At 08:41 AM 2/25/00, John Burke wrote:
>http://www.3kworld.com/newsroom.asp?sit_PK=&appmode=itemDetail&news_pk=1971
Relative to HP 3000 Performance Unit.
Rating
13.2 997/200 base: 13.2
23.7 997/400 gain: 10.5
39.0 997/600 gain: 15.3
36.4 997/800 gain: -2.6
48.4 997/1000 gain: 12.0
52.3 997/1200 gain: 3.9
So 13.2 units is what two CPUs by themselves can do. OK. Adding two more
isn't as efficient (which is normal and expected) so a gain of 10.5 looks
reasonable moving up to 4 CPUs; good so far. :)
What happened with the 997/600? Do they just have more fun when there are
six processors in the box? I doubt it. So where did a gain of 15.3 come
from?! And what's up with the 997/800?! 2.6 points *less* powerful than
the 997/600? I wonder if the sales guys have noticed no one wants 8 CPUs
in their 997. ;) My guess is that it's a typo on the 997/600. Perhaps
the rating should be 30.9 instead of 39.0. That would make the gains for
the 997/600 and 997/800 a more plausible 7.2 and 5.5 respectively.
And again... what's up with the gain of 12.0 moving to the 997/1000? That
seems pretty odd too. Should that be 44.8 instead of 48.4? If so then the
997/1000 only gains 8.4 and the 997/1200 steps up by 7.5. But if that's
right then why are the 997/600 and 997/800 smaller increments at only 7.2
and 5.5 assuming the last paragraph is correct (but in any case combining
for an average increase of 6.35 each)?
Perhaps there are more problems with the rating numbers on the 997 systems
in that table than these... If so, I don't expect to be able to puzzle
them out through this kind of analysis. If the only problem in the table
were believing that 39.0 should be 30.9 and all the numbers looked
plausible, then it would be one thing; i.e., a trivial typo (though in an
annoying place). But with the apparent implausibilities in that data, I
wouldn't trust any of those rating numbers for the 997 series.
"The truth is out there."
"Inquiring minds want to know." :)
--
Jeff Woods
[log in to unmask] (preferred)
[log in to unmask] (deprecated)
