P4 3Ghz. Xp120 - P4C800E-Dx. Nighthmare! Help please.

Cooling Processors quietly

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Thorz
Posts: 69
Joined: Sun Feb 13, 2005 8:01 pm

P4 3Ghz. Xp120 - P4C800E-Dx. Nighthmare! Help please.

Post by Thorz » Tue Mar 01, 2005 1:09 pm

Hi m8s.

I was really exited after finishing changing my 2 biggest sources of noice, the Intel stock fan and the ATI 9800pro stock fan on my Hercules card.
I have paired the XP120 with a 120 nexus modded fan as recommended several times in the forums. I had to bend 2 capacitors beside the CPU socket for that the XP120 could fit, it was stated at Thermanright website that this had to be done with my board.
To put this cooler was a pain, I have dismounted the motherboar and even with it put it was very difficult. Something really strange was that the 4rth. hook (last one) was very difficult to attach because the metal hook of the XP120 was quite far from the plastick hole of the retainer. I could place ot there at the end but it didn't feeled so safe locked. I tested several times to be sure that it was not going to fall off. The Nexus fan is blowing air down.
Now that the system is ready I have tried to run Prime95 stress test for some hours to see the temps and it doesn't complete the test! It returns an error after letting it run some minutes! This has never happened to me before. I could run the stress test + 3dMark2005 togheter all the night without a single problem before.
My temps are (from Asus PC Probe):
Idle:
CPU: 38C
MB: 36C

On load:
CPU: 53C
MB: 41C

Temperatures are agree with some thermal probes that I have in the case.
ATI 9800Pro is been cooled by the Zalman VF700-Cu that is been controlled with a Vantec Nexus Front Pannel..

What can be the problem here? This themperatures doesn't impress me so much compared with what I was getting woth the Intel stock heatsink, but my load temp is not that high for making Prime95 fail, or is it?

Please HELP ME people :( I am very dissapointed that my system is not working as espected.

Thank you very much.

The Instigator
Posts: 69
Joined: Mon Nov 22, 2004 7:43 pm
Location: Southern California

Post by The Instigator » Tue Mar 01, 2005 3:16 pm

Your temps really arent that out of line and are very safe. Some questions that need to be addressed are what kind of thermal compound and how much was used. What is your ambient air temp? What kind of numbers were you getting with the stock HSF? Is it a prescott or northwood CPU?

I personally have a 2.4C at 3.0GHZ which stays around 32C idle and about 45 full load with an xp-120 and 5V nexus. If you are using a prescott chip, that you are actually doing pretty for temps.

You might want to think about lapping the base to provide better cpu contact as the xp series of heatsinks seem to have an issue of not being vey flat and maybe getting some arctic silver 5.

Remember that the idea behind your heatsink is to allow for good cooling at low noise. If you want to bring your temps way down, you can put an 80cfm screamer fan and shave 5-10 more degrees, but thats not the point. As your system is now, you have safe temps and low noise which is a great place to be at.

Thorz
Posts: 69
Joined: Sun Feb 13, 2005 8:01 pm

Post by Thorz » Tue Mar 01, 2005 10:32 pm

Temps with the Intel stock fan were almost the same as these ones.
What worries me is that I am running the Neexus at 12V (directly connected to the MB because the Vantec Nexus Panel didn't like the low RPM of the Nexus fan and the alarm was on all the time.

This is a 3Ghz CPU northwood WITHOUT overclocking.
I have the PSU fan and an exhaust fan directly over the CPU fan.
I used Artic Silver 2 for contact with the method that I ALWAYS use: Use some at the heatsink base and rubber it with a plastic bag for covering any micro-hole in the heatsink surface, then use some more thin layer on the CPU core using a credit card for spreading it. Have not have any trouble with this method in the last 5 years.

This heatsink should produce lower temps whit this fan at 12V if I have read well the posts here. I like to test this setup ol at is max cooling capabilities and then put down the Nexus fan to 7V or so, but if it cannot run Prime95 with a 12V fan I am pretty doomed.

I am ruinning the tests again and Asus Probe shows CPU 52C and 32C for the MB temp,
My 3 real temp probes on the Vantec Nexus Panel show:
CPU: 50.5C
Chipset HS: 49.5C
Case: 41C

Room temp must be between 20 and 25C, I cannot find that easily, I don't have a thermometer here.

Can it be that the XP120 is wrong installed? It is IMPOSSIBLE to see if there is space between it and the CPU. I was ackward to install and the 4rth (last) hook looked very bad IMO. Can it be that the capacitors that I slightly beded have damaged the board?

Thanks.

EDIT: Prime95 fails even with the case open! I am pretty sure that temps are not going over 56C. I think something has got very screwed here :(

frankgehry
Posts: 1424
Joined: Sun Dec 12, 2004 5:00 am
Location: New York, NY

?

Post by frankgehry » Tue Mar 01, 2005 11:29 pm

T,

So the temps are ok, but the error from prime95 is what worries you? Is this correct. Maybe since you have the fan tied into the motherboard it is not keeping up and the mb is production an error. By the way. I think connecting the fan to the mb is the best way, because the fan is checked by the mb and not the separate fan controller. I may have misunderstood you question, but one thing that caught my attention was the way you described attaching your heatsink. Ususally you hook the two clips at one end and then push down on the clips at the other end and they lock at the same time. You shouldn't do one corner at a time. It is kind of tricky though, and maybe it doesn't really matter very much. The base of my xp-120 was square as determined with a drafting triangle so I didn't think I could improve the flatness by lapping. - FG

Thorz
Posts: 69
Joined: Sun Feb 13, 2005 8:01 pm

Re: ?

Post by Thorz » Wed Mar 02, 2005 1:04 am

frankgehry wrote:T,

So the temps are ok, but the error from prime95 is what worries you? Is this correct. Maybe since you have the fan tied into the motherboard it is not keeping up and the mb is production an error.
Temps look fine, but if Prime 95 is having an error it means that something is wrong. It was not giving any error with the Intel fan. That is weird. I don't understand your explanation about the error, what do you mean with "it is not keeping up"? What this has to do with Prime95 error? Errors from Prime are due to miscalculation from the CPU (normally heat related and power related). I discard my other componnents because they were working fine before the XP-120.
frankgehry wrote: Ususally you hook the two clips at one end and then push down on the clips at the other end and they lock at the same time. You shouldn't do one corner at a time. It is kind of tricky though, and maybe it doesn't really matter very much.
Hmmm... I did one clip at a time. I didn't try to hook the last 2 corners at the same time... it was very difficult to install anyway. Maybe I will pull this out again and try to attach the heatsink again? Has to be in the afternoon then.

IMOM
Posts: 77
Joined: Sat Mar 27, 2004 10:55 pm
Location: CA, USA

Post by IMOM » Wed Mar 02, 2005 4:06 am

Just a guess, but remember you mentioned about bent capacitors...how far did you bend the capacitors...enough to break them? I was nervous about the same thing...but my system have been stable and I am running my globes @5V with any problems...though granted I haven't recently ran any CPU benchmarking.

frankgehry
Posts: 1424
Joined: Sun Dec 12, 2004 5:00 am
Location: New York, NY

Post by frankgehry » Wed Mar 02, 2005 5:49 am

T,

If you install your xp-120 again, after you hook the clips on the first side (easy part) don't press down on the fins to hook the second side, but press down on the clips - both at the same time. This will make it a lot easier. - FG

Thorz
Posts: 69
Joined: Sun Feb 13, 2005 8:01 pm

Post by Thorz » Wed Mar 02, 2005 6:45 am

frankgehry wrote:T,

If you install your xp-120 again, after you hook the clips on the first side (easy part) don't press down on the fins to hook the second side, but press down on the clips - both at the same time. This will make it a lot easier. - FG
Thank you m8, I will try your advice. It was very difficult to press on the clips. I have to see if a screwdriver helps.

The Instigator
Posts: 69
Joined: Mon Nov 22, 2004 7:43 pm
Location: Southern California

Post by The Instigator » Wed Mar 02, 2005 10:11 am

If you are going to remount the HS, I would reccomend applying the thermal paste the way arctic silver reccommends for processors with heat spreaders: Put a small glob in the center of the processor and mount the HS. Pressure and heat will spread out the compund and give the best results. Using the credit card to apply an even coat is only reccomended for chips with an exposed core...P3, Athlon XP, etc. This will also help to give more contact in the center of the HS/CPU where the chip is the hottest in case the HS base is in fact a little warped.

Link here: http://www.arcticsilver.com/arctic_silv ... s_big2.htm

Thorz
Posts: 69
Joined: Sun Feb 13, 2005 8:01 pm

Post by Thorz » Thu Mar 03, 2005 1:53 am

The Instigator wrote:If you are going to remount the HS, I would reccomend applying the thermal paste the way arctic silver reccommends for processors with heat spreaders: Put a small glob in the center of the processor and mount the HS. Pressure and heat will spread out the compund and give the best results. Using the credit card to apply an even coat is only reccomended for chips with an exposed core...P3, Athlon XP, etc. This will also help to give more contact in the center of the HS/CPU where the chip is the hottest in case the HS base is in fact a little warped.

Link here: http://www.arcticsilver.com/arctic_silv ... s_big2.htm
That was a nice advice. Thanks.
I have leave the machine as it is. It appears that the only program that fails is Prime (knock on wood). It is temperature related because at the end I could run it the hole night woth the case open without any fail. I guess it is a combination of factors but mostly it is my case that has bad airflow. I will change it very soon for a new Lian Li case.

Thanks again.

JanW
Posts: 296
Joined: Fri Dec 03, 2004 12:38 pm
Location: France, Europe Folding for SPCR

Post by JanW » Thu Mar 03, 2005 4:49 am

On my silent rig, I have two fans affecting temperatures of the CPU and surrounding components: The HSF and an exhaust fan the HS is ducted to. Normally the system is 24h Prime stable. The voltage I'm running the exhaust fan at is below its starting voltage, though. Once I forgot to start the fan up manually after a reboot (usually the machine is running 24/7, folding). CPU temps didn't change at all (the HSF was working), but Prime didn't make it more than 10 or 20 minutes. I attributed that to increased temperature of the Northbridge or the voltage regulators, which are cooled by the exhaust (less by the HSF in my setup). Now I wouldn't have thought that would be a problem with the XP-120 (assuming its blowing down), but don't discard the possibility of other components overheating right away.

Marcus
Posts: 8
Joined: Sun Mar 20, 2005 1:44 pm

Post by Marcus » Tue Mar 22, 2005 1:12 pm

I just received my XP-120 and I can tell with a simple visual test that the base is indeed not flat. All I did was held it between my face and a light source so that the light was reflected in the base. Then I tilted it towards and away from me. The XP-120 looked pretty good side-to-side, but along the longitudinal axis (heat pipes facing me), you can see the light reflection shifting like waves in a pond. So I'll definitely be lapping mine before I install it. I'll lose the nickel plating, but at least I'll be getting full contact.

Actually, now that I think about it, it might be interesting to try a before-and-after test to see just how much of a difference it makes. I'll post the results if I decide to try it. Waiting on a Nexus fan for it right now, so it may not be until next week.

JanW
Posts: 296
Joined: Fri Dec 03, 2004 12:38 pm
Location: France, Europe Folding for SPCR

Post by JanW » Tue Mar 22, 2005 1:43 pm

Welcome to SPCR!!
I'd definately be interested to hear about the results of a comparison (unless others tell me that it's been done hundreds of time and that the result is known --- I don't know much about heatsink lapping).

wooglin
Patron of SPCR
Posts: 258
Joined: Thu Sep 09, 2004 9:54 am
Location: FOLDING IN TORONTO, CANADA
Contact:

Post by wooglin » Tue Mar 22, 2005 1:45 pm

Prime95 is useful as a memory tester first, a loading program 2nd.

Don't be suprised if P95 fails. It's just a sign that your memory is not 100% stable.

My original ram would fail P95 tests after 8-9 minutes, but no system stability issues with any other programs.

Replacement ram (Kingston HyperX matched pair) runs P95 perfectly for an extended period.

Just because P95 fails, doesn't mean there's anything wrong with your HS or CPU Cooling.

Now, if the CPU was reporting 75 degrees right before failure... OK, you might have a cooling problem. But your temps look fine. Don't worry, be happy.

Edit - sorry if late... I was responding to one of the earlier March 2nd posts.

sthayashi
*Lifetime Patron*
Posts: 3214
Joined: Wed Nov 12, 2003 10:06 am
Location: Pittsburgh, PA

Post by sthayashi » Tue Mar 22, 2005 2:06 pm

wooglin wrote:Prime95 is useful as a memory tester first, a loading program 2nd.
Actually, Memtest86 is a memory testing program. Prime95 is a stability testing program (amongst other things). It's better for testing your undervolts and underclocks.

wooglin
Patron of SPCR
Posts: 258
Joined: Thu Sep 09, 2004 9:54 am
Location: FOLDING IN TORONTO, CANADA
Contact:

Post by wooglin » Tue Mar 22, 2005 2:47 pm

sthayashi wrote:
wooglin wrote:Prime95 is useful as a memory tester first, a loading program 2nd.
Actually, Memtest86 is a memory testing program. Prime95 is a stability testing program (amongst other things). It's better for testing your undervolts and underclocks.
I'll defer to the gentleman with the exponentially higher post count. :wink:

Jan Kivar
Friend of SPCR
Posts: 1310
Joined: Mon Apr 28, 2003 4:37 am
Location: Finland

Post by Jan Kivar » Wed Mar 23, 2005 1:35 am

wooglin wrote:
sthayashi wrote:
wooglin wrote:Prime95 is useful as a memory tester first, a loading program 2nd.
Actually, Memtest86 is a memory testing program. Prime95 is a stability testing program (amongst other things). It's better for testing your undervolts and underclocks.
I'll defer to the gentleman with the exponentially higher post count. :wink:
Prime95 also stresses the memory (depending what stress test is being run). Too tight memory settings and Prime95 will fail, even when the computer seems to work OK otherwise. All Prime95 failures aren't related to CPU; this is mentioned in stress.txt (in the Prime95 dir).

Cheers,

Jan

Thorz
Posts: 69
Joined: Sun Feb 13, 2005 8:01 pm

Post by Thorz » Wed Mar 23, 2005 2:54 am

Thank you all for contributing more to this thread. As I stated before the fail appears to be heat related because with the case open Prime doesn't show any problems.

I have receive my new Antec case and will post a follow up in the site when I have finished the transplant to the new house for my machine :)

Post Reply