By: juanrga (nospam.delete@this.juanrga.com), January 2, 2015 11:21 am
Room: Moderated Discussions
Rob Thorpe (rthorpe.delete@this.robertthorpeconsulting.com) on January 2, 2015 8:20 am wrote:
> http://highscalability.com/blog/2014/12/31/linus-the-whole-parallel-computing-is-the-future-is-a-bunch.html
>
> I can see potential in this. Several journalists can summarize the same set of comment threads in parallel.
> We can ask Linus to post something angry on any thread that needs summarizing. The scalability problem is
> that there's only one Linus... and the journalists don't necessarily know that much about the subject.
Linus' points against manycores are plain wrong. He doesn't even know what a manycore is. I still recall when he pretended that Intel had abandoned manycores with the new Xeon Phi and then I gave him a quote from the KNL chief engineer stating the contrary.
I just published an article about future APUs from AMD. I am preparing a similar article about Nvidia, Intel, IBM, Fujitsu... plans. All them propose manycores for the year ~2020 because there is no other way to improve performance by 50x. Linus can continue ignoring the problems and the targets, the processor engineers cannot.
> http://highscalability.com/blog/2014/12/31/linus-the-whole-parallel-computing-is-the-future-is-a-bunch.html
>
> I can see potential in this. Several journalists can summarize the same set of comment threads in parallel.
> We can ask Linus to post something angry on any thread that needs summarizing. The scalability problem is
> that there's only one Linus... and the journalists don't necessarily know that much about the subject.
Linus' points against manycores are plain wrong. He doesn't even know what a manycore is. I still recall when he pretended that Intel had abandoned manycores with the new Xeon Phi and then I gave him a quote from the KNL chief engineer stating the contrary.
I just published an article about future APUs from AMD. I am preparing a similar article about Nvidia, Intel, IBM, Fujitsu... plans. All them propose manycores for the year ~2020 because there is no other way to improve performance by 50x. Linus can continue ignoring the problems and the targets, the processor engineers cannot.