Dual Processor overheat?

gaius

Member
Mar 6, 2001
99
0
0
We have these Dell workstations, the 420. They came as single cpu, but can be dualed. so we dualed many of them to do rendering recently because its cheap.

The problem is, it seems that the workstations with DUAL CPU have very high rate of video card dying. Also, some of the machine have random freezes. The single CPU ones has much lower rate of it going bad, by alot.

Also, some users said that the video in the screen would be messed up after a while. But if they shut it down and let it cool for a bit, it would be ok again.

I thought all these points to heat problem.
Dell's 420/410 cooling is set up this way. No fan on the CPU. They have a huge HS on their CPU (slot). Then they have a rear case fan with a wind tunnel from the fan to the cpu/HS. I put my hand on the fan, and it seem to be pretty weak.

I installed Sandra to look at the temperature. and its showing 67 degree Celcius!
I called Dell many times and complained about it. But they only would say "there have been no heat problems with these machines" and that the fans can cool the workstations fine, even after i told them the temperature and the symptoms. They just blame it on NT for the crashes. Then they blamed its Ghost images' fault.

I talked to my executive account manager and he set me up with a higher level support guy. He is much nicer than those level 1 people and dont just jump to conclusions right away. I told him the sympoms. He think its application/windows problem, because there is no error messages about heat in the event log.
He asked me to reboot to BIOS and leave it on after work and see if it would crash because of heat. He think if it crashes, it would be heat problem, not application/windows. I mentioned the 60s degree celcius problem, he said he is troubled by it, but want to do that test first.

I first thought it made sense, but the more i thought about it, i found it to be faulted. 1)when you just do bios, you are not doing much processing, which doesnt heat the cpu that much. that kinda made it pointless. 2) I've never heard of pc freeze/crash in bios mode. Is it even possible?


The user of the machine needed to do renderings for this whole week because of deadline, so he couldn't do this test yet. So during that time, i went online and ordered a CompuNurse to check the temperature. Since the HS is so big, i cant put the probe next to the core. so i just put it on the HS, and the temperature i got is 59 degree celcius. Thats with the case opened.
I then check the video card, but Dell's PSU is on the buttom which block me access to probe the graphic card's core. so i just put the probe on the top of the board, and it got 42 degree celcius.


What do you guys think about the temperature? Is that normal for a Dual? because for singles, normal temp should be 30-40s.
Also, do you think like me , that heat is the problem?

Lastly, do you think going into bios to see if it would freez/crash is a good test? like i said, is it even posssible to crash in bios because of heat?

thanks in advance.

oh yea, the spec of the 420.
they are P3 500 mhz to 1 ghz. duals. video card is usually tnt2 or firegl, few have GF 1. RAM are usually 256-512 mb rambus i belive.

 

Drakkhen

Senior member
Nov 9, 1999
824
0
71
Well, I think that what you need to look at is where it is messing up. If the images are becoming messed up when they are rendering, it would appear the the Videocards are the ones messing up.

Are the ones that are messing up based on a similar video card? If so, you might want to try and leave the case open and see if it messes up after that.

If the CPU was overheating, it would probably lock hard on you, not mess up the video output.

Just a thought.
 

gaius

Member
Mar 6, 2001
99
0
0
well, i think the heat damaged the video cards, which cause the video output corruptions.

Also, some people get random freezes/crashes 2-3 times per day.
 

ripthesystem

Senior member
Mar 11, 2002
571
0
0
I've run duallies for years now (even got one at home- see my sig) and I almost laughed outloud when you said the tech guy blamed the NT ghost images on this.. hAHAHAH!

wow.

also Many BIOS temps and temp monitor programs are only setup for 1 CPU. If this is the case and you have 2 CPUS it is not uncommon to see high temps because in order to show the temp it will add both temperatures together.. therefore 67C may actually be a reading of 33.5C for each CPU (or 1 @ 40 and 1@ 27.. you get the picture) although 33.5C would be pretty unlikely w/no HSF on the CPU.

If the 67C is for each CPU- that's BAD. way bad.
If 67C is for the Case temp.. that's not good either.

And either of those scenarios could cause your problems.
Add up a high CPU temp plus some warm Rambus and a Rendering Video Card and you've got some serious heat issues on hand.

If heat isn't the problem I have to ask the basics:
are the paired CPU's identical? Matched Stepping too?
all drivers in order? Are there other hardware issues at hand here.. some things (esp on older rigs) didn't play well with VIA chipsets.

I would start with the heat. If what Dell has in there isn't working.. ditch it and go for something you know will.

cold CPU's are happy CPU's
:D
ripthesystem
 

Colt45

Lifer
Apr 18, 2001
19,721
1
0
slap some extra fans on one of the rigs. have a decent 80mm intake and exhaust fan (if there are places for them)

try putting on some little 50mm fans on the cpus.

If there is more stability i guess youve found the problem..

does dell get pissed off if you modify their rigs?

Going into the BIOS to see if it is going to crash is a joke of a test.; because theres basicaly 0% load on the CPUs.
 

gaius

Member
Mar 6, 2001
99
0
0
ripthesystem,

67 degree celcius is with sandra.

59 degree cecius with a CompuNurse temperture probe. Like i said, i was not able to put it next to the core, so i can only put it on the HS. so the real core temp should be higher.

The paired CPU should be almost identicle, since both are from dell. i know the speed is the same at least.
And its Dell, so its Intel mobo.


Can you tell me your temperature of your dual is?
 

foofoo

Golden Member
Mar 5, 2001
1,344
0
0
hi,
the heat sounds like it but you also might look into power supplies. these symptoms could also be an inadequate power supply.
good luck
 

gaius

Member
Mar 6, 2001
99
0
0
Colt45 & ripthesystem,

Yea, I dont think Dell would be happy if i add fans and other stuff. Their tech prob will say since i mod it, they wont support it.
So i can't really add in components myself, unless they are from Dell.

I am glad people here confirms what i belives. :)
 

ripthesystem

Senior member
Mar 11, 2002
571
0
0
granted - dell may not appreciate it.. but do you appreciate paying for a computer that doesn't work?

Besides.. if you add on a few fans etc, they are the easiest things to remove if you have to send something to Dell for inspection. Dell doesn't even have to know about them. But they might save you some grief.

Try to narrow it down to a heat problem for sure though. As was stated above a heat problem would most likely lock up the whole system instead of giving video garbage.. and if you think the vid card got fried then why would it be ok when you reboot. There may be more than meets the eye here.

ripthesystem
 

gaius

Member
Mar 6, 2001
99
0
0
well, wouldnt running 60 degree celcius, on 2 CPU, not just one cause heating problem? I mean the PC is on 24/7, and its running 60 degree constantly.

and wouldn't it do damages to components near it? the video is RIGHT next to the cpu. so i thought thats why its having problem.
to clearfy the video card situation. We had to replace many of the video cards because you can't read anything on the screen. But of the people that has the video cards that is not completely dead, some people do have the problem of getting bad video output until they shutdown and let it cool.

Then, there is the random freez/crash problem, unrelated to video card.

I want to buy some fans too. But we have many duals. and we'd have to spent lots of money to put fan on all of them.
I want to buy some fan on this one pc, and see if it helps. But the heat could have damaged some chips and other componts already.

The rear case fan i want to get is this:
92x92 fan
73 CFM. but its 47 dBA, and its 32 mm thick, which is thicker than the DELL fans. so i am afraid it might not fit.

I also want to buy a system cooler:
cooler

But dell have their PSU from the agp down to buttom of the case. so i won't be able to hook it up to the slot screws. prob have to tape it or tying it with string or something.
 

ripthesystem

Senior member
Mar 11, 2002
571
0
0
Hmmm. Thanks for the clarification. I didn't totally understand the vid card problem.

WOW those fans are noisy.. AND expensive.
You can find 92x92 Panaflos here for $4.00 ea. They also have 80x80 Panaflos too and better deals on others as well. I think you need to spend $25 min. here..

Also BGmicro.com usu. has Panaflos for $2-5ea.

Basically I'm pretty sure you can get a better deal than what you suggested.. even if you use lower volume fans that means you could get 3 quiet panaflos for $12 and that would give you great coverage on your whole case (1 in from front, 2 out from back) That might make things tolerable as far as case temp goes.. if you are worried you might do well with 2-3 higher flow 80mm fans instead.. but the ones you suggested are gonna sound like a hairdryer..

also is there anyway you can move the vid card so it's not raping the CPU's?
hth
ripthesystem
 

ripthesystem

Senior member
Mar 11, 2002
571
0
0
another thing I'm not sure about is how the dell heatsink sits in there. Is it actually on the CPU? if so a small fan on the heatsink could definately cool the sink down and keep the CPU @ a good temp.

rts
 

gaius

Member
Mar 6, 2001
99
0
0
I really don't care about the noise right now. i want to get as much cooling as possible and see if i can lower the temp to normal.
Dell does alot of weird stuff that makes adding fans hard. ie, because of PSU at buttom and the stupid small speaker, you cant add a front case fan.
I just want to see if cooling helps to remove the problem of the random PC crashing. If it does, hopefully we would be able to get Dell to pay of the extra fans for all the dual workstations.

As for the videocard, i can't move it. there is only 1 AGP port and it has a bar holder over it to hold it in its place.

The DELL HS is big monster on the cpu. Its a mirror shinning metal. it pretty much took over the space of the fan.
I am hoping that Globalwin system fan will help circulate the air around it. Have to figure out a way to hook that fan though.

i ordered an extra 92x92 fan to put in the HD cage. Also ordered an extra 80x80 fan in the rear above the other rear fan.
not really sure if those extra fans can be fitted in right. but might as well order them. da boss OKed it already :)

heh, yea,those fans are very loud. avg 41 dBA each. heh.

One of the places i ordered from is NJ, and i picked ground UPS. hope i'll get it monday next week.
 

ripthesystem

Senior member
Mar 11, 2002
571
0
0
cool

let me know how this works for you. the fans don't have to sit in ther nicely.. just as long as you can get them to do their job.

After I started modding my first PC years ago I ended up with 2 fans being held on my tape, string, wire and a 4 inch screw. I've gotten much better asthetics over time but they did work.

Let me know how it turns out!

ripthesystem
 

Colt45

Lifer
Apr 18, 2001
19,721
1
0
It might be a good idea to get a pair of some low flow 60mm fans to pop on the cpus too.

if you just want to test and see if the problem is heat, I'd say open up the case, slap a 60mm on each CPU, and let keep the case off.
then do some high stress stuff.. 3D video, move big files, etc.

if the stability problems go away, you know that its heat. then you can figure out a suitable way to get more cross-case airflow, as well as the fans on the heatsinks, and impliment it on all your duallies.
if you look around, theres some places where you can get a whole case of 80mm fans for very reasonable prices
 

gaius

Member
Mar 6, 2001
99
0
0
OK.
I installed the 72.5 CFM fan as rear case fan and the 62 cfm as system fan.
now, its 38-39 degree celcius! When i did rendering, temp was still at 38-39!
And it haven't crashed once yet. The guy who use the pc is out sick. But i've been using remote control software to do some rendering.
Its great so far.
Hopefully when that guy is not sick anymore, he can test it more with what he usualy run.

The ball is in Dell's court, now i have all the proofs to back my claim.
 

ASK THE COMMUNITY