GHz and the latest Ryzen 7000's reaching 6GHz.
Post reply on HN