The game is changing and fast it is, as at 30,000 cores pointed at one problem it doesn’t take long to solve at all as that’s what a company called Cycle Computing demonstrate recently. As they took the potential of Amazon’s cloud to task by building increasingly large clusters on the Elastic Compute Cloud by joining together an amazing 30,000 CPU (central processing units) in order to form one massive parallel computer for a major Pharma company.
So whats new you ask as this happens all the time as even spent a good part of my career building these things. Well the kicker is the cost typically this is huge as your talking massive datacenters with millions, yes millions of watts of power enough to light a small city coupled with thousands and thousands of computers, the man-hours to hook it all together and so on. However the cost here was a mere $1,279 per hour, including the fees to Amazon and Cycle Computing! This is unbelievable as it ran for only 7 hours to solve a complex molecular modeling problem for a drug company.
Even with all the geekdom aside the amazing piece of this is what it means to solving problems as mentioned in the days past a problem like this would have cost hundreds of millions to solve and here not even pennies on the dollar, however fractions of the cent were achieved. As to further distill this down in seven hours Cycle Computing on this cluster performed the equivalent of 10.9 “compute years of work” for under $10,000 dollars an amazing feat to say the least.
As this is a real game changer for man as there have been puzzles which have eluded us in the past and through the answers where achievable the costs (as they always seem to do) remained as the barrier. As we embrace the cloud and the things it can to for us many things will change as there was a day you needed (or at least felt you did) to own a book for the knowledge it contain, then you found that after you had read it, it did nothing but gather dust as you had taken what you needed from it.
The same is now playing out with computing power as owning it is more akin to allowing dust to collect on a book, when you can buy what you need when you need and in a quantity which is unheard of in the past. In the end this isn’t about some geeky technology of let’s see if we can dim the lights and calculate Pi to some crazy decimal place. In fact it’s about how we will be able to help to transform how science itself is done…