Interesting Article

NTDEV
1

Regarding the Penitum IV
http://www.emulators.com/pentium4.htm

You are currently subscribed to ntdev as: $subst(‘Recip.EmailAddr’)
To unsubscribe send a blank email to leave-ntdev-$subst(‘Recip.MemberIDChar’)@lists.osr.com

2

At 11:39 AM 1/2/2001 -0800, Paul Bunn wrote:

Regarding the Penitum IV
http://www.emulators.com/pentium4.htm

That’s a pretty negative article. I think readers might want to keep in
mind the context of the article writer, which sounds like a developer who
does instruction emulator products.

Personally, I think some people will find things like the greatly improved
memory bandwidth will be a terrific plus for their application performance.

Whenever there is a major architecture change, some people will be happy
and some will be unhappy, usually depending on how optimized the new
architecture is for THEIR specific context.

As a developer, it’s seems pretty certain I will have to write code that’s
optimal for the P4, so my time seems better spent learning about it than
having a religious war about it.

My personal preference would be an architecture change that made systems
never crash, and I would give up a bunch of performance if needed. This
would probably be a system software change though. It seems like I adjust
to whatever performance is, but seem to never adjust to having my work
periodically discarded at random times.

  • Jan

You are currently subscribed to ntdev as: $subst(‘Recip.EmailAddr’)
To unsubscribe send a blank email to leave-ntdev-$subst(‘Recip.MemberIDChar’)@lists.osr.com

3

Actually, I’ve met the author of that article – he actually works at MS (the
emulator stuff is just stuff he does in his own time for “fun”). He’s one of the
brightest guys I’ve ever met, and certainly has no ax to grind against Intel or
any undue bias. It is a rather negative article, but it is highly objective.
This is borne out by my memory benchmark program that you ran for me to compare a
P3-600 against a P4-1.4Ghz. While the P4 offered 2 to 4 times memory bandwidth
(depending on cache utilization) than the P3, in the all-important context-switch
test (at least relevant to NT/Win2K) the P4 was slightly slower than the P3, which
is pathetic since the P4 was running at 800Mhz faster than the P3.

Regards,

Paul Bunn, UltraBac.com, 425-644-6000
Microsoft MVP - WindowsNT/2000
http://www.ultrabac.com

-----Original Message-----
From: Jan Bottorff [mailto:xxxxx@pmatrix.com]
Sent: Tuesday, January 02, 2001 2:18 PM
To: NT Developers Interest List
Subject: [ntdev] Re: Interesting Article

At 11:39 AM 1/2/2001 -0800, Paul Bunn wrote:

Regarding the Penitum IV
http://www.emulators.com/pentium4.htm

That’s a pretty negative article. I think readers might want to keep in
mind the context of the article writer, which sounds like a developer who
does instruction emulator products.

Personally, I think some people will find things like the greatly improved
memory bandwidth will be a terrific plus for their application performance.

Whenever there is a major architecture change, some people will be happy
and some will be unhappy, usually depending on how optimized the new
architecture is for THEIR specific context.

As a developer, it’s seems pretty certain I will have to write code that’s
optimal for the P4, so my time seems better spent learning about it than
having a religious war about it.

My personal preference would be an architecture change that made systems
never crash, and I would give up a bunch of performance if needed. This
would probably be a system software change though. It seems like I adjust
to whatever performance is, but seem to never adjust to having my work
periodically discarded at random times.

You are currently subscribed to ntdev as: $subst(‘Recip.EmailAddr’)
To unsubscribe send a blank email to leave-ntdev-$subst(‘Recip.MemberIDChar’)@lists.osr.com

4

So, the article is saying that my brand new PIII at 1.0Ghz is faster than a
PIV at 1.5.

Gary

You are currently subscribed to ntdev as: $subst(‘Recip.EmailAddr’)
To unsubscribe send a blank email to leave-ntdev-$subst(‘Recip.MemberIDChar’)@lists.osr.com

5

Probably by quite a margin – it all depends on how much the application you are
using is doing bulk memory reads/writes (where the P4 would be *much* better).
The program I referred to in an earlier email is available at:
ftp://ftp.ultrabac.com/pub/utils/bm_mem/x86/bm_mem.zip
The most important figure given (IMHO) is the context-switch nightmare test which
tests a system’s ability to swap 8 bytes between two threads on each processor
(use the /L switch to limit the test to just one processor in an SMP system). I
believe this test is important since it determines how well a system handles in a
multithreaded OS, how fast interrupts can be dispatched/handled, and how well
KM<->UM transitions are handled (each 8 byte copy results in 4 system calls for
synchronization object manipulation). As a guideline, a P4 system (1.4GHz)
reported 0.994 MB/sec.
It’s astounding that a P3-600 can report 7% better score on this test than a
P4-1400.

Regards,

Paul Bunn, UltraBac.com, 425-644-6000
Microsoft MVP - WindowsNT/2000
http://www.ultrabac.com

-----Original Message-----
From: Gary Little [mailto:xxxxx@delphieng.com]
Sent: Tuesday, January 02, 2001 2:48 PM
To: NT Developers Interest List
Subject: [ntdev] Re: Interesting Article

So, the article is saying that my brand new PIII at 1.0Ghz is faster than a
PIV at 1.5.

You are currently subscribed to ntdev as: $subst(‘Recip.EmailAddr’)
To unsubscribe send a blank email to leave-ntdev-$subst(‘Recip.MemberIDChar’)@lists.osr.com

6

That’s what VMS and VAX/Alpha were all about. They lost out to the Unix
philosophy of faster and cheaper at all costs.

Just my $0.02 worth.
Greg

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com]On Behalf Of Jan Bottorff
Sent: Tuesday, January 02, 2001 4:18 PM
To: NT Developers Interest List
Subject: [ntdev] Re: Interesting Article

At 11:39 AM 1/2/2001 -0800, Paul Bunn wrote:
>Regarding the Penitum IV
>http://www.emulators.com/pentium4.htm

That’s a pretty negative article. I think readers might want to keep in
mind the context of the article writer, which sounds like a developer who
does instruction emulator products.

Personally, I think some people will find things like the greatly
improved
memory bandwidth will be a terrific plus for their application
performance.

Whenever there is a major architecture change, some people will be happy
and some will be unhappy, usually depending on how optimized the new
architecture is for THEIR specific context.

As a developer, it’s seems pretty certain I will have to write
code that’s
optimal for the P4, so my time seems better spent learning about it than
having a religious war about it.

My personal preference would be an architecture change that made systems
never crash, and I would give up a bunch of performance if needed. This
would probably be a system software change though. It seems like I adjust
to whatever performance is, but seem to never adjust to having my work
periodically discarded at random times.

  • Jan

You are currently subscribed to ntdev as: xxxxx@pdq.net
To unsubscribe send a blank email to leave-ntdev-$subst(‘Recip.MemberIDChar’)@lists.osr.com

You are currently subscribed to ntdev as: $subst(‘Recip.EmailAddr’)
To unsubscribe send a blank email to leave-ntdev-$subst(‘Recip.MemberIDChar’)@lists.osr.com