Hacker News Reader

Comment by bluedino | Hacker News Reader

It's interesting seeing these machines being built by Tesla, Meta, private companies.

Typically these are things built by national laboratories, university research labs...

It's a testament to the immediate applications of AI/ML. In the past, supercomputers were mainly a tool for research, with expected returns from that research in the order of years to decades. Now, as demonstrated by how large language models are generative image models are scaling, there's a reasonable expectation of short term returns from investment.

I'm not sure that adoption by that particular list of companies is evidence for "a reasonable expectation of short term returns from investment". VC-backed startups and cresting tech giants are traditionally willing to splurge huge amounts of cash on mad plans which have a slim chance of delivering universal domination.

If all of my local pizza parlours were installing (well, renting) AI/ML rigs, it would be a different story.

There's a good chance that the parent franchises of all of your local chain pizza-places are doing precisely that.

Private companies likely would have clusters that dwarf the majority of computers on the public lists. Except they don't tell you about them.

Why would they? So their competitors can copy their business plan?

IBM has had large clusters for decades.

Any large bank or trading company would have clusters that would be at home in that list.

You don't want to know what the Apple cluster is like: they have to train for Siri, in multiple languages, etc. I be that's several orders of magnitude larger than Tesla.

AWS itself is probably several orders larger than most university clusters, probably larger than most country's combined computational capacity.

Similar classes of machines have been build and used for simulations in industries for decades now.

In fact if you look through the historic top500 lists a lot of the space outside the top10 is occupied by industry clusters. Classically oil and gas has been one of the big consumers. In recent years there are more and more cloud clusters in there as well.

So this Tesla cluster isn't really anything unusual.

I work for a large chemical company and we only have a single DGX

The entire company only has a single DGX or your section/office/division has only a single DGX? Many of these large companies are so big the left hand doesn't know what the right hand is doing. One department may be starving for cash while another one across the office cubic is awash in funding.

There are some companies that don't have centralized high-performance compute resources so each department goes out and builds or buys and their own performance compute resources and strings it along. Many universities do this too where physics has a compute cluster and economics and bio-sciences have another cluster and sometimes they are underpowered and if all the departments grouped together and pooled their resources they would have adept system administrators and developers and hardware to support their computing needs but it doesn't always happen with poor governance so they end up with stuff that appears to be built from someone's basement and when that person leaves, retires, graduates there's so much poor documentation none know what to make of anything.

The privately owned machines are undercounted on public lists, and always have been. Before it was ML it was big engineering firms and oil companies who owned the huge computers.

Of the top500 supercomputers, 47% are owned by for-profit industry [1]. They are underrepresented in the top ranks: they only make up 16.7% of the total performance of the list, and from a quick skim it seems like of the top 25 only 5 are from private companies (two energy companies, Samsung, Yandex and MS Azure).

1: https://www.top500.org/statistics/list/ (select the "segments" category)

In fact, you are seeing the Trillion dollar companies not just make their own datacenters, but also chips; Dojo, TPU (Amazon, Meta ).

Bank of America too

This is symbolic of the maturation of academic research.

Academics publish their research even when it contributes nothing to society. Private enterprises and government agencies aren't as aggressive in publishing even when the publication of their research may benefit society more than many of the stuff coming out of academia. Some do for various reasons but sometimes they don't. That's why when companies go belly up and historians have their field day, we find out stuff such as Kodak discovering the benefits of digital photography since the 1960's and certain employees within different departments were pushing for Kodak to take the lead, yet Kodak executives didn't know what to make of it and where afraid of digital photography eating into their film business.

We also learn stuff such as ATT developing video phones since the 1960's but they were too early and it never go anywhere until decades later when ATT was a lesser relevant player in the telecom industry. And the Post Office foresaw the coming trend of online bill pay and email eating into their mail volume since the early 90's, yet when their they had projects that would enable senders and receivers to received mail scanned to a USPS account or pay bank statements through a USPS account for those early bill pay adopters, Postal Management did nothing with it.

Reply on news.ycombinator.com