By: Michael S (already5chosen.delete@this.yahoo.com), August 16, 2014 1:54 pm
Room: Moderated Discussions
Maynard Handley (name99.delete@this.name99.org) on August 16, 2014 2:24 pm wrote:
>
> I stated in a previous post that David and I maybe ultimately agreed that the x86 tax was worth
> about two development years, and that continues to me to seem like a good way to view it.
> Does this mean two extra years to create an equivalent performing CPU, or that, *with the same sized team
> and same process*, an ARM or POWER device would lead an x86 device by about two years? I'd say, from the
> very limited evidence we have, both interpretations are reasonable. IF, for example (we'll see soon enough)
> an Apple A8 performs generally at the level of a Broadwell-Y (a flexible metric --- there's absolute single-threaded
> performance, multi-threaded performance, AVX-assisted FLOPS, performance/watt, GPU performance, dynamic
> range of performance etc --- but let's ignore the details for now) one could reasonably argue that the
> x86 complexity tax is the equivalent of about two years in process improvement.
That's also flexible metric because there will be several Broadwell-Y devices - from i7 to i3.
i7 does not really look touchable in any absolute (not per-watt) CPU-related metric, but especially in single-thread.
Well, not just Broadwell, Haswell-Y is almost certainly untouchable as well.
http://browser.primatelabs.com/geekbench3/compare/712255?baseline=603360
Anyway, comparison of Apple A8 vs Brodwell-Y is not exactly the most illuminating, because designs targets differ quite significantly. Broadwell-Y is still Broadwell, which still supposed to collect majority of its revenue in 15-80 W space rather than in 8-11 W space. Also the fact that Broadwell is not much more than a shrink of Haswell can't be ignored.
>
> I stated in a previous post that David and I maybe ultimately agreed that the x86 tax was worth
> about two development years, and that continues to me to seem like a good way to view it.
> Does this mean two extra years to create an equivalent performing CPU, or that, *with the same sized team
> and same process*, an ARM or POWER device would lead an x86 device by about two years? I'd say, from the
> very limited evidence we have, both interpretations are reasonable. IF, for example (we'll see soon enough)
> an Apple A8 performs generally at the level of a Broadwell-Y (a flexible metric --- there's absolute single-threaded
> performance, multi-threaded performance, AVX-assisted FLOPS, performance/watt, GPU performance, dynamic
> range of performance etc --- but let's ignore the details for now) one could reasonably argue that the
> x86 complexity tax is the equivalent of about two years in process improvement.
That's also flexible metric because there will be several Broadwell-Y devices - from i7 to i3.
i7 does not really look touchable in any absolute (not per-watt) CPU-related metric, but especially in single-thread.
Well, not just Broadwell, Haswell-Y is almost certainly untouchable as well.
http://browser.primatelabs.com/geekbench3/compare/712255?baseline=603360
Anyway, comparison of Apple A8 vs Brodwell-Y is not exactly the most illuminating, because designs targets differ quite significantly. Broadwell-Y is still Broadwell, which still supposed to collect majority of its revenue in 15-80 W space rather than in 8-11 W space. Also the fact that Broadwell is not much more than a shrink of Haswell can't be ignored.