[Beowulf] K Computer built for speed, not use

Lux, Jim (337C) james.p.lux at jpl.nasa.gov
Thu Oct 11 08:29:17 PDT 2012


-----Original Message-----
From: beowulf-bounces at beowulf.org [mailto:beowulf-bounces at beowulf.org] On Behalf Of Prentice Bisbal
Sent: Thursday, October 11, 2012 6:44 AM
To: beowulf at beowulf.org
Subject: Re: [Beowulf] K Computer built for speed, not use

On 10/11/2012 12:37 AM, Mark Hahn wrote:
>> Any general purpose system will inevitably underperform for some 
>> people, and many might argue that the art of managing such a project 
>> is making sure everyone squawks equally loud about how the stake is 
>> being driven into their heart.
> I think of it from the other direction: a specialized machine would 
> need to demonstrate really significant savings.

This is the logic behind DE Shaw's Anton computer for molecular dynamics, except instead of saving money, the item being saved is time. 
In their papers they argue that designing the unique processors for the Anton put them 5 years ahead of waiting for general-purpose processors to achieve the same performance. Of course, money was not a concern.
>> yes... They're a very intriguing company in many ways.  It would be an interesting place to work (if it weren't in the hellhole of the east coast.. my California upbringing coming to the fore with some geographic trash talk)


This whole discussion of the K computers practicality reminds me of autoracing. The saying is "Win on Sunday, sell on Monday", with the logic that winning on Sunday will boost car sales for the "brand" that one on the following Monday. The other argument is that the racing competition spurs engineering advances that trickles down to passenger cars. And we all know that cars that race on Sunday, whether Indy Car, Formula 1, NASCAR, or whatever, don't make good passenger cars or pickups, and are actually illegal to use on streets.

>> I don't know that car manufacturers do this so much any more. I think it's more like the space program.  If you want to recruit top talent, telling that mechanical engineer that they will be third assistant door latch designer isn't nearly as attractive as saying "you get to work on Formula 1 engines".  Same in the space biz: we don't pay wages competitive with, say, Silicon Valley, but we pay in "space dollars" and the chance to say "yep, I helped build the landing radar that landed MSL successfully".  The same applies to aerospace manufacturers and why they do NASA work (which is generally not profitable for them, since our volumes are small compared to Defense or Commercial Space).. it's a lure for new hires: "You can work on a big antenna for a satellite that will help us understand climate change" is more attractive than "you can help work on a big antenna that will be used to intercept cellphone calls"  (and you can talk about what you do at work to your friends and parents.. not generally possible with the latter)


IMHO, the only computer companies doing the "Win on Sunday, sell on Monday" thing right are the companies like Intel and IBM, and Mellanox. 
These companies have many products in the top 500, but are also available to the "general public" (in quotes because I'm not sure you can call cluster purchasers the general public), and make enough money to fund their own R&D.

While supercomputers like the K are awesome feats of engineering, if those innovations don't trickle down into products with more widespread availability, I have to wonder what's in it for the company and the rest of the world (other than government grants keeping them afloat, which isn't a great business model in the end)?

>> Government grants and contracts are a fine business model. It's just different than commercial. The government always pays the invoice (slowly), and they are ALWAYS going to be buying certain kinds of services.  Basic logistics and maintenance functions, for instance. There are lots of companies that do quite nicely on providing, say, janitorial services to the government.  There are also companies that specialize in providing technical services (SAIC, Booz-Allen-Hamilton, for instance).   Lots of little jobs, each yielding a nice 5-10% profit, and once you've developed the institutional processes to do the contracts administration, etc., it works very nicely.  And then, there's tonnes of small research organizations that survive nicely on a continuing stream of small contracts and grants (in the 50k to 1M sort of range).  I used to work for a company that fit in the latter category.. we ranged from 5-15 employees depending on what work was in house.  We wrote lots of proposals, got 10% or so of them, and managed to keep everyone busy doing lots of different things.  Granted, that company would never grow to 100 people, but the owners didn't really want to, either. And, when the owners die or retire (more likely the former than the latter), the company will cease to exist.

I hope I didn't go too far off-topic just then.

--
Prentice
_______________________________________________
Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf


More information about the Beowulf mailing list