More Noble Truths About Exascale

but is it faster than high frequency trade?

I couldn't read the report you linked to without signing up to another sight. I'm tired of the spam already that I refuse to sign up for anything now.

It seems to me that code bloat has crippled hardware advances. I believe a revolution in programming approach rather than any hardware advances will be the next significant improvement in computing. To make an analogy to the history of physics, I think that all of the different coding approaches and languages (C++, C#, Python, Java, Escher, Euler, etc.) out there are like the variety of electrostatic and magnetic laws discovered before Maxwell rolled them all into one elegant yet baffling set of equations. Maybe a better analogy is that a new mathematics needs to be born that allows for analyzing how to make "tight" effective coding. Kind of like the state of physics before Newton and Leibniz invented Calculus and differential equations.

I agree with you premise. I can remember when 64K was all we would ever need, now a "simple" program consumes much more then that. I am convinced the size of current software is a result of sloppy programing. Rather than looking at how efficient we can code something, we develop something that works regardless of the "bugs" in the code.