On Thu, 27 Dec 2007 19:28:32 GMT, Robert Redelmeier
<redelm@ev1.net.invalid> wrote:

>In comp.sys.ibm.pc.hardware.chips kony <spam@spam.com> wrote in part:
>>>AFAIR, PSU efficiency is ~50%. 70% for some of the best units.
>>>No way a faster computer gets total work done that much faster.
>>
>> Actually typical efficiency might be 70%, better units now
>> approaching 85%.
>
>Thank you for the correction
>
>> Yes a faster computer really is at least 3X faster than the
>> Via processor based ones that were suggested. They give up
>> all hopes of performance in order to achieve low power.
>
>No, they do not. Yes, the fast box benchmarks 3x. But I said
>"gets total work done that much faster". Does the user get up
>out of the chair in 1/3rd the time? No!

Yes, if we're goig to talk strict numbers then we would have
to say that since the faster system gets the job done
sooner, the user does get up sooner.


>Outside HPC, the only
>thing that speed affects is the user experience -- the faster
>system appears more responsive. She might wait 0.5 second less.
>We pay a heavy price for impatience.

You are only counting uses that would not have justified the
purchase of the more expensive semi-modern system (which is
required, to attain the power figures originally cited). In
other words, If someone bought a newer system to have this
power consumption, they have already _decided_ to either
keep their old PC, making the new purpose mostly a waste, or
abandon it because they consider it's low performance (like
the new proposed system would have) unacceptible.

There just isn't any thought beyond an overly simplistic
"one number is lower" concept here. It ignores all the
reaslistic scenarios and to spend more time on this
half-baked idea is a waste of time (and power, for that
matter, because it is another reason people would leave
their systems on longer reading instead of being done and
turning the system off).