By: David Kanter (dkanter.delete@this.realworldtech.com), June 4, 2013 5:00 pm
Room: Moderated Discussions
> I wouldn't expect them to sabotage Saltwell (that would be crazy), but instead show Silvermont
> in the best possible light. For example show the best-case improvements rather than average
> improvements over a large range of benchmarks,
They showed a geo-mean and disclosed the benchmarks.
> and for iso-power/performance choose the point
> in the graph where the difference is maximal rather than the average over the full operating
> range.
The iso-power numbers were at very sensible power levels that are consistent with phones and tablets. The iso-perf numbers are normalized to whatever the Atom core could hit (I can't recall the specific model).
I agree it would be nice to see performance as power sweeps from say, 100mW-1.5W or something. But that would have been more complex and probably too much of a give away.
>Simply showing the power/performance graphs for Saltwell and Silvermont would be far
> more informative and interesting but I'm sure there is a reason they weren't shown.
Probably because they don't want to give away too much information to competitors.
> > Most of Intel's test numbers were comparing with Saltwell, and those numbers alone
> > paint a strong picture. As for comparisons between Saltwell and various ARM cores,
> > I'd feel better using third party tests - fortunately those exist.
>
> I suppose from those ARM calculated the projected Silvermont results.
Except that the comparisons between Saltwell and Silvermont were point comparisons, and IIRC, ARM was showing curves. I'm not sure how accurate the interpolation is going to be.
David
> in the best possible light. For example show the best-case improvements rather than average
> improvements over a large range of benchmarks,
They showed a geo-mean and disclosed the benchmarks.
> and for iso-power/performance choose the point
> in the graph where the difference is maximal rather than the average over the full operating
> range.
The iso-power numbers were at very sensible power levels that are consistent with phones and tablets. The iso-perf numbers are normalized to whatever the Atom core could hit (I can't recall the specific model).
I agree it would be nice to see performance as power sweeps from say, 100mW-1.5W or something. But that would have been more complex and probably too much of a give away.
>Simply showing the power/performance graphs for Saltwell and Silvermont would be far
> more informative and interesting but I'm sure there is a reason they weren't shown.
Probably because they don't want to give away too much information to competitors.
> > Most of Intel's test numbers were comparing with Saltwell, and those numbers alone
> > paint a strong picture. As for comparisons between Saltwell and various ARM cores,
> > I'd feel better using third party tests - fortunately those exist.
>
> I suppose from those ARM calculated the projected Silvermont results.
Except that the comparisons between Saltwell and Silvermont were point comparisons, and IIRC, ARM was showing curves. I'm not sure how accurate the interpolation is going to be.
David
