By: juanrga (nospam.delete@this.juanrga.com), January 10, 2015 4:14 pm
Room: Moderated Discussions
Eric Bron (eric.bron.delete@this.zvisuel.privatefortest.com) on January 7, 2015 7:41 am wrote:
> > Manycores use throughput-optimized cores, multicores use latency-optimized cores.
>
> KNL cores are both, aren't they ?
Overall, the KNL cores are throughput optimized and designed to compete against Nvidia/AMD GPGPUs (manycores).
Due to this you will need multicore Xeons (with latency optimized cores: Broadwell/Skylake) for latency sensitive applications.
> btw I was meaning an official definition, not your oversimplified take on it
>
It is a de-facto standard.
> > Manycores use throughput-optimized cores, multicores use latency-optimized cores.
>
> KNL cores are both, aren't they ?
Overall, the KNL cores are throughput optimized and designed to compete against Nvidia/AMD GPGPUs (manycores).
Due to this you will need multicore Xeons (with latency optimized cores: Broadwell/Skylake) for latency sensitive applications.
> btw I was meaning an official definition, not your oversimplified take on it
>
It is a de-facto standard.
