The industry expects processors with 64 cores or more will arrive by 2015, forcing the need for parallel software, said David Patterson of the Berkeley Parallel Lab. Although researchers have failed to create a useful parallel programming model in the past, he was upbeat that this time there is broad industry focus on solving the problem.
– Source: EE Times
That lack of a programming model is already a major issue. Where is the equivalent of a Hadoop for multicore? I’m not the only one who’s wondered that. Or is it just a problem that’s too hard to solve
Posted via email from Flashing Neurons!!!
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_b.png?x-id=f2e4d5ee-de84-440f-af90-78085e7dd277)
3 Comments
Do you think the current crazy for functional programming and nested data parallelism won't get us there?
I think signs are pointing in that direction. The abstraction are tough for long time object oriented programmers but younger people seem to be getting a hang of them fast.
Big companies are also putting their muscle there. ie. Microsoft and .Net 4.0
I'd love to know. It's something I'd like to explore given time. Functional programming has been around for a while, but somehow never made it into this realm. There are challenges to be sure. For one a lot of these problems are task parallel not data parallel. What you will see is people not using task parallel approaches to solve data parallel problems.
The other aspect is speed. The libraries, etc that have been developed in Fortran, C, C++ etc are important to big numerical computing.
As for big companies, if any will be successful here, it will be Intel, IBM, HP, since they understand many of these problems more than any, at least on the programming model side
I'd love to know. It's something I'd like to explore given time. Functional programming has been around for a while, but somehow never made it into this realm. There are challenges to be sure. For one a lot of these problems are task parallel not data parallel. What you will see is people not using task parallel approaches to solve data parallel problems.
The other aspect is speed. The libraries, etc that have been developed in Fortran, C, C++ etc are important to big numerical computing.
As for big companies, if any will be successful here, it will be Intel, IBM, HP, since they understand many of these problems more than any, at least on the programming model side