Log in

No account? Create an account

Previous Entry | Next Entry

A busy day of Supercomputing

I didn't get any writing done last night apart from on the shuttle bus: I got into my hotel room around 10pm, lay on the bed, and fell straight to sleep.

Today started with the Keynote Talk. It was preceded by an entertaining video, saying “Ladies and Gentlemen. Before this morning's production, Please Extinguish all Laptops; Set Cellphones to Stun; Unbuckle your Seatbelt; and Let your Hair Down!”

The talk was by Neil Gershenfeld, who leads MIT's Center for Bits and Atoms. He started by talking about "Paintable Computing", which was a fascinating concept - where computing power comes in the form of miniature interexchangeable and peer-to-peer-connecting units that come in a barrel, and if you need more computing power you can paint an extra coat of compute units on your wall. He gave some nice examples of cellular automata, including physical ones (bubbles travelling through tiny tubes). He talked a bit about what he called "mathematical programming", where the route from problem to solution never involves going near a computer - but to me it seemed a bit dubious to call this "programming".
He spent a lot of his time talking about his Fab Labs, where people (in Boston, India, Norway, or wherever) can design and make (almost) anything. E.g. 8-year-old kids building circuit boards, or designing a 3D construction kit that you can build from a cardboard box using a laser cutter. Like Ray Kurzweil last year, he think 20 years' time will be astonishingly different; in this case he thinks we'll have pocket molecular assemblers by 2027. His talk was very interesting, enjoyable and thought-provoking, as the keynote talks at SC conferences tend to be.

I then spent an hour on the MathWorks booth, where I was able to answer a few people's questions on our Distributed Computing tools, which was quite fun. Being on the booth at SC is pretty much the only time I ever meet customers, and although I wouldn't want to do it for a living, it is fun occasionally getting put on the spot and solving customer problems on the fly.

I found a very very cool thing on the exhibition floor today. You know I said there weren't any Porsches? Well, there was a Camaro, even if it was only a foot long. A company called Univa UD had a Transformers toy – the Ultimate Bumblebee – which they're giving away in a prize draw. I spent over half an hour figuring out how to transform him from his car form into his robot form. Then I transformed him back in about five minutes, which made the Univa people's jaws drop - "you definitely win the award for fastest transform so far!" They had other Transformers toys too, including a hilarious Mister Potato Head dressed up like Optimus Prime - they called him Opti-Mash Prime. I meant to go back to their booth to enter the prize draw, but the rest of the day was too busy.

Then I went to a "Masterworks" session, which was in two parts. The first was on the genetics of plant breeding, analysing the effects of certain genes in maize. They're sequencing a lot of maize DNA, and the HPC comes in when they analyse the changes in DNA between thousands of individuals, analyse the behaviour of those same individuals, and look for correlations; and they're now using HPC to make predictions about how genes will express themselves in certain circumstances. It was interesting, but doesn't seem too different to things Morag has done :)

Then there was a talk by General Motors on energy efficiency, the way they're trying to reduce fuel consumption in the short term, and move completely away from petrol/diesel in the longer term (before oil prices go to $100/barrel because there are no more oil sources). He says they're aiming for electric cars which can be recharged at the petrol station in 5 minutes, and that GM have 140 people whose job is just working on making practical fuel cells; and apparently the US Government has a project called FreedomCAR which will be hydrogen-powered, expected by 2015.
The (very little) HPC-related content in his talk related to the question of how to store hydrogen, so that a car can have range >300miles and yet refuel in <5min. They've used chemical modelling to figure out some compounds which absorb hydrogen, store it in solid form, then release it at near-room temperature and pressure. (Li2NH + H2 <-> LiNH2 + LiH was one of them.)

I walked around the so-called Disruptive Technologies exhibit hall, about upcoming technologies that are expected to have significant impact on the industry. The most interesting things there were company called Nantero who have a non-volatile RAM based on carbon nanotubes, and one called D:Wave who apparently have a commercially viable quantum computer - a superconducting adiabatic quantum processor, apparently.

I've got almost no novel writing done today - only about 300 words :( I did have a Three Card Blind round to judge, which took over half an hour, and I spent a certain amount of time playing with Transformers and randomly wasting time online. I need to be more disciplined with the small chunks of time I get during the day, or I'll get massively behind on my novel.

There are two things about being in America that I really like. One is the ubiquity of drinking fountains, in airports, conference centres, hotels and so on. Being someone who drinks a lot/ of water, this makes me happy. The other is the way that those buildings all tend to be air conditioned. Despite the weather here in Nevada being almost as cold as England, I haven't needed to go out in my coat, because I never spend more than a minute or two outside.

I discovered an interesting difference in attitude yesterday between myself and the other MathWorkers. As I've mentioned, I spend the time when I'm not doing SC07y things writing a novel, exploring the local arcade, surfing the net and suchlike. (Well, mostly just writing a novel.) A few of the more keen MathWorkers here spend the time when they're not doing SC07y things carrying on work on MathWorks stuff - spending in some cases several hours a day carrying on working. As far as I'm concerned, given how much of my time here is spent on effectively work-related things, I view their extra work as equivalent to working weekends, and I don't have a problem with them doing that, as long as they don't have a problem with me not doing it.

Here's the view I get when I get out of the elevator on the 25th floor and turn to look down the corridor to my room: A corridor stretching off to infinity...
My room is three from the far end of that corridor. Infinity minus three is a long walk.

Right, now I'm off to the hotel to do some more writing, and hopefully get to at least a thousand. There's no reason why I shouldn't be able to get to 1667 today, actually, given it's only 8:20pm local time. Except that I'm running out of chapters I've plotted-ahead, but hey, that's part of what NaNo is all about...


( 3 comments — Leave a comment )
Nov. 14th, 2007 08:20 am (UTC)
The Transformers sound very cool. Yay for you impressing the designers :)

There are two things about being in America that I really like. ... The other is the way that those buildings all tend to be air conditioned. Despite the weather here in Nevada being almost as cold as England, I haven't needed to go out in my coat, because I never spend more than a minute or two outside.
Tut tut. George Monbiot wouldn't approve.

Good luck with your writing! I was a bit ahead on Sunday, but I'm way behind now, having written 74 words on Monday and nothing on Tuesday. I need to figure out how to get my characters out of the difficult situation I've cunningly trapped them in, before I can write any more.
Nov. 15th, 2007 08:05 pm (UTC)
Tut tut. George Monbiot wouldn't approve.
Well... good point. But they're not going to turn off the aircon if I ask them to, and given that they have it on, I can either wear jumpers for no reason, or take advantage of it.
I confess I've been to hot bits of the US more than cold, and I appreciate the aircon more in hot bits than in cold. (In cold bits, you could just use central heating.)
Nov. 14th, 2007 09:23 am (UTC)
A few months ago there was a talk at the Genome Campus where I work on quantum computing. The speaker mentioned that the current scientific consensus is that to perform any useful quantum computation will require over 200 qubits which is completely beyond what the experimentalists think may be possible. At present quantum computers have only been built with a very small number of qubits (I can't remember if the most achieved was 3, 5, or 8).

D:Wave's claims aren't being taken seriously by the scientific community because:
a) The workings of Orion are secret (how do we know they're not making it up?)
b) They don't seem to understand key things about how quantum computers work

D:Wave seem to think that linking their 'quantum chips' together will result in a quantum computer with a higher number of qubits. Their roadmap calls for an increase to 32 qubits by the end of the year, 512 qubits in early 2008, and an astonishing 1,024 qubits by the middle of 2008. This is just astonishing, entanglement between such chips would not occur, if it was that easy we'd have built quantum computers ages ago.
( 3 comments — Leave a comment )