Infiscale

Gigaflops, teraflops, petaflops, exaflops. Every step up is a scale increase of one thousand. Presently, the largest supercomputers deployed perform not much more than 30 petaflops. More than five years ago our software was used by IBM in the booting of the first public petascale system, Roadrunner, at just over 1 petaflops [link]. This massive and amazing system was only 1/1000th of an exascale system. Today's best supercomputers would have to be scaled up at least 30 times their current peak to meet that exascale mark. More transistors and more "me too" companies won't cut it - innovation will.

Don't get us wrong. We are not saying that there aren't good companies doing real work towards exascale development. There are. We work with a few great ones. It just is not common.

The industry is not there yet...

Tripling the scale of the largest current supercomputers is a huge jump. So why are there so many companies out there promoting "exascale" technologies? Seriously, Google it. There is a great deal of marketing hype out there for something that is distant on the horizon.

Look at all the articles on topics like exascale programming. How can this be known when the hardware to provide it is not yet available? Likely those are people who do not have a clue about the complexity of even a petascale system and the serious effort it takes to make it work. We have done these systems and it is a seriously hard thing to do if really done right.

Exascale will get there, and we are working with the people who will help bring it to you. However, there are others that saw how big of news petascale was, and lacking real experience or expertise with it they decided to move beyond that and promote exascale vapor. Yes, any company that is talking about exascale that has not pulled off petascale successfully is pushing vaporware.

Exascale can't become another contemporary buzzword victim like "cloud" or "green" - words that have had their meaning diluted by questionable marketing with the intent to deceive. This nonsense shifts public opinion and takes the technology away from science, putting it into the wasteful domain of moving iron by any means possible.

Exascale represents a real, tangible technology threshold that could offer radical advancements in many scientific disciplines. The potential for societal benefit are enormous, and we want and will see this happen.

We are ready for the challenge. Our software is waiting for the first exascale-capable batch of hardware to become available.