Reply To: Wealth of Nations, and Economics of Abundance

[Alex Pollard]

I recently saw a fascinating presentation from the company Simmersion:

http://www.simmersion.com.au/

The company is having a lot of success selling urban landscape modelling tools to local councils to help with planning decisions. The interesting
stuff is what else you can do with high-power off-the-shelf graphics hardware, in real-time.

What they are now doing at Simmersion is to utilisise the massively parallel capabilities of a typical graphics “shader” to do very intensive computations in about 1/1000th the time a typical CPU could do it. So for example, in one simulation the sky is coloured according to a scattering and refraction formula about two pages in length. Every pixel in the sky is painted using this algorithm in real-time – it is not pre-rendered.

A clearer example of this idea is a galaxy simulation. On screen a galaxy is displayed with a very bulbous massive sphere at the centre of a uniform disc of stars orbiting around it. Then the simulation sends a black hole through the disc and you see on the screen all the hundreds of millions of simulated stars being flung around in the black hole’s gravity field. Another simulation (this one was from Nvidia) showed real-time computational fluid dynamics – you could manipulate a puff of dust and a pool of water in real time. The graphics card is doing the simple drawing, but importantly, it’s also doing the intensive computational fluid dynamics calculations in real time.

My point, finally, is that if the average PC user has Windows Vista, they will have a very powerful parallel numerical processor. Your average technology enthusiast could model highly complicated plasma effects on their home PC. The power of graphics cards is doubling every 6 months so it won’t be long before realistic real-time particle-in-cell plasma simulations are possible on desktop PCs.

You could have a virtual plasma science workshop on your desktop, showing comparisons of Tokamak fusion attempts vs focus fusion vs the Bussard concept for example.

For the first time we may also have amateur astronomers creating highly orthodox simulations of galaxies and solar systems. As people play around with this stuff, they may come to realise how completely inadequate the models are. Previously only the institutional researchers have access to this power.

Enabling such a direct comparison could convince a lot of people that the conventional models are seriously lacking. I know computer models are garbage-in-garbage-out, which is why enabling a comparison would be best.

I’d suggest finding someone from the games industry to kick-start this effort. They should have lots of graphics cards know-how. They may be looking for another outlets for their talents, like many others in the gaming industry who are finding the competitiion getting intense – the games industry is now bigger than Hollywood.