By: Brendan (btrotter.delete@this.gmail.com), May 15, 2013 4:10 am
Room: Moderated Discussions
Hi,
RichardC (tich.delete@this.pobox.com) on May 15, 2013 3:20 am wrote:
> Brendan (btrotter.delete@this.gmail.com) on May 15, 2013 2:26 am wrote:
> > Current games are designed for 4 threads or less. This doesn't mean that games can't be designed
> > to use 8 or more threads effectively; it only means that it takes time for software developers
> > to get adapt to changes.
>
> I'm guessing that you aren't actually a software developer.
You guess wrong.
> I am. And I have a
> shorthand for this argument, which I've heard many many times in various forms
> through my career: "It's just a software problem". And it's BS. Software
> performance optimization is hard; parallel programming is hard; if a hardware
> feature has been around for several years and most software isn't using it,
> then it's because the feature is not widely applicable, for whatever reason,
> and it isn't "just a software problem".
I don't have much trouble designing or writing scalable code - it just takes a little foresight in the planning/design stages. From what I can tell, a lot of the people that post on these forums are also able to do parallel programming without whining about it being too hard.
What is hard is trying to retro-fit parallel programming into software that was never designed for it (or trying to retro-fit support for an arbitrary number of threads in software design for "4 threads only"); and this is what causes adoption to take ages.
It's not a software problem - more like a "recycling old stuff rather than redesigning from scratch problem".
- Brendan
RichardC (tich.delete@this.pobox.com) on May 15, 2013 3:20 am wrote:
> Brendan (btrotter.delete@this.gmail.com) on May 15, 2013 2:26 am wrote:
> > Current games are designed for 4 threads or less. This doesn't mean that games can't be designed
> > to use 8 or more threads effectively; it only means that it takes time for software developers
> > to get adapt to changes.
>
> I'm guessing that you aren't actually a software developer.
You guess wrong.
> I am. And I have a
> shorthand for this argument, which I've heard many many times in various forms
> through my career: "It's just a software problem". And it's BS. Software
> performance optimization is hard; parallel programming is hard; if a hardware
> feature has been around for several years and most software isn't using it,
> then it's because the feature is not widely applicable, for whatever reason,
> and it isn't "just a software problem".
I don't have much trouble designing or writing scalable code - it just takes a little foresight in the planning/design stages. From what I can tell, a lot of the people that post on these forums are also able to do parallel programming without whining about it being too hard.
What is hard is trying to retro-fit parallel programming into software that was never designed for it (or trying to retro-fit support for an arbitrary number of threads in software design for "4 threads only"); and this is what causes adoption to take ages.
It's not a software problem - more like a "recycling old stuff rather than redesigning from scratch problem".
- Brendan