By: Megol (golem960.delete@this.gmail.com), May 13, 2017 1:26 pm
Room: Moderated Discussions
Heikki kultala (heikki.kultala.delete@this.gmail.com) on May 2, 2017 1:42 pm wrote:
> Here is example what you get when you go from small 2-wide in-order to small 2-wide OoOE:
>
> Silvermont vs in-order atom
I thought I responded to this ages ago?
It isn't a fair comparison.
Newer process, larger L2 caches, faster memory, better system interface, better prefetch, shorter pipeline, better branch prediction and more cores. Most likely faster crypto instructions too.
None of those improvements are due to going to OoO execution.
> Here is example what you get when you go from small 2-wide in-order to small 2-wide OoOE:
>
> Silvermont vs in-order atom
I thought I responded to this ages ago?
It isn't a fair comparison.
Newer process, larger L2 caches, faster memory, better system interface, better prefetch, shorter pipeline, better branch prediction and more cores. Most likely faster crypto instructions too.
None of those improvements are due to going to OoO execution.