By: Rohit (a.delete@this.b.c), April 25, 2013 8:21 am
Room: Moderated Discussions
default (no.delete@this.thanks.com) on April 24, 2013 9:29 am wrote:
> > I am surprised that this part is too power hungry to go into Ultrabooks. Once you
> > bring the DRAM close by, shouldn't it decrease the IO power? Any thoughts why?
> >
>
> It's probably due to the fact that the chip would never be able to fully utilize the eDRAM and still
> keep the 17w TDP. There are already cases of IVB Ultrabooks throttling when you fully-load the GPU and
> CPU (i.e. games), and that's just HD 4000. Powering something with 2.5x the computational throughput
> on the exact same process node would be impossible (or wasted if you truly caged it inside 17w).
>
> The entire stated purpose of GT3 on Ultrabooks is to improve the performance/watt of
> the GPU by going wider. You can be sure that it won't be clocked anywhere near the
> speed of the GT3e parts - performance will be faster than HD 4000, but not top-end.
>
>
That makes sense. Thanks!
> > I am surprised that this part is too power hungry to go into Ultrabooks. Once you
> > bring the DRAM close by, shouldn't it decrease the IO power? Any thoughts why?
> >
>
> It's probably due to the fact that the chip would never be able to fully utilize the eDRAM and still
> keep the 17w TDP. There are already cases of IVB Ultrabooks throttling when you fully-load the GPU and
> CPU (i.e. games), and that's just HD 4000. Powering something with 2.5x the computational throughput
> on the exact same process node would be impossible (or wasted if you truly caged it inside 17w).
>
> The entire stated purpose of GT3 on Ultrabooks is to improve the performance/watt of
> the GPU by going wider. You can be sure that it won't be clocked anywhere near the
> speed of the GT3e parts - performance will be faster than HD 4000, but not top-end.
>
>
That makes sense. Thanks!