dual xeon w/passive heatsinks overheating

rk2004 · Post by **rk2004** » Tue May 10, 2005 3:40 pm

i've just installed passive coller master heatsinks
on my dual 2.4 xeon CPU's, and after starting up the system
the temp. went up to 60 degree within less than 5 minutes.
What could I do wrong? Is it thermal conductivity issue between
cpu and the heatsinks or fan behind it. Which by the way is
typical case fan, 80mm Sunon , blowing away from heatsinks.
Should I reverse that fan or add another one to blow air directly
on to the heatsinks? i know that i did something stupid.
Please let me know what.

halcyon · Post by **halcyon** » Tue May 10, 2005 10:25 pm

I suggest you read a bit in this forum. Old posts about what kind of CPUs can be used with passive heatsinks.

Your Xeons don't qualify.

It's just not feasible.

People are running passively CPUs with 1/4 of the thermal output of your Xeons AND still having problems with heat.

I don't think there is a such a miracle passive cooler on the market (yet) that would enable you to cool dual Xeons (2.4GHz) passively.

If you are using a fan (I'm not sure I understood your description), then try to position it so that it gets fresh air from outside the case and blows it onto the heatsinks.

Also, if you have not already tried, remove all thermal "tape" from the bottom of the heatsinks and replace that with proper thermal paste (e.g. Arctic Silver 5).

yeha · Post by **yeha** » Tue May 10, 2005 10:28 pm

those 'passive' 1u and 2u heatsinks are only passive in the sense that there's no fan attached - they still require huge amounts of air to be shifted through them to adequately cool.

1u/2u servers i've stood near created a fearful amount of noise, which was front/back fans forcing air through the chassis internals and thus the 'passive' heatsinks.

rk2004 · Post by **rk2004** » Wed May 11, 2005 7:22 am

Now you got me totally confused, so let me ask you this: is there a way to achieve quiet Dual 2.4 Xeon System at all? Since I'm using CAD & Studio 3D Max apps. for most of the time i believe that dual system is a must. And I see big difference in performance between my current dual system and my old single P4.
Spending 8-14 hours a day next to typical Intel's Setup (60 mm fans on CPU's) is unbearable for ears. So right now I’m about to do anything to get rid of this noise, including complete PC replacement, case replacement or building insulated enclosure around the whole case. As far as PC replacement goes the question is: Would 3.7 Ghz system w/hyper-threading enabled performed as well as my current setup in CAD & 3d Max?

My quest for quiet dual Xeon system began since I bought this PC, about 1.5 year ago and it looks like is going nowhere. If you look at my previous posts you will see that I also tried swiftechMCX603-V heatsinks w/80 mm fans on it, the result was none, the noise was still the same as w/60 mm intel fans. Even though someone suggested to undervolt the cpu fans w/zalman fanmates I just gave up, I don’t think It would make much difference in noise level. Am I wrong? Some SPCR users claim that they achieved relatively quiet dual Xeons, and I’ve seen SPCR using Cooler Master E3W-NPTXC-01 or E3W-NPTXS-04 heatsinks and reporting low temps and low noise, so what’s the cut?
Any detailed info, guidlines on designing quiet dual Xeon greatly appreciated.

Now as far as my last post “overheating dual xeon” goes, the cpu’s were overheating because I didn’t lock the levers on the sockets. I corrected the problem, reversed the rear case fan so it’s blowing directly onto the heatsinks. The CPU temps look as follow: At the boot-up (relatively idle) 38C and 40C, remote temps around 40 after running Cad for about 2 hours the CPU temps went up about 9C on each cpu and remote temps went up 4C. Is that ok?, is it safe to run it like that for another 10 hours? And I did use Arctic Silver 5 in first place. I understand that setup like that requires a lot of air to go through the heatsinks fins. Would replacing my rear current case fan (80mm) with 120mm nexus make any sense? What about exhausting the air from case? I’ve Seasonic S12-500W. mounted almost right above the heatsinks, does it help to exhaust hot air outside? Would adding another 120mm fan in front of case improve air circulation? Or I should just stop it here, go back to standard 60mm fans?

Regards,

------------------------------------------------------------------------------------
CASE: Chenbro SR20503 Mid-Tower.
CPU: Dual Xeon 2400 Mhz
MOBO: Iwill DPI533-SATA 2
MEM: 1GB pc2100 ddr sdram
VIDEO: Nvidia Quadro FX1100

alglove · Post by **alglove** » Wed May 11, 2005 8:18 am

Looking at http://processorfinder.intel.com , it appears that your Xeons have a thermal guideline of 65W or 77W (depending on the stepping) and a thermal spec of 71-73 degC. Your temperatures sound like they are OK, though obviously, you will want to keep an eye on your temperatures until you are comfortable the behavior of your new cooling setup.

Exhausting hot air outside does help, by the way. How does the computer sound now?

sthayashi · Post by **sthayashi** » Wed May 11, 2005 8:19 am

The temps seems reasonable to me.

I have to admit, your subject seemed like an obvious problem, like: "Basketball cannot fit in thimble, please help"

Anyways, yes fanmates probably would have helped in your old setup, depending on the fans that you used. I used a dual Athlon XP setup for a while and those processors have a similar heat output as your Xeons (or so I believe). I used Thermalright SLK-800As w/ Panaflo M1As on them. At 12V, they're very obtrusive. At about 5-6v, they're reasonably quiet (I used to say silent, but my noise floor has dropped in the last year).

Glad you got things working.

Would a 3.7 GHz setup w/ Hyperthreading work as well? I have no idea since I don't use CAD or 3D Max. A more interesting question would be: Would a dual core processor work just as well? I think the answer to that would probably be yes.

One last thought: Your video card might start being a limiting factor in terms of noise. I don't know how the FX1100 sounds, but you should still keep that in mind.

rk2004 · Post by **rk2004** » Wed May 11, 2005 9:29 am

thanks for your responses fellas,

The noise level right now is fantastic, the difference is like day and night, no questions about it. The only buzz i hear now is the one comming from the 80mm Sunnon fan blowing onto the heatsinks w/3000 rpm speed. Although this is acceptable, i'll replace the fan with 120mm Nexus and undervolt it if
necessary.

The graphic card has about 2 dia. fan on it but it's not that noisy and I can live with it.

My only concern now is the airflow and the temp inside the case. Would
another 120mm nexus fan in front (setup to exhaust the air outside)
make a difference? Since all the hot air accumulates in top of case.

well, I guess i got to play with it little.

Once again, thanks a lot.

kesv · Post by **kesv** » Wed May 11, 2005 10:07 am

rk2004 wrote: My only concern now is the airflow and the temp inside the case. Would
another 120mm nexus fan in front (setup to exhaust the air outside)
make a difference? Since all the hot air accumulates in top of case.

It might help some, but most cases are built to suck air from bottom front and exhaust it from top back. If you are going to install a fan to the front you should try to and return the airflow in the 'right' direction. The fact that turning the exhaust fan to blow in helped the temperature, makes me conclude that your case does not get enough air, possibly due to a restrictive front panel in your case.

Keep in mind that a fan up front is often more audible than one in the back, so using other options to increase airflow is preferable to just adding fans.

One option would be to duct air from the sidepanel to your cpus. Making sure the CPUs always get air that has not been heated by any other components allows you to run the CPU fans with lower speed. If you do this ducting make sure to again reverse the fan in the back to blow air out of the case.

Many cases actually have a pretty bad frontpanel that doesn't let enough air through. So if not from the side then you could open the case from the bottom. This is likely to be the best option from the noise perspective. If you go this route, putting some feet under the case that raises it a centimeter or two helps the airflow.

tomcat · Post by **tomcat** » Wed May 11, 2005 12:51 pm

i watercool my dual xeon (2.4Ghz) rig. Radiator with two Nexus Case fans at 12V .. very much silent. There's another 12cm case fan in the back of the case. CPU temperatures right now: cpu1: 36C cpu2: 38C (100% load, folding for SPCR)

Markus

Pjotor · Post by **Pjotor** » Wed May 11, 2005 8:38 pm

tomcat wrote:i watercool my dual xeon (2.4Ghz) rig. Radiator with two Nexus Case fans at 12V .. very much silent. There's another 12cm case fan in the back of the case. CPU temperatures right now: cpu1: 36C cpu2: 38C (100% load, folding for SPCR)

You know, those temperatures at load suggest you can lower the speed, and hence the noise, on those fans quite a bit. Anything below 50 means your fans spin faster than they need.

rk2004 · Post by **rk2004** » Wed May 11, 2005 9:05 pm

Now as far as my last post “overheating dual xeon” goes, the cpu’s were overheating because I didn’t lock the levers on the sockets

Well I was thinking about it again and it didn't make much sense to me as it probably didn't make much sense to you the first place. Quite impossible
unless I smashed the pins on cpus. So I decided to take out one of the heatsinks, and... the CPU came out together with the heatsink.
All pins are fine. How come? If it was locked in the first place. Too much thermal compound? Is there some emergency unlocking mechanism in sockets?

frostedflakes · Post by **frostedflakes** » Wed May 11, 2005 10:01 pm

You could try undervolting your Xeons to reduce heat output. Intel has low-voltage Xeons that operate at 1.3v, so your board should have undervolting options at least that low, maybe even lower.

I used to own a pair of 1.6GHz LV Xeons. They ran at 1.3v (35w thermal spec), and were very cool. I never ran them passive, but they stayed extremely cool with the Intel retail HS/fan assemblies. Also overclocked to 2.8GHz at the default voltage. Not too bad for $100 for the pair.

Only problem is that I doubt your motherboard support voltage adjustments in the BIOS, so you'll have to do socket mods if you want to undervolt. Check out this guide at Datamine.tk (great source for Xeon and ASUS server board info).

http://www.datamine.tk/pages/article.asp?TID=102

ckolivas · Post by **ckolivas** » Thu May 12, 2005 1:55 am

rk2004 wrote:the question is: Would 3.7 Ghz system w/hyper-threading enabled performed as well as my current setup in CAD & 3d Max?

I don't know your software, but basically if your applications are single threaded they derive no advantage from the dual Xeons or the hyperthreading, but would be much faster with a 3.7Ghz processor. If, however, your applications are multi-threaded, the dual xeon setup will be faster than the 3.7Ghz P4HT.

monkeh · Post by **monkeh** » Fri May 13, 2005 1:23 pm

i have a dual 2.8 xeon system, with the same 3u passive heatsinks.

I found it was safer to fix a 80mm nexus fan to each heatsink, and run them at 5v, this pushes just enough air though to keep my xpu's at room temp.

I do however have 2 120mm nexus fans on the case, 1 in (on side paned feeding the cpu's with cool air) And the other is a exhaust fan.

I glued a 80x80mm a piece of perspex to the top of the fans and they hang on the heatsinks perfectly, and aid with directing the airflow to the back of the case.

it was pretty quiet, until i installed 6 hard drives

jamesavery22 · Post by **jamesavery22** » Mon May 16, 2005 10:44 am

just to backup ckolivas's post,
Yes 3d studio max favors multiple processors,

3d studio max bench of a p4ee 3.73

Dual xeon 3.0 setup

Not quite sure if they are the exact same bench. Their descriptions seem to say so.

The dual 3.0 gets twice the performance according to their tests.