Underclocking P4 3.0E Prescott

[EdwardATeller]

I dropped this P4 3.0E chip:

http://www.newegg.com/Product/Product.aspx?Item=N82E16819116027

into my socket 478 Asus P4R800-VM motherboard:

http://dlsvr03.asus.com/pub/ASUS/mb/sock478/p4r800-vm/e1662_p4r800-vm.pdf

As advertised, the chip runs hot. I have this Zalman fan/heat sink on
it:

http://www.newegg.com/Product/Product.aspx?Item=N82E16835118114

The fan runs at full speed all the time. The chip is about 51C at
idle. I used to have a Celeron 2.4 Ghz in there, and the fan would
idle at a much slower, and quieter, speed. I have the BIOS set to
drop the fan speed when possible, but clearly it thinks 51C is too
warm, so it runs full speed.

If I underclock the CPU, this might make it run cooler and quieter.
The computer is a DVR, and if I can play HD video with a slightly
slower CPU, that's all I need it to do.

Never played with this stuff before, but I'd be curious what people
here have to say. Thanks.

 

[philo]

I dropped this P4 3.0E chip:

http://www.newegg.com/Product/Product.aspx?Item=N82E16819116027

into my socket 478 Asus P4R800-VM motherboard:

http://dlsvr03.asus.com/pub/ASUS/mb/sock478/p4r800-vm/e1662_p4r800-vm.pdf

As advertised, the chip runs hot. I have this Zalman fan/heat sink on
it:

http://www.newegg.com/Product/Product.aspx?Item=N82E16835118114

The fan runs at full speed all the time. The chip is about 51C at
idle. I used to have a Celeron 2.4 Ghz in there, and the fan would
idle at a much slower, and quieter, speed. I have the BIOS set to
drop the fan speed when possible, but clearly it thinks 51C is too
warm, so it runs full speed.

If I underclock the CPU, this might make it run cooler and quieter.
The computer is a DVR, and if I can play HD video with a slightly
slower CPU, that's all I need it to do.

Never played with this stuff before, but I'd be curious what people
here have to say. Thanks.

Sure, try under clocking it.

You would not want it to get as hot as one of your H. bombs

 

[John McGaw]

I dropped this P4 3.0E chip:

http://www.newegg.com/Product/Product.aspx?Item=N82E16819116027

into my socket 478 Asus P4R800-VM motherboard:

http://dlsvr03.asus.com/pub/ASUS/mb/sock478/p4r800-vm/e1662_p4r800-vm.pdf

As advertised, the chip runs hot. I have this Zalman fan/heat sink on
it:

http://www.newegg.com/Product/Product.aspx?Item=N82E16835118114

The fan runs at full speed all the time. The chip is about 51C at
idle. I used to have a Celeron 2.4 Ghz in there, and the fan would
idle at a much slower, and quieter, speed. I have the BIOS set to
drop the fan speed when possible, but clearly it thinks 51C is too
warm, so it runs full speed.

If I underclock the CPU, this might make it run cooler and quieter.
The computer is a DVR, and if I can play HD video with a slightly
slower CPU, that's all I need it to do.

Never played with this stuff before, but I'd be curious what people
here have to say. Thanks.

I'd be more curious as to why your CPU is running so hot at idle with
that heatsink and fan combination on it. I have the same CPU in my
Shuttle compact system (SB65G2) with the stock "ICE" cooler, not known
for its great cooling capacity, and I can run my 80mm fan at the slowest
speed and the CPU runs at around 42-45C at idle. It is not silent but it
makes no more noise than the Galaxy 8800GT video card which I have in
the new HTPC system I built.

Are you sure that you installed the cooler correctly? Peeled off any
coverings on the bottom if required? Applied the proper amount of
heatsink compound if required? Got the plate locked down properly? Is
the CPU fan truly compatible with the motherboard's control system? Is
it plugged into the correct header? Sorry if all of these seem obvious
but sometimes things get overlooked...

But, that said, you should be able to underclock the CPU by some amount
although I don't know if you will be able to reduce the voltage which
would also reduce the power dissipation even more.

 

[Paul]

I dropped this P4 3.0E chip:

http://www.newegg.com/Product/Product.aspx?Item=N82E16819116027

into my socket 478 Asus P4R800-VM motherboard:

http://dlsvr03.asus.com/pub/ASUS/mb/sock478/p4r800-vm/e1662_p4r800-vm.pdf

As advertised, the chip runs hot. I have this Zalman fan/heat sink on
it:

http://www.newegg.com/Product/Product.aspx?Item=N82E16835118114

The fan runs at full speed all the time. The chip is about 51C at
idle. I used to have a Celeron 2.4 Ghz in there, and the fan would
idle at a much slower, and quieter, speed. I have the BIOS set to
drop the fan speed when possible, but clearly it thinks 51C is too
warm, so it runs full speed.

If I underclock the CPU, this might make it run cooler and quieter.
The computer is a DVR, and if I can play HD video with a slightly
slower CPU, that's all I need it to do.

Never played with this stuff before, but I'd be curious what people
here have to say. Thanks.

On a Prescott, at least some of the power dissipation is leakage
current. Back when CMOS had really large feature size (3 micron),
power would be close to perfectly proportional to FCV**2. With
the Prescott, I might not have the right figure, but I thought
leakage was somewhere around 25%. When you underclock,
don't expect a linear improvement from F alone, as the
leakage is a constant factor added to the FCV**2. So
even at DC (clock stopped - if that was possible), the
thing would still consume 25% of TDP.

I'd run it full speed, and throw a better cooler at it.
The Tuniq Tower would be an example, if it'll fit in
the computer case . This is a BIG cooler.

http://www.newegg.com/Product/ProductReview.aspx?Item=N82E16835154001

Also, make sure your case ventilation is adequate. I have a
CNPS7000 on my current computer, and noticed a "warm cloud"
all around the CPU area. I increased the capacity of my
rear case fan, to improve the movement of air past the
CPU area. That might convince your CPU fan speed to drop a
bit.

The rear fan on a computer case, cannot work unless there is
sufficient intake vent area on the front of the case. By removing
the lower plastic bezel on my Antec Sonata, I achieved that objective,
without having to cut more holes in it.

Paul

 

[kony]

I dropped this P4 3.0E chip:

http://www.newegg.com/Product/Product.aspx?Item=N82E16819116027

into my socket 478 Asus P4R800-VM motherboard:

http://dlsvr03.asus.com/pub/ASUS/mb/sock478/p4r800-vm/e1662_p4r800-vm.pdf

As advertised, the chip runs hot. I have this Zalman fan/heat sink on
it:

http://www.newegg.com/Product/Product.aspx?Item=N82E16835118114

The fan runs at full speed all the time. The chip is about 51C at
idle. I used to have a Celeron 2.4 Ghz in there, and the fan would
idle at a much slower, and quieter, speed. I have the BIOS set to
drop the fan speed when possible, but clearly it thinks 51C is too
warm, so it runs full speed.

If I underclock the CPU, this might make it run cooler and quieter.
The computer is a DVR, and if I can play HD video with a slightly
slower CPU, that's all I need it to do.

Never played with this stuff before, but I'd be curious what people
here have to say. Thanks.

The most effective way to reduce CPU heat is to reduce CPU
voltage, plus underclocking it enough to retain stability at
the lower voltage (runs Prime95 stabily for at least an hour
with no errors). Whether your board has suitable features I
don't know but it probably does being an Asus. That will
definitely reduce the temps, though I can't say how far you
can underclock that chip before it struggles with HD video.

Today modern video card GPUs have hardware HD decoding
support so one way to offload some of the processing is use
of such a video card. Personally I would have bought a new
video card towards this end before buying a new processor,
but you may need both since DVR entails recording as well as
playback. However it seems your board has AGP slot not PCIe
and I'm not familiar with what the best option is on a
current-gen AGP card for your purpose, nor if you were
already using a separate video card instead of the
integrated video.

 

[Ned B]

I dropped this P4 3.0E chip:

http://www.newegg.com/Product/Product.aspx?Item=N82E16819116027

into my socket 478 Asus P4R800-VM motherboard:

http://dlsvr03.asus.com/pub/ASUS/mb/sock478/p4r800-vm/e1662_p4r800-vm.pdf

As advertised, the chip runs hot. I have this Zalman fan/heat sink on
it:

http://www.newegg.com/Product/Product.aspx?Item=N82E16835118114

The fan runs at full speed all the time. The chip is about 51C at
idle. I used to have a Celeron 2.4 Ghz in there, and the fan would
idle at a much slower, and quieter, speed. I have the BIOS set to
drop the fan speed when possible, but clearly it thinks 51C is too
warm, so it runs full speed.

If I underclock the CPU, this might make it run cooler and quieter.
The computer is a DVR, and if I can play HD video with a slightly
slower CPU, that's all I need it to do.

Never played with this stuff before, but I'd be curious what people
here have to say. Thanks.

I would recommend to try lapping the bottom of the Zalman heat sink
plate by carefully rubbing it flat against 600 sandpaper masking taped
down to a flat table. Follow this with a sheet of 1000 sandpaper and it
may make a big improvement for you. If your case is well ventilated you
should be getting a temperature of 35-38C with your cooling fan.
You only need about a half a drop of silicone heat transfer grease on
the top of your CPU and the Zalman heat sink. I have a Abit 478 socket
P4 3.2 Pentium with 800FSB and my CPU at idle is 34C using a NinjaPlus
revision B heat sink. I am using a well ventilated Antec Sonata case
with extra air able to come into the front lower bezel by cutting away a
lot of the bottom of it and also both sides about 1/2 inch. My exhaust
fan is 120mm running at about 700rpm. My Radeon HIS X1950Pro video card
exhausts through the rear of the case. My power supply runs very low
fan speed, it is a PC Power & Cooling Silencer 750.

 

[EdwardATeller]

I'd be more curious as to why your CPU is running so hot at idle with
that heatsink and fan combination on it. I have the same CPU in my
Shuttle compact system (SB65G2) with the stock "ICE" cooler, not known
for its great cooling capacity, and I can run my 80mm fan at the slowest
speed and the CPU runs at around 42-45C at idle. It is not silent but it
makes no more noise than the Galaxy 8800GT video card which I have in
the new HTPC system I built.

Are you sure that you installed the cooler correctly? Peeled off any
coverings on the bottom if required? Applied the proper amount of
heatsink compound if required? Got the plate locked down properly? Is
the CPU fan truly compatible with the motherboard's control system? Is
it plugged into the correct header? Sorry if all of these seem obvious
but sometimes things get overlooked...

But, that said, you should be able to underclock the CPU by some amount
although I don't know if you will be able to reduce the voltage which
would also reduce the power dissipation even more.

Thanks for all the good replies. Turns out I can't change the
multiplier on my board. I guess that means I can't underclock, but
all the info here leads me to think that I must not have installed the
heat sink properly. Maybe it's not seated flat, so I'll take a look
at that first.

I tried buying a new graphics card first, but that didn't seem to do
the trick. I finally got my system to play HD with a new processor
and the new graphics card, but today the new card's fan started
buzzing, so I tried my old card, and guess what, it worked fine. That
means the 7600 GS is going back to Circuit City. I did change my
memory configuration today to have the DDR RAM in the proper slots
(1+3, instead of 1+2). Not sure if that made a difference, but it's
working fine now, just a little noisy.

Thanks again.

 

[kony]

Thanks for all the good replies. Turns out I can't change the
multiplier on my board. I guess that means I can't underclock, but
all the info here leads me to think that I must not have installed the
heat sink properly. Maybe it's not seated flat, so I'll take a look
at that first.

I tried buying a new graphics card first, but that didn't seem to do
the trick. I finally got my system to play HD with a new processor
and the new graphics card, but today the new card's fan started
buzzing, so I tried my old card, and guess what, it worked fine. That
means the 7600 GS is going back to Circuit City. I did change my
memory configuration today to have the DDR RAM in the proper slots
(1+3, instead of 1+2). Not sure if that made a difference, but it's
working fine now, just a little noisy.

Thanks again.

If memory serves me, 7nnn series only does about 30% HD
decoding while 8nnn series does a lot more. Here's an
interesting article, Google will find others.
http://arstechnica.com/news.ars/pos...u-usage-during-hd-playback-ati-vs-nvidia.html

 

[kony]

When you underclock,
don't expect a linear improvement from F alone, as the
leakage is a constant factor added to the FCV**2. So
even at DC (clock stopped - if that was possible), the
thing would still consume 25% of TDP.

.... but isn't that a great reason to underclock, in that it
allows decreasing V?

 

[Paul]

... but isn't that a great reason to underclock, in that it
allows decreasing V?

My point is, for the Prescott, that the improvement might not
be as good as you might expect. Yes, I agree, that dialing
down Vcore helps, as long as the thing is still stable. But
there will always be that leakage term in the power equation,
preventing the kind of improvements that the basic frequency
related term predicts. As far as I know, Prescott is the
worst for this kind of thing.

When F is zero here, the Prescott will still have the leakage power,
whatever the right number happens to be.

Power = FCV**2 + I_leakage*V

The leakage term on CMOS, used to be so small, that it was
used as a diagnostic. The IDDQ test used to be used at wafer
sort, to detect defective CMOS. Back in those days, even a
little leakage at DC, indicated a defective chip. But when
the chip burns 25% power at DC, such tests are no longer
meaningful.

http://en.wikipedia.org/wiki/Iddq_testing

Paul

 

[GT]

I dropped this P4 3.0E chip:

http://www.newegg.com/Product/Product.aspx?Item=N82E16819116027

into my socket 478 Asus P4R800-VM motherboard:

http://dlsvr03.asus.com/pub/ASUS/mb/sock478/p4r800-vm/e1662_p4r800-vm.pdf

As advertised, the chip runs hot. I have this Zalman fan/heat sink on
it:

http://www.newegg.com/Product/Product.aspx?Item=N82E16835118114

The fan runs at full speed all the time. The chip is about 51C at
idle. I used to have a Celeron 2.4 Ghz in there, and the fan would
idle at a much slower, and quieter, speed. I have the BIOS set to
drop the fan speed when possible, but clearly it thinks 51C is too
warm, so it runs full speed.

I'm running the same cooler on my Core 2 Duo. I'm not impressed with its
cooling, but it is quieter than the stock cooler! I can understand it
struggling with the P4.

Simple test - run the PC with the side off the case and what temp does the
CPU reach then and what speed does the CPU fan run at? If it is more than a
few degrees lower then your current temperatures, then just invest in more
case cooling. Failing that, then go with Kony's suggestion of reducing the
CPU voltage and speed. Adjust in the smallest steps possible.

 

[EdwardATeller]

I'm running the same cooler on my Core 2 Duo. I'm not impressed with its
cooling, but it is quieter than the stock cooler! I can understand it
struggling with the P4.

Simple test - run the PC with the side off the case and what temp does the
CPU reach then and what speed does the CPU fan run at? If it is more than a
few degrees lower then your current temperatures, then just invest in more
case cooling. Failing that, then go with Kony's suggestion of reducing the
CPU voltage and speed. Adjust in the smallest steps possible.

Thanks for the feedback on my cooler. Maybe that's the problem.

This morning, I added a little thermal paste, and the temperature came
down a lot. I added a little more, and it went up a bit. I scraped
some off, and it didn't go back down to the lowest, but still lower
that it was.

I am out of thermal paste right now, so I'll just wait until I have
something else to order from NewEgg, and I'll get a tube of Tuniq TX-2
Thermal Compound:

http://www.newegg.com/product/product.aspx?item=N82E16835154003

I think I have a good feel for the right amount of paste to use.
Watching HD video runs my CPU at about 54C. I can live with that for
now.

I also put my Celeron back on as a test, and the fan was just as loud,
so maybe I just mis-remembered how loud it used to run. The Celeron
idled around 34C, and the lowest I got my P4 to idle was 38C.

 

[EdwardATeller]

If memory serves me, 7nnn series only does about 30% HD
decoding while 8nnn series does a lot more. Here's an
interesting article, Google will find others.http://arstechnica.com/news.ars/post/20071230-measuring-cpu-usage-dur...

Right now, I'm getting smooth HD playback with an ATI Radeon 9200
128MB AGP card. This card has no fan and doesn't pull a lot of
power. That's why the 7600 GS is heading back to the store for an $80
refund.

I was able to upgrade to a HD-DVR for about $140, $80 for the P4 and
$60 for the Hauppauge HVR-1600. The large amount of time I put into
this project goes down as an educational expense. The most important
thing I learned, after getting the hardware right, is to use the
CyberLink codecs included in the latest version of PowerDVD.

Thanks for all the info and help.

 

[kony]

My point is, for the Prescott, that the improvement might not
be as good as you might expect. Yes, I agree, that dialing
down Vcore helps, as long as the thing is still stable. But
there will always be that leakage term in the power equation,
preventing the kind of improvements that the basic frequency
related term predicts. As far as I know, Prescott is the
worst for this kind of thing.

When F is zero here, the Prescott will still have the leakage power,
whatever the right number happens to be.

Power = FCV**2 + I_leakage*V

You have a point, and yet that right number depends on V as
well. I feel ultimately it makes a Prescott a poor choice
for low noise/energy-conservation at all, but now that the
chip is acquired in the real-world implementation whatever
that can be done to reduce temp (while keeping at least the
minimal level of performance needed) would come closer to
the goal.

The leakage term on CMOS, used to be so small, that it was
used as a diagnostic. The IDDQ test used to be used at wafer
sort, to detect defective CMOS. Back in those days, even a
little leakage at DC, indicated a defective chip. But when
the chip burns 25% power at DC, such tests are no longer
meaningful.

http://en.wikipedia.org/wiki/Iddq_testing

Perhaps, but 25% of 100W versus 25% at 65W is a difference.

 

[GT]

Thanks for the feedback on my cooler. Maybe that's the problem.

This morning, I added a little thermal paste, and the temperature came
down a lot. I added a little more, and it went up a bit. I scraped
some off, and it didn't go back down to the lowest, but still lower
that it was.

I am out of thermal paste right now, so I'll just wait until I have
something else to order from NewEgg, and I'll get a tube of Tuniq TX-2
Thermal Compound:

http://www.newegg.com/product/product.aspx?item=N82E16835154003

I think I have a good feel for the right amount of paste to use.
Watching HD video runs my CPU at about 54C. I can live with that for
now.

I also put my Celeron back on as a test, and the fan was just as loud,
so maybe I just mis-remembered how loud it used to run. The Celeron
idled around 34C, and the lowest I got my P4 to idle was 38C.

How long have you had the cooler for? Perhaps the fan is starting to fail -
they get noisy before popping off to the big computer case in the sky. Do
you have another fan (a desk fan perhaps) that you could point at the cooler
while you unplug the cooler fan. See if the temperature drops or rises when
you change the cooling fan. DON'T run it with no fan for more than a few
seconds, but there is nothing to say you have to use the fan that came with
the cooler - I use a larger fan, mounted on a bracket above my cooler. The
fan is larger, so runs slower and is silent.

 

[EdwardATeller]

How long have you had the cooler for? Perhaps the fan is starting to fail -
they get noisy before popping off to the big computer case in the sky. Do
you have another fan (a desk fan perhaps) that you could point at the cooler
while you unplug the cooler fan. See if the temperature drops or rises when
you change the cooling fan. DON'T run it with no fan for more than a few
seconds, but there is nothing to say you have to use the fan that came with
the cooler - I use a larger fan, mounted on a bracket above my cooler. The
fan is larger, so runs slower and is silent.

That is a great idea. The cooler is coming up on 4 years old. I have
a large case fan sitting around, and I can lay it across the top of
the fins.

One thing I forgot to mention is that there is some dust build up on
the heatsink blades. It's better than it was when I started this
project, but maybe that is an issue. I'll get some compressed air and
clean it up.

 

[EdwardATeller]

That is a great idea. The cooler is coming up on 4 years old. I have
a large case fan sitting around, and I can lay it across the top of
the fins.

One thing I forgot to mention is that there is some dust build up on
the heatsink blades. It's better than it was when I started this
project, but maybe that is an issue. I'll get some compressed air and
clean it up.

Tried the new fan, but it was WAY noiser, running at 3600 rpm instead
of 2400. It had a 4 wire connector, but was able to fit on the
motherboard's CPU fan connector. Blew the dust off with a straw, and
that made a lot of difference with the CPU temp. DOH! But the fan
still does not slow down when the processor is at 37C. Something must
have happened to the fan-slowing mechanism which is BIOS controlled.
Even the little manual control on the power cord doesn't do anything.

I'm going to look for a utility that will slow down the fan from the
OS (XP sp2). That might do the trick. Thanks again for all the help.

 

[John McGaw]

EdwardATeller wrote:
snip...

Tried the new fan, but it was WAY noiser, running at 3600 rpm instead
of 2400. It had a 4 wire connector, but was able to fit on the
motherboard's CPU fan connector. Blew the dust off with a straw, and
that made a lot of difference with the CPU temp. DOH! But the fan
still does not slow down when the processor is at 37C. Something must
have happened to the fan-slowing mechanism which is BIOS controlled.
Even the little manual control on the power cord doesn't do anything.

I'm going to look for a utility that will slow down the fan from the
OS (XP sp2). That might do the trick. Thanks again for all the help.

If the fan is four-pin then you have a PWM fan which is the way that
most modern MBs seem to be going. Honestly, I don't know if such a fan
is controllable by a three-pin header on an older MB. The PWM has its
speed controlled by a string of pulses put out by the MB while the older
sort have their speed varied with a variation in voltage.

 

[EdwardATeller]

EdwardATeller wrote:

snip...





If the fan is four-pin then you have a PWM fan which is the way that
most modern MBs seem to be going. Honestly, I don't know if such a fan
is controllable by a three-pin header on an older MB. The PWM has its
speed controlled by a string of pulses put out by the MB while the older
sort have their speed varied with a variation in voltage.

That is an old fan I had lying around. Might have pulled it out of a
junked server I found in a dumpster back in New York.

Really seems like something has disabled my Q-Fan control that is set
in the BIOS. Any idea what kind of trigger temps are used to set the
fan speeds? If 37C is considered hot, then my chip starts off hot,
and that would explain the constant high fan speed.

 

[GT]

Tried the new fan, but it was WAY noiser, running at 3600 rpm instead
of 2400. It had a 4 wire connector, but was able to fit on the
motherboard's CPU fan connector. Blew the dust off with a straw, and
that made a lot of difference with the CPU temp. DOH! But the fan
still does not slow down when the processor is at 37C. Something must
have happened to the fan-slowing mechanism which is BIOS controlled.
Even the little manual control on the power cord doesn't do anything.

I'm going to look for a utility that will slow down the fan from the
OS (XP sp2). That might do the trick. Thanks again for all the help.

You can buy fan controllers - have a google and see what you think.