By: David Kanter (dkanter.delete@this.realworldtech.com), July 25, 2012 11:11 am
Room: Moderated Discussions
Adrian (a.delete@this.acm.org) on July 25, 2012 4:39 am wrote:
> David Kanter (dkanter.delete@this.realworldtech.com) on July 25, 2012 1:37 am
> wrote:
> > New computational efficiency data shows GPUs with a clear edge over
> CPUs, but
> > the gap is narrowing as CPUs adopt wide vectors (e.g. AVX).
> Surprisingly, a
> > throughput CPU is the most energy efficient processor,
> offering hope for future
> > architectures. Our data also shows some
> advantages of AMD's Bulldozer, and the
> > overhead associated with highly
> scalable server CPUs.
> >
> > Comments and feedback
> > welcome!
> >
>
> > David
>
> The value for Gflops/W that appears in your charts implies that
> the fastest NVIDIA Fermi cards @ 665 DP Gflops are 225 W PCI-E cards (not, e.g.,
> 300 W cards).
>
> Is that correct? I could not find any mention about their power
> consumption in the NVIDIA web site.
Yes, 225W is correct. If you look in some of their detailed spec sheets (PDF form), some of the older ones include power numbers.
Actually, one of my biggest concerns about the 2013 edition is that Nvidia won't publish any power numbers for Kepler...which means I cannot put them on the chart.
I was positively thrilled when I turned up that IBM systems paper that disclosed the POWER7 power dissipation.
David
> David Kanter (dkanter.delete@this.realworldtech.com) on July 25, 2012 1:37 am
> wrote:
> > New computational efficiency data shows GPUs with a clear edge over
> CPUs, but
> > the gap is narrowing as CPUs adopt wide vectors (e.g. AVX).
> Surprisingly, a
> > throughput CPU is the most energy efficient processor,
> offering hope for future
> > architectures. Our data also shows some
> advantages of AMD's Bulldozer, and the
> > overhead associated with highly
> scalable server CPUs.
> >
> > Comments and feedback
> > welcome!
> >
>
> > David
>
> The value for Gflops/W that appears in your charts implies that
> the fastest NVIDIA Fermi cards @ 665 DP Gflops are 225 W PCI-E cards (not, e.g.,
> 300 W cards).
>
> Is that correct? I could not find any mention about their power
> consumption in the NVIDIA web site.
Yes, 225W is correct. If you look in some of their detailed spec sheets (PDF form), some of the older ones include power numbers.
Actually, one of my biggest concerns about the 2013 edition is that Nvidia won't publish any power numbers for Kepler...which means I cannot put them on the chart.
I was positively thrilled when I turned up that IBM systems paper that disclosed the POWER7 power dissipation.
David
