Re: Micro$loth Windoze 95 Badges - how they should be

[publicareauser@swarthmore.edu]

In article <Pine.SOL.3.91.950912215736.18857D-100000@howard>, Patrick
O'Neil <patrick@howard> wrote:


> Actually, it depends on what benchmark software you use.  Floating point, 
> for instance, is not a Pentium strong point, though the P120s and P133s 
> are roughly equivalent in fp with 602s (certainly it is when compared to 
> a 601).

Wow, that's pretty good.  Let's see, last I checked the PPC 602 was
designed to drive video game machines.  A Pentium has the same fp marks as
a video game player?  Boy, I knew it was slow, but not THAT slow!  As for
the PPC 601 vs. the P133, the P133 has SPEC92 fp of about 105 whereas the
PPC 601, 110 Mhz, has SPEC92 fp of about 145.  Deduce what you want from
that, but remember that Intel optimizes their chips for SPEC92, and that
IA chips performed much worse under the new SPEC95 benchmarks (which Intel
has not managed to build a cheat box for yet).

> That said, it is also rather unfair to strike at the P6 for 
> being slower operating a 16 bit app...a PowerPC is also slow with such 
> software, being optimized for 32 bit apps.

PC Week tested a 150 Mhz P6 against a 133 Mhz P5.  The P5 outperformed the
P6 when running the 16 bit programs that are the mainstay of the PC
world.  When running either WIn95 with 32 bit native programs OR Win NT
with 32 bit native programs, the P6 outperformed the P5 by approximately
20%.  But there is a 13% difference in Mhz, so at equivalent clock speeds,
running all of the 32 bit apps, the P6 is less than 10% faster than the
P5!  How is Intel supposed to justify the extremely high price of the P6
with that kind of performance gain?  Answer, they can't.  Which explains
why they have announced that P6's will not ship in quantity for the
desktop market until next summer; there is simply no way that they are
going to be able to convince someone to shell out an extra couple thousand
buck for a machine that is only marginally faster than a P5 (after the
consumer spends money buying all of the 32 bit apps).  Of course, how they
are going to convince consumer to buy it a year from now is beyond me. 
They're probably stalling for time to develop a new marketing technique or
something (Remember how they used to claim that a P6 would be about 125%
faster than a P5?  Now they'd be happy just to get it to 25% faster...). 
A 10% performance gain now would not even come close to catching up with
PPC, a 10% performance gain a YEAR from now...they might as well not even
bother.  Don't get me wrong, I have NO DOUBT that Intel will continue to
be a successful company since they control such a large percentage of the
market share and since they have such good marketing (both Microsoft and
Intel have extremely clever management...they realized long ago that they
don't actually need to put much, if any, money into R & D as long as they
make it up by putting hundreds of millions into advertising) but as far as
the technology race goes, it seems they are quickly losing ground.