Re: P4 2.8 FSB 800 overclock. Any ideas?

One problem is you started with a 2.8ghz versus a 2.4ghz chip. The Northwood
core is hitting the end of its cycle from Intel @ 3.2ghz. Without a choice chip
and serious cooling, hitting 3600mhz+ is not going to happen that often.

You want to hit 275mhz FSB on a 2.8ghz (14 x 200mhz) = 3850mhz. Good luck on
that one without a choice chip and serious cooling.

A 2.4ghz chip (12x200), would hit 3300mhz @ 275mhz FSB. That is a much more
reasonable goal to reach.

Your memory is another issue. You need to gauge memory bandwidth versus core CPU
frequency to obtain optimal performance (which is the goal). While your chip
may have topped out @ FSB242mhz, you at least are able to run your memory at
full bandwidth (synchronously). That's a plus. I would scale back the memory to
spec (233mhz = DDR466mhz = PC3700) since it is only a 9mhz (126mhz gain)
increase and run it with the better timings and see which one overall provides
better performance and efficiency.

Mark

"Frank Weston" <> spewed forth in such a glorious tone
the following gruel:

>Setup:
>P4 2.8 FSB 800 SL6WJ
>ASUS P4P Deluxe
>OCZ DDR PC-3700 Performance Memory 256x2
>
>The best I'm able to do is 3.4 Ghz at FSB 242 with memory at 193.6. I guess
>I should be happy, temps are low, V-core is 1.525, and the system is sitting
>at 43C now after running Prime95 for an hour, but I want more.
>
>No matter what combo of V-core and memory I run, FSB tops out at about 242.
>What's holding me back? I read of systems like this running at 275+. Also,
>memory does not want to run too much more than spec, and then at very
>relaxed timings. Spec is 233 at 2.75 vdimm, 2,3,3,7 (1T). I'm having to
>run 2.5,4,4,8.
>
>What's holding me back? A fan on the 865 perhaps?
>
>
>

P4-3.06ghz @ 3335mhz (FSB=144, Stock 1.55v 44c Max Load)
ATI Radeon 9700 Pro @ 343.5/343.5
SONY GDM-F520 21" Trinitron Monitor
Gigabyte 8PE667 i845PE Motherboard
Corsair PC3500 (DDR433) @ DDR375 2-2-2-6-1T
Maxtor Ultra/133 6mb cache Gold Series 120GB HD
Creative SB Audigy Platinum
16771 in 3DMark 2001SE 330
5715 in 3DMark 2003

"Mark Bell" <> wrote in message
news:...
> One problem is you started with a 2.8ghz versus a 2.4ghz chip. The
Northwood
> core is hitting the end of its cycle from Intel @ 3.2ghz. Without a choice
chip
> and serious cooling, hitting 3600mhz+ is not going to happen that often.

I'm not so much bumping the limit of the chip as I am the FSB limit of the
motherboard. I don't really need or want to run at 3.6 Ghz, but I am
curious as to why I can't get an FSB of greater than 242 when I read of
people attaining 275+ with relative ease. Could it be that the reviewers
are unrealistically hypeing this board?

>
> You want to hit 275mhz FSB on a 2.8ghz (14 x 200mhz) = 3850mhz. Good luck
on
> that one without a choice chip and serious cooling.
>
> A 2.4ghz chip (12x200), would hit 3300mhz @ 275mhz FSB. That is a much
more
> reasonable goal to reach.
>

Again, I don't need to run at 275 FSB. I'm just curious as to why I can't.
I'm reading reviews where people are topping 300 FSB with 3.0 Ghz chips.


> Your memory is another issue. You need to gauge memory bandwidth versus
core CPU
> frequency to obtain optimal performance (which is the goal). While your
chip
> may have topped out @ FSB242mhz, you at least are able to run your memory
at
> full bandwidth (synchronously).

No. As I stated, at 242 FSB I'm running memory at 193.6 which is a 4:5
memory/FSB ratio.



That's a plus. I would scale back the memory to
> spec (233mhz = DDR466mhz = PC3700) since it is only a 9mhz (126mhz gain)
> increase and run it with the better timings and see which one overall
provides
> better performance and efficiency.

Yes, when it comes time to settle in on a configuration, I'll probably do
just that. But, now, I'm pushing to find the limits of the system
everywhere, and the FSB question is the big one.

BTW, you might think about changing your lead in line below. Some people
might not find it amusing and there's no use insulting those who might
seek/offer advice from/to you.

>
> Mark
>
> "Frank Weston" <> spewed forth in such a glorious
tone
> the following gruel:
>
> >Setup:
> >P4 2.8 FSB 800 SL6WJ
> >ASUS P4P Deluxe
> >OCZ DDR PC-3700 Performance Memory 256x2
> >
> >The best I'm able to do is 3.4 Ghz at FSB 242 with memory at 193.6. I
guess
> >I should be happy, temps are low, V-core is 1.525, and the system is
sitting
> >at 43C now after running Prime95 for an hour, but I want more.
> >
> >No matter what combo of V-core and memory I run, FSB tops out at about
242.
> >What's holding me back? I read of systems like this running at 275+.
Also,
> >memory does not want to run too much more than spec, and then at very
> >relaxed timings. Spec is 233 at 2.75 vdimm, 2,3,3,7 (1T). I'm having to
> >run 2.5,4,4,8.
> >
> >What's holding me back? A fan on the 865 perhaps?
> >
> >
> >
>
> P4-3.06ghz @ 3335mhz (FSB=144, Stock 1.55v 44c Max Load)
> ATI Radeon 9700 Pro @ 343.5/343.5
> SONY GDM-F520 21" Trinitron Monitor
> Gigabyte 8PE667 i845PE Motherboard
> Corsair PC3500 (DDR433) @ DDR375 2-2-2-6-1T
> Maxtor Ultra/133 6mb cache Gold Series 120GB HD
> Creative SB Audigy Platinum
> 16771 in 3DMark 2001SE 330
> 5715 in 3DMark 2003

"Frank Weston" <> spewed forth in such a glorious tone
the following gruel:

>
>"Mark Bell" <> wrote in message
>news:.. .
>> One problem is you started with a 2.8ghz versus a 2.4ghz chip. The
>Northwood
>> core is hitting the end of its cycle from Intel @ 3.2ghz. Without a choice
>chip
>> and serious cooling, hitting 3600mhz+ is not going to happen that often.
>
>I'm not so much bumping the limit of the chip as I am the FSB limit of the
>motherboard. I don't really need or want to run at 3.6 Ghz, but I am
>curious as to why I can't get an FSB of greater than 242 when I read of
>people attaining 275+ with relative ease. Could it be that the reviewers
>are unrealistically hypeing this board?

Because as the FSB increases, so does your core CPU speed. I am sure the board
can hit an FSB of 275+, but only with a CPU that can obtain that speed.
Core_CPU_Speed = FSB x multiplier, so you are stuck, potentially, at your
chips maximum overclocking potential. You are asking your 2.8ghz chip to do
3850mhz. Using a chip with a lower multiplier would afford you a greater chance
of hitting an FSB>=275mhz. I assume you are monitoring your temps vigorously.

What I am basically saying is it is most likely your CPU, not your motherboard
that is limiting your FSB.

>> You want to hit 275mhz FSB on a 2.8ghz (14 x 200mhz) = 3850mhz. Good luck
>on
>> that one without a choice chip and serious cooling.
>>
>> A 2.4ghz chip (12x200), would hit 3300mhz @ 275mhz FSB. That is a much
>more
>> reasonable goal to reach.
>>
>
>Again, I don't need to run at 275 FSB. I'm just curious as to why I can't.
>I'm reading reviews where people are topping 300 FSB with 3.0 Ghz chips.

I never said you, "need," to run at 275mhz FSB. I said you want to.

As for 3ghz chips, I find that hard to believe, unless they are running serious
cooling (phase, water, etc...) and a choice chip. A 3.0ghz chip running at
300mhz FSB = 4.5ghz. That's the extreme exception, not the norm. I'm not saying
it is impossible; it is just improbable.

>> Your memory is another issue. You need to gauge memory bandwidth versus
>core CPU
>> frequency to obtain optimal performance (which is the goal). While your
>chip
>> may have topped out @ FSB242mhz, you at least are able to run your memory
>at
>> full bandwidth (synchronously).
>
>No. As I stated, at 242 FSB I'm running memory at 193.6 which is a 4:5
>memory/FSB ratio.

You are running PC3700 memory which is rated at 233mhz FSB. It won't do 1:1
even with a mild OC to 242mhz? Give it a little more juice if needed. I can't
see good OCZ not hitting an extra 9mhz with mildly relaxed timings. 4:5 timings
with your FSB @ 242mhz running FSB 233mhz @ 193.6mhz is a waste of good memory.

OR....

As I said, scale back to 233mhz FSB then and run it at 1:1 and see what your
benchmarks produce. The objective is optimal performance, not the highest
core clock speed (which does not always corrolate to optimal performance).

>That's a plus. I would scale back the memory to
>> spec (233mhz = DDR466mhz = PC3700) since it is only a 9mhz (126mhz gain)
>> increase and run it with the better timings and see which one overall
>provides
>> better performance and efficiency.
>
>Yes, when it comes time to settle in on a configuration, I'll probably do
>just that. But, now, I'm pushing to find the limits of the system
>everywhere, and the FSB question is the big one.

Well, not all CPUs will hit those extreme FSBs. Just as some get a choice
chip, others get one that will not overclock that much when tested. It
happens. The Northwood core is hitting its EOL. 3.2-3.4ghz is all Intel
expected out of it, so you can't be surprised when some chips (even D1
steppings) do not hit 3500mhz+. 3388mhz is respectable.

Make sure to run Prime95 AND CPUBurn (burnmmx) at the same time to fully
load the chip. Prime95, on it's own, will not give the new multithreaded
CPUs a full 100% load. It helps to throw CPUBurn on top of it to obtain a
sustained 100% load. I let it run for at least 4hrs, and usually 24hrs when I am
finalizing my CPU settings.

>
>BTW, you might think about changing your lead in line below. Some people
>might not find it amusing and there's no use insulting those who might
>seek/offer advice from/to you.

Well, then they need a better sense of humor or the need to not take life so
seriously. >;-P I've been using that for, oh, 9yrs now.

Mark


P4-3.06ghz @ 3335mhz (FSB=144, Stock 1.55v 44c Max Load)
ATI Radeon 9700 Pro @ 343.5/343.5
SONY GDM-F520 21" Trinitron Monitor
Gigabyte 8PE667 i845PE Motherboard
Corsair PC3500 (DDR433) @ DDR375 2-2-2-6-1T
Maxtor Ultra/133 8mb cache Gold Series 120GB HD
Creative SB Audigy Platinum
16771 in 3DMark 2001SE 330
5715 in 3DMark 2003

"Mark Bell" <> wrote

> Because as the FSB increases, so does your core CPU speed. I am sure the
board
> can hit an FSB of 275+, but only with a CPU that can obtain that speed.
> Core_CPU_Speed = FSB x multiplier, so you are stuck, potentially, at your
> chips maximum overclocking potential. You are asking your 2.8ghz chip to
do
> 3850mhz. Using a chip with a lower multiplier would afford you a greater
chance
> of hitting an FSB>=275mhz. I assume you are monitoring your temps
vigorously.
>
> What I am basically saying is it is most likely your CPU, not your
motherboard
> that is limiting your FSB.
>

check this: http://www.hexus.net/review.php?review=556&page=10
"it's an FSB monster. 285FSB came and went with utter stability; a testament
to Intel's impressive engineering prowess. Fast and relatively inexpensive,
just how we like it. Recommended."

These guys were testing with a 3.0 FSB 800 processor.

Or, how about this:
http://www.lostcircuits.com/motherbo...p4p800/9.shtml
"Despite the limited CPU voltage range, we had no problems running the
P4P800 overclocked to 3.6 or 3.7 GHz processor speed. In terms of overall
board stability, the P4P800 was rock solid up to 300 MHz"

These guys were also running a 3.0 FSB 800 processor. I don't put too much
stock in reviews, but these guys are either totally full of crap or I've got
something holding my system back in the FSB department.

> As for 3ghz chips, I find that hard to believe, unless they are running
serious
> cooling (phase, water, etc...) and a choice chip. A 3.0ghz chip running at
> 300mhz FSB = 4.5ghz. That's the extreme exception, not the norm. I'm not
saying
> it is impossible; it is just improbable.

Although I'm not shooting for 4.5 Ghz, heat seems to be no problem. At
3.4Ghz, with Prime95 running under most of my applications, my CPU temp is
44 C with a motherboard temp of 35 C. That's cooler than idle on any AMD
setup I've ever run. At idle, I'm running 27C. BTW I'm running air cooling
with a Zalman 7000, and the sound is undetectable.


> As I said, scale back to 233mhz FSB then and run it at 1:1 and see what
your
> benchmarks produce. The objective is optimal performance, not the highest
> core clock speed (which does not always corrolate to optimal performance).

I thought this was a good idea, so I tried it. Sandra benchmarks reflected
a slower CPU, and memory bandwidth didn't increase by any significant
amount. Worse yet, I got tearing and artifacts when running 3DMark3 and in
some of my games. When I run the memory at the equivalent of maxed PC3200,
these problems go away. I'm back to 242 FSB and a 4:5 divider. Maybe I
should have just bought cheaper memory.

> Make sure to run Prime95 AND CPUBurn (burnmmx) at the same time to fully
> load the chip. Prime95, on it's own, will not give the new multithreaded
> CPUs a full 100% load. It helps to throw CPUBurn on top of it to obtain a
> sustained 100% load. I let it run for at least 4hrs, and usually 24hrs
when I am
> finalizing my CPU settings.

I'm running Win2K which does not support Hyperthreading, so all that's
wasted on me. Running CPU Burn on top of Prime95 seems a little severe. If
I can get a system to survive Prime95 overnight, I find that it can pretty
much take anything else I can throw at it.

Don't get me wrong. I'm very pleased with the performance I'm getting. My
3DMark3 scores are pushing 6000 and I haven't even thought about
overclocking the video card (yet). Games are what I want this box to be
best at, and so far I haven't seen anything else that comes close (at least
for the price and running cool, quiet and stable)

"Mark Bell" <> wrote in message
news:...
> "Frank Weston" <> spewed forth in such a glorious
tone
> the following gruel:
>
> >
> >"Mark Bell" <> wrote in message
> >news:.. .
> >> One problem is you started with a 2.8ghz versus a 2.4ghz chip. The
> >Northwood
> >> core is hitting the end of its cycle from Intel @ 3.2ghz. Without a
choice
> >chip
> >> and serious cooling, hitting 3600mhz+ is not going to happen that
often.
> >
> >I'm not so much bumping the limit of the chip as I am the FSB limit of
the
> >motherboard. I don't really need or want to run at 3.6 Ghz, but I am
> >curious as to why I can't get an FSB of greater than 242 when I read of
> >people attaining 275+ with relative ease. Could it be that the reviewers
> >are unrealistically hypeing this board?
>
> Because as the FSB increases, so does your core CPU speed. I am sure the
board
> can hit an FSB of 275+, but only with a CPU that can obtain that speed.
> Core_CPU_Speed = FSB x multiplier, so you are stuck, potentially, at your
> chips maximum overclocking potential. You are asking your 2.8ghz chip to
do
> 3850mhz. Using a chip with a lower multiplier would afford you a greater
chance
> of hitting an FSB>=275mhz. I assume you are monitoring your temps
vigorously.
>
> What I am basically saying is it is most likely your CPU, not your
motherboard
> that is limiting your FSB.
>
> >> You want to hit 275mhz FSB on a 2.8ghz (14 x 200mhz) = 3850mhz. Good
luck
> >on
> >> that one without a choice chip and serious cooling.
> >>
> >> A 2.4ghz chip (12x200), would hit 3300mhz @ 275mhz FSB. That is a much
> >more
> >> reasonable goal to reach.
> >>
> >
> >Again, I don't need to run at 275 FSB. I'm just curious as to why I
can't.
> >I'm reading reviews where people are topping 300 FSB with 3.0 Ghz chips.
>
> I never said you, "need," to run at 275mhz FSB. I said you want to.
>
> As for 3ghz chips, I find that hard to believe, unless they are running
serious
> cooling (phase, water, etc...) and a choice chip. A 3.0ghz chip running at
> 300mhz FSB = 4.5ghz. That's the extreme exception, not the norm. I'm not
saying
> it is impossible; it is just improbable.
>
> >> Your memory is another issue. You need to gauge memory bandwidth versus
> >core CPU
> >> frequency to obtain optimal performance (which is the goal). While your
> >chip
> >> may have topped out @ FSB242mhz, you at least are able to run your
memory
> >at
> >> full bandwidth (synchronously).
> >
> >No. As I stated, at 242 FSB I'm running memory at 193.6 which is a 4:5
> >memory/FSB ratio.
>
> You are running PC3700 memory which is rated at 233mhz FSB. It won't do
1:1
> even with a mild OC to 242mhz? Give it a little more juice if needed. I
can't
> see good OCZ not hitting an extra 9mhz with mildly relaxed timings. 4:5
timings
> with your FSB @ 242mhz running FSB 233mhz @ 193.6mhz is a waste of good
memory.
>
> OR....
>
> As I said, scale back to 233mhz FSB then and run it at 1:1 and see what
your
> benchmarks produce. The objective is optimal performance, not the highest
> core clock speed (which does not always corrolate to optimal performance).
>
> >That's a plus. I would scale back the memory to
> >> spec (233mhz = DDR466mhz = PC3700) since it is only a 9mhz (126mhz
gain)
> >> increase and run it with the better timings and see which one overall
> >provides
> >> better performance and efficiency.
> >
> >Yes, when it comes time to settle in on a configuration, I'll probably do
> >just that. But, now, I'm pushing to find the limits of the system
> >everywhere, and the FSB question is the big one.
>
> Well, not all CPUs will hit those extreme FSBs. Just as some get a choice
> chip, others get one that will not overclock that much when tested. It
> happens. The Northwood core is hitting its EOL. 3.2-3.4ghz is all Intel
> expected out of it, so you can't be surprised when some chips (even D1
> steppings) do not hit 3500mhz+. 3388mhz is respectable.
>
> Make sure to run Prime95 AND CPUBurn (burnmmx) at the same time to fully
> load the chip. Prime95, on it's own, will not give the new multithreaded
> CPUs a full 100% load. It helps to throw CPUBurn on top of it to obtain a
> sustained 100% load. I let it run for at least 4hrs, and usually 24hrs
when I am
> finalizing my CPU settings.
>
> >
> >BTW, you might think about changing your lead in line below. Some people
> >might not find it amusing and there's no use insulting those who might
> >seek/offer advice from/to you.
>
> Well, then they need a better sense of humor or the need to not take life
so
> seriously. >;-P I've been using that for, oh, 9yrs now.
>
> Mark
>
>
> P4-3.06ghz @ 3335mhz (FSB=144, Stock 1.55v 44c Max Load)
> ATI Radeon 9700 Pro @ 343.5/343.5
> SONY GDM-F520 21" Trinitron Monitor
> Gigabyte 8PE667 i845PE Motherboard
> Corsair PC3500 (DDR433) @ DDR375 2-2-2-6-1T
> Maxtor Ultra/133 8mb cache Gold Series 120GB HD
> Creative SB Audigy Platinum
> 16771 in 3DMark 2001SE 330
> 5715 in 3DMark 2003


Mark, I give you an A+ for patience

> check this: http://www.hexus.net/review.php?review=556&page=10


"A semi-unlocked Pentium 4 helped matters here"

"Frank Weston" <> spewed forth in such a glorious tone
the following gruel:

>
>"Mark Bell" <> wrote
>
>> Because as the FSB increases, so does your core CPU speed. I am sure the
>board
>> can hit an FSB of 275+, but only with a CPU that can obtain that speed.
>> Core_CPU_Speed = FSB x multiplier, so you are stuck, potentially, at your
>> chips maximum overclocking potential. You are asking your 2.8ghz chip to
>do
>> 3850mhz. Using a chip with a lower multiplier would afford you a greater
>chance
>> of hitting an FSB>=275mhz. I assume you are monitoring your temps
>vigorously.
>>
>> What I am basically saying is it is most likely your CPU, not your
>motherboard
>> that is limiting your FSB.
>>
>
>check this: http://www.hexus.net/review.php?review=556&page=10
>"it's an FSB monster. 285FSB came and went with utter stability; a testament
>to Intel's impressive engineering prowess. Fast and relatively inexpensive,
>just how we like it. Recommended."
>
>These guys were testing with a 3.0 FSB 800 processor.

They also had an engineering sample which is not multiplier locked. As a
result, they can take that 15x200mhz = 3ghz processor and run it at
10x300mhz = 3ghz and push up the FSB. All non-engineering samples (aka, what
we, the consumer, receive), are multiplier locked, which means the 15 is
constant, but the FSB is variable. The side effect is as the FSB rises, so does
the core CPU speed.

>
>Or, how about this:
>http://www.lostcircuits.com/motherbo...p4p800/9.shtml
>"Despite the limited CPU voltage range, we had no problems running the
>P4P800 overclocked to 3.6 or 3.7 GHz processor speed. In terms of overall
>board stability, the P4P800 was rock solid up to 300 MHz"

If you do the math, "rock solid up to 300mhz" would mean that processor was
running at 4.5ghz if it were a standard, consumer level 3ghz processor with a
FIXED multiplier of 15. Obviously they say up to 3.6ghz or 3.7ghz, so we can
surmise that they, too, have in their possession engineering samples which
allows manipulation of the multiplier.

>These guys were also running a 3.0 FSB 800 processor. I don't put too much
>stock in reviews, but these guys are either totally full of crap or I've got
>something holding my system back in the FSB department.

Well, what you have is a misinterpretation of the facts. It happens to all of
us.

>> As for 3ghz chips, I find that hard to believe, unless they are running
>serious
>> cooling (phase, water, etc...) and a choice chip. A 3.0ghz chip running at
>> 300mhz FSB = 4.5ghz. That's the extreme exception, not the norm. I'm not
>saying
>> it is impossible; it is just improbable.
>
>Although I'm not shooting for 4.5 Ghz, heat seems to be no problem. At
>3.4Ghz, with Prime95 running under most of my applications, my CPU temp is
>44 C with a motherboard temp of 35 C. That's cooler than idle on any AMD
>setup I've ever run. At idle, I'm running 27C. BTW I'm running air cooling
>with a Zalman 7000, and the sound is undetectable.

Then you can rule out temps, but that is but one factor in determining your OC
potential. You also have Vcore (which you already tested @ max), and simply put,
the limit of your chip. It appears you have hit the limit of your chip. Even
with good temps and a max'd Vcore within desired tolerances, your chip has
simply hit the proverbial wall.

>> As I said, scale back to 233mhz FSB then and run it at 1:1 and see what
>your
>> benchmarks produce. The objective is optimal performance, not the highest
>> core clock speed (which does not always corrolate to optimal performance).
>
>I thought this was a good idea, so I tried it. Sandra benchmarks reflected
>a slower CPU, and memory bandwidth didn't increase by any significant
>amount. Worse yet, I got tearing and artifacts when running 3DMark3 and in
>some of my games. When I run the memory at the equivalent of maxed PC3200,
>these problems go away. I'm back to 242 FSB and a 4:5 divider. Maybe I
>should have just bought cheaper memory.

If at all possible, I would explore returning the memory as defective if it will
not run at its rated speed without problems. PC3700 should run at PC3700 specs
(DDR466), not PC3200 specs (DDR400).

>> Make sure to run Prime95 AND CPUBurn (burnmmx) at the same time to fully
>> load the chip. Prime95, on it's own, will not give the new multithreaded
>> CPUs a full 100% load. It helps to throw CPUBurn on top of it to obtain a
>> sustained 100% load. I let it run for at least 4hrs, and usually 24hrs
>when I am
>> finalizing my CPU settings.
>
>I'm running Win2K which does not support Hyperthreading, so all that's
>wasted on me. Running CPU Burn on top of Prime95 seems a little severe. If
>I can get a system to survive Prime95 overnight, I find that it can pretty
>much take anything else I can throw at it.

Well, the idea is absolute stability. If you present your CPU with a sustained
100% load, and it passes, then you know it will not fail in any given situation
in terms of CPU stability and functionality. Sometimes overkill *is* good.

>Don't get me wrong. I'm very pleased with the performance I'm getting. My
>3DMark3 scores are pushing 6000 and I haven't even thought about
>overclocking the video card (yet). Games are what I want this box to be
>best at, and so far I haven't seen anything else that comes close (at least
>for the price and running cool, quiet and stable)

Agreed.

Mark

P4-3.06ghz @ 3335mhz (FSB=144, Stock 1.55v 44c Max Load)
ATI Radeon 9700 Pro @ 343.5/343.5
SONY GDM-F520 21" Trinitron Monitor
Gigabyte 8PE667 i845PE Motherboard
Corsair PC3500 (DDR433) @ DDR375 2-2-2-6-1T
Maxtor Ultra/133 8mb cache Gold Series 120GB HD
Creative SB Audigy Platinum
16771 in 3DMark 2001SE 330
5715 in 3DMark 2003

"Mark Bell" <> wrote

> They also had an engineering sample which is not multiplier locked. As a
> result, they can take that 15x200mhz = 3ghz processor and run it at
> 10x300mhz = 3ghz and push up the FSB. All non-engineering samples (aka,
what
> we, the consumer, receive), are multiplier locked, which means the 15 is
> constant, but the FSB is variable. The side effect is as the FSB rises, so
does
> the core CPU speed.

Although one review mentioned a "semi-unlocked" processor, one did not, and
there are other reviews that use stock processors that I didn't quote.
Enlighten me. On an engineering sample is the multiplier set mechanically?
If not, the unlocked processor point is irrelevant, because the reviews
were for stock ASUS P4P motherboards and that BIOS has no capability to
adjust the multiplier.

> If you do the math, "rock solid up to 300mhz" would mean that processor
was
> running at 4.5ghz if it were a standard, consumer level 3ghz processor
with a
> FIXED multiplier of 15. Obviously they say up to 3.6ghz or 3.7ghz, so we
can
> surmise that they, too, have in their possession engineering samples which
> allows manipulation of the multiplier.

No question about the math, I did it a long time ago. I agree that the
numbers just don't add up. On the other hand they do claim 3.6 to 3.7 Ghz
which is still more than I'm getting and would correlate to an FSB of over
260 for my processor. Why am I stuck at 242? That's the question I'm still
picking at. I don't imagine my system will be stable or cool at 3.7, but
I'd like to find out if I can get a little closer to that speed.

> If at all possible, I would explore returning the memory as defective if
it will
> not run at its rated speed without problems. PC3700 should run at PC3700
specs
> (DDR466), not PC3200 specs (DDR400).

I've been wondering about this. At PC3700 specs and tight timing, the
memory tests fine, the system doesn't hang, Prime95 runs well. But, I get
display problems (tearing, streaking, etc.) in the CPU test portion of
3DMk3, and in some games. Are these memory problems or could they be a
function of the relationship between video card and memory speed, or could
they even be a function of the monitor I'm using? I need to do more testing
and research before I can blame the memory. In any case, it's from Newegg,
and I'm not worried about a return if that becomes necessary.

> Well, the idea is absolute stability. If you present your CPU with a
sustained
> 100% load, and it passes, then you know it will not fail in any given
situation
> in terms of CPU stability and functionality. Sometimes overkill *is* good.
>

Here's where our philosophy differs. Nothing is absolute. To guarantee
absolute stability one would have to test at 100% capacity for an infinite
duration. That wouldn't leave much time to use the computer for anything
else. I prefer to test to a level that guarantees reasonable stability and
leaves me a little time for work and games.

I think it's more a combination of board and ram than anything.
Clearly it's evident that just by different bios revision, that the
board hardware and bios ver are limiting factors. And as far as ram,
the whole deal with i875P chipsets not working with many brands (GeIL
and Corsair) Ultra/Low Latency DDR modules is self explanitory. I
chalk it up to mostly luck. You can repeat results by buying the same
hardware to some extent over and over once you find something that you
know works, but sometimes you can buy all the same stuff and get very
different (good or bad) results.

I ended up with some plain jane Samsung CL2.5 DDR400 memory modules.
The 1GB GeIL Dual DDR400 ULD kit I bought wouldn't even post. Hell,
it wouldn't even turn on. But they work good enough to run 3.45G on
my 3.00 perfectly stable (230FSB) with vcore 1.7, ddr voltage 2.65,
agp voltage 1.75. I can run max memory timings, and all settings on
the fastest on an MSI 875P Neo board with bios v1.6.

But the machine runs the same when using the SPD setting. Besides,
irregardless when you set to SPD doesn't it read the rom off the DDR
module and pull it's timings from there anyway? So if it's faster
timed memory, SPD should see and set accordingly? At least that's
what Sandra reports in mainboard information under memory modules.

Anyway I just thought I'd interject ;-)

Good luck on going faster,
Kevin "I NEED WATER COOLING" Keenan

P4 3.00GHz at 3.45GHz (15 x 230 FSB = 3452MHz) vcore 1.7V - Stock
Intel Cooler
All In Wonder 9700 Pro @ 355/355
Viewsonic 21" P225f+ Monitor
MSI 875P Neo FIS2R
1G Samsung 2x512MB DDR400 Memory CL 2,2,2,5,8
WD 120GB 8MB Buffer HDD
2 x 200GB 8MB Buffer HDD in RAID-0 on Promise Fasttrak TX2000 IDE Raid
Adapter
Antec SmartBlue350 Power Supply

3DMark2001SE 18353
3Dmark03 6011
Q3 Demo001 404
Sandra2003Pro CPU 10600
Santra2003Pro Mem 5299

CPU is 28C at idle, and 55C at 100% load (encoding DiVX, benching, and
D3D gaming