RichardC (tich.delete@this.pobox.com) on May 13, 2013 3:48 am wrote:

> I think Intel has a plan: they've put considerable effort into tools for creating
> apps with more parallelism, and they would love to see more apps using more cores.
> It just hasn't been an adequate plan. They've also - wisely, I think - moved
> functionality on-chip so that they can keep selling bigger chips at high prices.
>
> The recent trends in PC and tablet market suggest that the plan isn't working
> well enough. But I don't really have a good alternative suggestion for Intel -
> a move from $1000 laptops (with Windows) to $250 tablets (with Android or iOS)
> is bound to be rather disruptive for both Intel and Microsoft.

I think part of the problem is that common PC nowadays has 2C/4T processor (and not so long ago we were at 2C/2T). In the best case, you can have about 2.5 times the performance using 4 threads. Considering turbo boost, even less than that. Realistically, with some sensible effort for not totally embarrassingly parallel case, you can probably expect less than two times performance. I'm not sure that's worth it in most cases.

Also, if you are targeting some processor X with lowish performance and low thread count (== two), it may be easier to do some hacky offloading for one additional thread and get maybe 20-50 % boost from that, instead of supporting even four threads. If your low end target has N hardware threads, motivation to support N threads is a lot higher.

-JLarja