It’s no secret that multi-core computing is the future; Intel and AMD have told us they can’t make a single core go much faster, but they can pack more and more cores onto a single chip.
Current operating systems, compilers, and frameworks do little to assist the application architect or developer in taking advantage of this radical change in hardware.
Digipede’s grid computing software is remarkably good at distributing application workloads — not just across multiple machines on a grid, but also across multiple cores on a chip. Here’s a video I made this weekend that shows how we use the same exact technology to distribute calculations first across multiple cores on a single server, then across a larger grid. Please have a look — I managed to keep it under four minutes!
As a complement to this video, you may also want to check out Dan Ciruli’s earlier demonstration that shows the code changes required to grid-enable an application — just 20 lines of code!
http://www.digipede.net/products/whitepaper.html (scroll down to the “Videos” section)
I’ll have lots more about this topic in the coming few weeks.