By: Stubabe (Stubabe.delete@this.nospam.com), May 16, 2013 11:24 am
Room: Moderated Discussions
RichardC (tich.delete@this.pobox.com) on May 15, 2013 3:26 am wrote:
> Exophase (exophase.delete@this.gmail.com) on May 14, 2013 8:24 pm wrote:
> > Here's another set of tests:
> >
> > http://forums.anandtech.com/showthread.php?t=2274887
> >
> > Enabling HT even on a high end 4 core processor shows tangible benefit in half of the eight games tested.
>
> The conclusion at that link "Hyperthreading neither helps nor hurts when gaming".
> It wins a few, it loses a few.
>
>
The point is it does win a few - which you claim it doesn't. The are plenty of games that are almost completely GPU bound or run at much higher minimum FPS than the monitor can show so can we conclude that high clock speeds, multicore, and high single thread IPC don't help the desktop either?
Considering newer games tend to make better utilisation than older ones is it not more reasonable to argue that those games (written then high core/thread counts were rarer) were not optimised for threading. Especially, since most games pitch to the lowest common hardware config they can tolerate without significantly impacting performance (Game Devs don't make money by excluding most of their potential customer base).
Please don't allow the failures of software devs (that have to pitch to 5year old hardware and release code by last Tuesday) to exploit new features to dismiss them out of hand.
I write desktop code and it scales very well thank you, SMT multicore or whatever. Then again it is no longer my primary source of income so I have time to fiddle.
> Exophase (exophase.delete@this.gmail.com) on May 14, 2013 8:24 pm wrote:
> > Here's another set of tests:
> >
> > http://forums.anandtech.com/showthread.php?t=2274887
> >
> > Enabling HT even on a high end 4 core processor shows tangible benefit in half of the eight games tested.
>
> The conclusion at that link "Hyperthreading neither helps nor hurts when gaming".
> It wins a few, it loses a few.
>
>
The point is it does win a few - which you claim it doesn't. The are plenty of games that are almost completely GPU bound or run at much higher minimum FPS than the monitor can show so can we conclude that high clock speeds, multicore, and high single thread IPC don't help the desktop either?
Considering newer games tend to make better utilisation than older ones is it not more reasonable to argue that those games (written then high core/thread counts were rarer) were not optimised for threading. Especially, since most games pitch to the lowest common hardware config they can tolerate without significantly impacting performance (Game Devs don't make money by excluding most of their potential customer base).
Please don't allow the failures of software devs (that have to pitch to 5year old hardware and release code by last Tuesday) to exploit new features to dismiss them out of hand.
I write desktop code and it scales very well thank you, SMT multicore or whatever. Then again it is no longer my primary source of income so I have time to fiddle.