Xeon 533 FSB or P4 800 FSB ? (Corrected post)

[Manish M.]

Hi Guys,

I am planning to build a high performance workstation. So I know I want a
dual chip system with SCSI.

Form Intel's site it appears that Xeon is a better chip BUT the current
Xeon available here in Toronto is mainly 533 MHz FSB.

These days the P4 with 800 FSB AND higher clock rate is very common in fact.
So I am wondering weather I should really bother going with Xeon.

Xeon has parallel instruction (in some instructions) BUT then P4 is higher
FSB.

Also if I build the system on 800 FSB I can get a faster BUS on the
motherboard.

Another problem with Xeon is (I think not sure though) that it has different
pin config so I cannot move to P4 later if I want to .

Any comments ?

-Thanks
Manish M

 

[Nate Edel]

I am planning to build a high performance workstation. So I know I want a
dual chip system with SCSI.

Form Intel's site it appears that Xeon is a better chip BUT the current
Xeon available here in Toronto is mainly 533 MHz FSB.

These days the P4 with 800 FSB AND higher clock rate is very common in fact.
So I am wondering weather I should really bother going with Xeon.

There's very little reason to get the Xeon unless you're going to get duals;
in fact, off the top of my head, I can't think of any single-socket-604
motherboards, only dual-socket-604 ones.

Dual Xeon 2.4ghz is a very attractive price right now, if you're doing work
that can take advantage of duals, but otherwise you'd do much better with a
single, slightly faster 800mhz FSB Pentium IV.

And if you _would_ take advantage of duals, consider dual Opterons; for
(about) the cost of dual Xeon 2.4s, you can get dual Opteron 240s... a
little slower each, but with a lot more room for growth, and a 333mhz FSB
and mobo that should be able to take 400mhz FSB Opterons when those are
available.

 

[John McGaw]

Hi Guys,

I am planning to build a high performance workstation. So I know I want a
dual chip system with SCSI.

Form Intel's site it appears that Xeon is a better chip BUT the current
Xeon available here in Toronto is mainly 533 MHz FSB. snip...
-Thanks
Manish M

My memory, admittedly vague at times, tells me that P4 chips cannot be used
in multi-processor systems as the P3 chips could. Have you considered a
single processor machine with one of the new "extreme" P4 processors? With
the 800mHz bus and huge cache they would seem to have the best of both
worlds for many tasks. Or, you might just consider going to one of the new
AMD 64-bit machines which seem to be really great performers. Performance is
roughly on par with (but often better than) the best Xenon systems depending
on the task. And there are a lot of great new MBs coming out to support
multi-processor configurations. Tom's Hardware http://www6.tomshardware.com/
has had a couple of test articles on various MBs using these chips. Their
most current test found one that they though very highly of.
--
John McGaw
[Knoxville, TN, USA]

Return address will not work. Please
reply in group or through my website:
http://johnmcgaw.com

 

[Muttly]

If you were gonna go for an Athlon FX51, DON'T!!!! Wait a few weeks, maybe a
month or 2. This chip is being replaced. And its getting a slightly new
design that means to upgrade a FX51 will require a new motherboard. The new
FX53 has a different pin arrangement and will start at the same speed as the
51, but later a slower chip and faster ones will come online. When the 53 is
released the 51 will be finishing production as far as I know. While there
is no doubt its an excellent chip just be aware that its out of date a few
weeks after its release.
Martin.

Hi Guys,

I am planning to build a high performance workstation. So I know I want a
dual chip system with SCSI.

Form Intel's site it appears that Xeon is a better chip BUT the current
Xeon available here in Toronto is mainly 533 MHz FSB. snip...
-Thanks
Manish M

My memory, admittedly vague at times, tells me that P4 chips cannot be used
in multi-processor systems as the P3 chips could. Have you considered a
single processor machine with one of the new "extreme" P4 processors? With
the 800mHz bus and huge cache they would seem to have the best of both
worlds for many tasks. Or, you might just consider going to one of the new
AMD 64-bit machines which seem to be really great performers. Performance is
roughly on par with (but often better than) the best Xenon systems depending
on the task. And there are a lot of great new MBs coming out to support
multi-processor configurations. Tom's Hardware http://www6.tomshardware.com/
has had a couple of test articles on various MBs using these chips. Their
most current test found one that they though very highly of.
--
John McGaw
[Knoxville, TN, USA]

Return address will not work. Please
reply in group or through my website:
http://johnmcgaw.com

 

[Robert Myers]

Hi Guys,

I am planning to build a high performance workstation. So I know I want a
dual chip system with SCSI.

Form Intel's site it appears that Xeon is a better chip BUT the current
Xeon available here in Toronto is mainly 533 MHz FSB.

These days the P4 with 800 FSB AND higher clock rate is very common in fact.
So I am wondering weather I should really bother going with Xeon.

Xeon has parallel instruction (in some instructions) BUT then P4 is higher
FSB.

Also if I build the system on 800 FSB I can get a faster BUS on the
motherboard.

If you want a dual chip system, you're stuck with xeon. If you
rummage through Specbench CPU2000 results for HP and Intel:

http://see.sun.com/Apps/DCS/mcp?r=700428rR477fe0120003N1k0428rR0mLAKVLAe_

you can draw your own conclusion about relative performance. Of the
published results, the P4 3.2GHz 800MHz FSB is just about neck and
neck with Xeon 3.2GHz 533MHz FSB if the xeon has the 1 Megabyte L3
Cache. Dell Precision workstation with P4EE womps both of them on
CPU2000, which is a good predictor for technical workstations. That
is to say, a bigger cache makes up for a faster bus, but a bigger
cache with a faster bus is best of all. What a surprise. For the
P4EE you get about a 20% boost in performance for a single processor.
With hyperthreading and all, price no object, that would probably be
my choice. Be sure the credit line on your platinum card is in good
shape.

Another problem with Xeon is (I think not sure though) that it has different
pin config so I cannot move to P4 later if I want to .

You're asking about dual Xeon's and thinking about upgrading already?
If you're not going to wait for Prescott, don't even think about the
future. Make your decision on what you can buy today.

RM

 

[*Vanguard*]

in news:[email protected]:

My memory, admittedly vague at times, tells me that P4 chips cannot
be used in multi-processor systems as the P3 chips could. Have you
considered a single processor machine with one of the new "extreme"
P4 processors? With the 800mHz bus and huge cache they would seem to
have the best of both worlds for many tasks. Or, you might just
consider going to one of the new AMD 64-bit machines which seem to be
really great performers. Performance is roughly on par with (but
often better than) the best Xenon systems depending on the task. And
there are a lot of great new MBs coming out to support
multi-processor configurations. Tom's Hardware
http://www6.tomshardware.com/ has had a couple of test articles on
various MBs using these chips. Their most current test found one that
they though very highly of.

"The Pentium 4 Extreme Edition is a reconfigured Xeon server chip
designed to work in a desktop, ...", according to InfoWorld
(http://snurl.com/39wd).

"Intel was able to beef up the caches quickly because the new Pentium 4
is actually the same basic chip as the Xeon MP with 2MB of level-three
cache, a chip for multiprocessor servers that has been on the market for
months.", according to News.com (http://snurl.com/39wg) and ZdNet
(http://snurl.com/39wh).

The Pentium 4 Extreme is *NOT* Intel's next Intel Pentium 5 (dubbed
Prescott). It's a repackaged processor they already had! They wanted
to fold the Xeon family under the much better known Pentium 4 brand
name. You haven't gotten used to Microsoft's hype yet? Athlon finally
comes out with their 65-bit processor and steals the news, so Microsoft
takes an existing 64-bit processor that they've had for awhile and slaps
on a new name to give it some press to dampen Athlon's thunder. They
wanted something to announce for the Christmas crowd.

If you don't have applications and games that actually support dual
processors then it's a waste of money to build a system for software you
don't even have yet or won't be using. Same goes for Intel's
hyperthreading (when you hit the store, how many software boxes do you
see "P4 Hyperthreaded Enabled"?). And some applications, and
*especially* games will NOT run with dual processors and/or
hyperthreading. You have to specify that the game's process run only
under one processor or hyperthreaded model (so the other processor sits
idle).

 

[daytripper]

"The Pentium 4 Extreme Edition is a reconfigured Xeon server chip
designed to work in a desktop, ...", according to InfoWorld
(http://snurl.com/39wd).

Well, now that it's public....It's a Gallatin processor with the MP 'nads
castrated...

/daytripper (It's all marketing magic ;-)

 

[jamotto]

Hi Guys,

I am planning to build a high performance workstation. So I know I want a
dual chip system with SCSI.

Form Intel's site it appears that Xeon is a better chip BUT the current
Xeon available here in Toronto is mainly 533 MHz FSB.

These days the P4 with 800 FSB AND higher clock rate is very common in fact.
So I am wondering weather I should really bother going with Xeon.

Xeon has parallel instruction (in some instructions) BUT then P4 is higher
FSB.

Also if I build the system on 800 FSB I can get a faster BUS on the
motherboard.

Another problem with Xeon is (I think not sure though) that it has different
pin config so I cannot move to P4 later if I want to .

Any comments ?

-Thanks
Manish M

You should look into the P4's with the HT (hyperthreading tech.) HT
seem's to be very similar to Parallel Instruction execution. I think
you would also get the 800MHz FSB. The 800MHz FSB will only help you
if you use memory intensive apps. If you are running ONLY business
apps then AMD's Athlon XP's might be better/cheaper. Go here
http://www.rojakpot.com/ to see some benchmarks.
If money is no object then look into the Athlon 64 chips.

Also unless you need to daisy chain 7-14 drives or the queueing
abilities. IDE ATA133 should be cheaper and most Motherboards have
RAID controllers that support levels 0 and 1. Or look into SATA150
RAID.

good luck building your computer

 

[Nate Edel]

"The Pentium 4 Extreme Edition is a reconfigured Xeon server chip
designed to work in a desktop, ...", according to InfoWorld
(http://snurl.com/39wd).

The Pentium 4 Extreme is *NOT* Intel's next Intel Pentium 5 (dubbed
Prescott).
Right.

It's a repackaged processor they already had!

Yes and no; it's old in the sense that it's the Xeon technology repackaged,
but it's new in the sense that (A) it runs at 800fsb and (B) it's socket 478
not socket 604.

see "P4 Hyperthreaded Enabled"?). And some applications, and
*especially* games will NOT run with dual processors and/or
hyperthreading. You have to specify that the game's process run only
under one processor or hyperthreaded model (so the other processor sits
idle).

I've yet to see a game that won't run on a dual or hyperthreaded system;
they rarely take advantage of the second process/threads (Quake3 and other
games based on the Q3 engine are a notable case that does) and may even run
slightly slower if hyperthreading is on (though this is unusual) but I've
never seen one that simply wouldn't run.

 

[David Schwartz]

[snip stuff about the P4 extreme that I agree with]

If you don't have applications and games that actually support dual
processors then it's a waste of money to build a system for software you
don't even have yet or won't be using.

No specific application support is needed for dual processors. The OS
has sufficient support to get much of the benefit of dual processors whether
or not the application was designed with dual processors in mind.

Same goes for Intel's
hyperthreading (when you hit the store, how many software boxes do you
see "P4 Hyperthreaded Enabled"?).

No need for the software to know about HT. The OS makes it mostly
invisible. There are some very obscure exceptions (for example, if the
application implements its own spinlocks), but they just sap performance a
bit, they're not fatal.

And some applications, and
*especially* games will NOT run with dual processors and/or
hyperthreading.

I have heard only a single example of a program that didn't work
correctly on an SMP machine and that wasn't a game. I also know of one
driver, but an alternate driver was available. Do you have any examples to
back this up?

You have to specify that the game's process run only
under one processor or hyperthreaded model (so the other processor sits
idle).

The other processor will not sit idle. It will do all the other things
that need to be done. The graphics driver will use it. The disk driver will
use it. The network driver will use it. In fact, you may see a very
significant benefit as the game application isn't constantly interrupted to
service peripherals.

DS

 

[kony]

On Thu, 4 Dec 2003 23:56:22 -0800, "David Schwartz"

[snip stuff about the P4 extreme that I agree with]

If you don't have applications and games that actually support dual
processors then it's a waste of money to build a system for software you
don't even have yet or won't be using.

No specific application support is needed for dual processors. The OS
has sufficient support to get much of the benefit of dual processors whether
or not the application was designed with dual processors in mind.

Nope, most people aren't running more than one processor-intensive
application at a time, so the 2nd CPU would be barely used if the
(whole purpose of the workstation) application support isn't there.
For those who are running multiple intensive applications the better
solution is a 2nd system.

The other processor will not sit idle. It will do all the other things
that need to be done. The graphics driver will use it. The disk driver will
use it. The network driver will use it. In fact, you may see a very
significant benefit as the game application isn't constantly interrupted to
service peripherals.

In fact, the benefit is slim-to-none, even a performance decrease in
Quake 3: http://www6.tomshardware.com/cpu/20030811/dual_xeon-12.html

Dual CPUs are ideal for a high-end workstation running the right apps,
but not worthwhile for any box appropriately called a "PC".


Dave

 

[David Schwartz]

Dual CPUs are ideal for a high-end workstation running the right apps,
but not worthwhile for any box appropriately called a "PC".

I have strong anecdotal evidence to the contrary. In a previous thread
on a similar subject, numerous people posted that they've switched to
multiprocessor PCs for normal desktop use and have no intention of ever
switching back. The difference is huge.

DS

 

[Nate Edel]

In fact, the benefit is slim-to-none, even a performance decrease in
Quake 3: http://www6.tomshardware.com/cpu/20030811/dual_xeon-12.html

That's not an apples-to-apples comparison; they're comparing a 3.2ghz,
800FSB chip to a pair of 3.06ghz 533FSB chips.

Given a 50% improvement in FSB rate, it doesn't surprise me that the
difference in speed is significantly greater than the ~5% increase in core
clock rate. Memory bandwidth makes a _big_ difference. (As you'll also note
in some cases with the increase in speed between the 512k and 1M cache
Xeons)

Apples-to-Apples would be Xeon 3.06/533 vs. P4 3.06/533 on a comparable
motherboard chipset.

 

[Nate Edel]

I have strong anecdotal evidence to the contrary. In a previous thread
on a similar subject, numerous people posted that they've switched to
multiprocessor PCs for normal desktop use and have no intention of ever
switching back. The difference is huge.

The difference is throughput vs. latency; duals rarely scale linearly for
throughput on desktop apps or games, which is what most benchmarks measure.

They often have an _easily_ noticeable effect on latency.

 

[John McGaw]

Hi Guys,

I am planning to build a high performance workstation. So I know I want a
dual chip system with SCSI.

Form Intel's site it appears that Xeon is a better chip BUT the current
Xeon available here in Toronto is mainly 533 MHz FSB. snip...
Manish M

This article, along with the eight MB comparison article at Tom's Hardware I
referenced before might provide some food for thought:

http://www.extremetech.com/article2/0,3973,1403009,00.asp

--
John McGaw
[Knoxville, TN, USA]

Return address will not work. Please
reply in group or through my website:
http://johnmcgaw.com

 

[kony]

I have strong anecdotal evidence to the contrary. In a previous thread
on a similar subject, numerous people posted that they've switched to
multiprocessor PCs for normal desktop use and have no intention of ever
switching back. The difference is huge.

DS

Most often people are using an older, slower system, then they do this
upgrade... of course it's faster, a single CPU system would be too.

Your "strong anecdotal evidence" is voodoo superstition _IF_ it isn't
based on benchmarks appropriate to the use of the system. There are
benchmarks of real applications that show clearly, only a minor
performance increase (or decrease) in most uses. These are not
isolated synthetic benchamarks, but reproducible (and reproduced) many
many times. If you have benchmarks to the contrary that are confirmed
by a 3rd party then supply them, we're always hungry for more data.


Dave

 

[David Schwartz]

Most often people are using an older, slower system, then they do this
upgrade... of course it's faster, a single CPU system would be too.

Not in the cases I'm personally familiar with. In these cases, a person
inherits a server that was obsoleted and uses it a desktop machine. I have a
dual P3-1Ghz and dual P3-750 machine that I inherited in just this way.
They're more usable desktops than a single CPU P4-3Ghz.

Your "strong anecdotal evidence" is voodoo superstition _IF_ it isn't
based on benchmarks appropriate to the use of the system. There are
benchmarks of real applications that show clearly, only a minor
performance increase (or decrease) in most uses. These are not
isolated synthetic benchamarks, but reproducible (and reproduced) many
many times. If you have benchmarks to the contrary that are confirmed
by a 3rd party then supply them, we're always hungry for more data.

How do you benchmark productivity and frustration? Seriously, sit down
in front of a dual-CPU desktop and try some normal tasks. You will rapidly
by astonished by the absence of frustrating periods of non-responsiveness
that you didn't even realize were there. Kind of like how you don't notice
cars exhaust until you come back from a place that has no cars.

DS

 

[kony]

Not in the cases I'm personally familiar with. In these cases, a person
inherits a server that was obsoleted and uses it a desktop machine. I have a
dual P3-1Ghz and dual P3-750 machine that I inherited in just this way.
They're more usable desktops than a single CPU P4-3Ghz.

You have a serious configuration problem then, a single P4 3Ghz will
run circles around a dual P3 750-1000 even when running multiple
high-CPU demand apps.

If you have applications that inproperly assign high priority that
would explain it, but the "fix" is not more CPUs, it's decent
software.

How do you benchmark productivity and frustration? Seriously, sit down
in front of a dual-CPU desktop and try some normal tasks. You will rapidly
by astonished by the absence of frustrating periods of non-responsiveness
that you didn't even realize were there. Kind of like how you don't notice
cars exhaust until you come back from a place that has no cars.

The systems must have very significant configuration problems to see
"non-responsiveness". On a properly working single-CPU system, the
CPU can run at 100% load on a task and be immediately responsive on
another, so long as the former isn't inproperly assigned a priority
higher than it should have. This is not a performance benefit from
the 2nd CPU in normal use, it's a software problem that still isn't
fixed.

If you simply cannot do without the buggy software or would rather
throw $$$ into a dual CPU system than the time to resolve the problem,
that's your choice, if your time is THAT valuable it might even be the
right choice for that specific situation, but it is NOT a performance
benefit per se, it is a pseudo-patch for bad code.


Dave

 

[David Schwartz]

The systems must have very significant configuration problems to see
"non-responsiveness". On a properly working single-CPU system, the
CPU can run at 100% load on a task and be immediately responsive on
another, so long as the former isn't inproperly assigned a priority
higher than it should have. This is not a performance benefit from
the 2nd CPU in normal use, it's a software problem that still isn't
fixed.

I should point that I'm specifically talking about PC hardware running
Windows operating systems.

DS

 

[*Vanguard*]

in news:[email protected]:

No specific application support is needed for dual processors.
The OS has sufficient support to get much of the benefit of dual
processors whether or not the application was designed with dual
processors in mind.

If all you want is a faster mass storage subsystem, use RAID. From what
I've seen of the benchmarks at tomshardware.com, programs that don't use
nor cannot use dual processors don't benefit much (and, like you said,
it's probably the OS that got sped up so the apps tag along on the coat
tails of the OS speedup). Lots of money for little benefit. Remember
to not just look at the numbers but the percentage of difference in
performance versus the percentage of difference you pay.

I have heard only a single example of a program that didn't work
correctly on an SMP machine and that wasn't a game. I also know of one
driver, but an alternate driver was available. Do you have any
examples to back this up?

I'm still playing Thief Gold and Thief 2. I like stealth. In a forum
that I frequent, other users got new dual processor systems and were
forced to assign the game to just one processor to get the game to play.
Since I haven't bothered wasting lots of my money on little performance
gain (i.e., no real bang for the buck), I personally haven't bothered
with dual processor systems. I'm not running CAD, file servers, huge
databases with thousands of consecutive user connects, etc.

The other processor will not sit idle. It will do all the other
things that need to be done. The graphics driver will use it. The
disk driver will use it. The network driver will use it. In fact, you
may see a very significant benefit as the game application isn't
constantly interrupted to service peripherals.

Of the other games that I play, they don't much go back to the hard
drive. Usually you get to a spot where there is a separation in the
gameplay, like changing levels or missions and there's a big load at
that time from the hard drive. Otherwise, during gameplay, I see the
hard drive light flash very little.