🚀 go-pugleaf

King Neptune <dmo...@globalx.net> wrote:


>I think it would be a cold day in hell before you and your industry pundits are proved 
>correct.  Sure MS may have some problems with P6 performance, but don't doubt that Bill 
>Buddy could lull Big Blue to sleep again, and slip out Cairo. 

This is still up in the air.  I was surprised to hear this, but
instead of spamming the net in a "P6 Panic", I went to the Intel
conference and asked an Intel representative.

16 bits apps will not take advantage of the P6's speed, on 95 or NT.
32 bit apps I was told however, DO take advantage of the P6 _under
95_.  I think the PC Week article may have been speculation, Intel
stated that the P6 does not run 16-bit code well, and since it is
known that Win95 contains some 16 bit code, they may have assumed that
Win95 will run poorly.  Anyways, the Intel rep stated 32-bit apps
under 95 and NT do show significant speed gains with the P6.

With 95 however, it depends on the application.  Some may call more
16-bit code, some may call less.  To say Windows 95 is "mostly" 16 bit
is riduclous however, it's actually mostly 32-bit.  But the 16-bit
portions can be critical in terms of speed.  We'll see.

To sum up: Lets wait until we get some actual benchmarks to panic.  I
don't hold the reasoning that this isn't a problem because you can
"just go to NT", as NT resource requirements are still quite high
compared to 95 and there is no guarantee it will run all 95 apps,
specifically games.  P6 may be priced for NT systems with 16 megs NOW,
but as always, it will drop.  And from past experience, I wouldn't be
surprised to see RAM prices stay the same while the P6 costs half as
much as its introduction.


>interface.  It is already in Beta from what I have heard.  MS is also looking at the 

To clarify: Windows NT with the _95_ interface is in beta, and likely
will be released very early '96.  As for the object-oriented NT, MS
stated some time ago that it would be a series of add-ons, rather than
a large "Cairo" roll-out.

>Bill buddy doesn't have the best marketing analyists working for him - because he does.  
>Bill is going to grab a major chunk of cash with this forward and backward WIN95 OS by 
>serving it up to the home market, and then tweek-up NT for the next generation of chips.

What I don't get is how some people see this as a "conspiracy".  Of
course MS would like you to move to NT, it's a more robust OS, and its
ability to work across Power PC/Mips/Alpha/Intel machines just widens
MS's market.

But MS's realizes it needs apps, and people just can't make the
_large_ jump to NT.  So 95 is created.  It has compromises, but it
works, and works well for most machines where NT wouldn't.  You want
to upgrade to NT?  Do it when you get the hardware.  There's nothing
backhanded about this, it's serving your target markets while giving
them an upgrade path they don't even have to _wait for_.  Want a
better Windows 95 when you get the hardware?  No need to wait, it's
already here!

>the Bible...however, I find it hard to understand why an OS with portions of GDI and USER 
>in 16 bit having only 10% performance gain on a chip with twice the transistors? 

The # of transistors does not necessarily translate into performance
gain. The problem is that by changing _modes_ of memory "prediction"
in the P6 causes a large speed hit because of it's 14 pipelines, as
compared to 6 in the P5.  Don't quote me on any technical details
however.

>same problem, if it runs 16 bit apps on a P6 and yes 16 bit apps will dominate for 
>sometime. 

I question this actually.  With MS releasing a 32-bit version of
office with 95, and 16-bit apps performing (already) quite poorly
under NT, it's not in a companies best competitive interests to
release a 16-bit app when it likely will be raked through the coals in
PC reviews and suffer against the competitions 32-bit apps.  MS sets
the standard here, and they'll be the ones out the door first with
32-bit apps.  Others will follow if they want to survive.