and in support of Stefan's language/software observations, the hardware aspect is nearly as important, but perhaps more subtle ...
AMD and ARM are committed to the HSA movement (bit.ly/1E2mOWq) recognizing that CPUs, GP/GPUs, FPGAs and perhaps other new SoC developments will increasingly become part of a dynamic computational system that becomes more broadly accessible to people. the efficiency of any particular solution across the problem lifecycle, from problem realisation to problem solution, will be metric defining the tools in use at any particular time. best, cdm On Friday, December 26, 2014 3:59:52 PM UTC-8, Stefan Karpinski wrote: > > I think we need more languages if only because I think we've barely > started to explore the programming language design space – it's a very > non-convex optimization problem. Sure, some languages are close to local > optima, but you never know what's going to happen when you start combining > things in new ways and find yourself in a totally new part of the space. > Clojure, Go, Rust, Ceylon, Scala, Swift and lots of others are trying > really new and innovative things, some of which are working out really > nicely. People need to try these things if we're going have better tools. >
