Gpu overheat alarms & shutdowns...

chuckberg

Distinguished
Jan 10, 2011
22
0
18,510
Hello,
My friend just put a Corsair 850 watt power supply in his HP P6404y machine and tried to upgrade the Nvidia 9100 stock video card to an Nvidia GTX 460 and the machine would run fine, however when he tried to play games the machine would start beeping and shut down which is usually a symptom of a GPU overheat condition. He then returned the GTX460 and upgraded to a PNY GTX 470 thinking his first card was from a no-name vendor and the same problem occurs. He then put a large fan next to his computer with the side off and if the machine starts beeping he turns that on and can continue to run, but I told him that option is not recommended. There is nothing in the Bios to see what the settings are for heat monitoring on a GPU, and there is no log generated showing what the cause of the shutdown is. We have looked online for help and there is none available for this Pegatron m2n78-la motherboard made by Asus for HP. Is there any software
that allows us to monitor and change the settings for heat ranges on a GPU, or is there any other way to identify what the machine is monitoring for heat so we can try to fix this. The bios options are very limited so we really are stuck on this one. Please help!! Thanks





Chuck

P.S. Here is a link to the system specs. from HP.

http://h10025.www1.hp.com/ewfrf/wc/document?docname=c02053862&tmp_task=prodinfoCategory&lc=en&dlc=en&cc=us&product=4176379&lang=en
 
The jump from a GTX 460 to a GTX 470 is a good performance upgrade, however will make things WORSE temperature wise. The GTX 470's run VERY hot to begin with.

What you need to do however, is run a Temp monitoring program to log temperatures of your GPU and CPU so you can determine what's actually overheating. The case used for the HP P6404y isn't designed for high end cooling by any means. So you may very well just not have adequate cooling. Not to mention, they are probably using minimal/stock cooling for the CPU to begin with. If that CPU cooler is clogged with dust, the CPU could be what's overheating.

HP P6404y : AMD Phenom II X4 820 + 8GM RAM + M2N78-LA (Violet6) Motherboard
http://h10025.www1.hp.com/ewfrf/wc/document?docname=c02053862&tmp_task=prodinfoCategory&lc=en&dlc=en&cc=us&site=null&key=null&product=4176379#N67
c02540588.jpg


Just looking at the case, I'd say there's not enough air intake to keep the CPU or GPU cool when gaming. Most aftermarket cases have a fan in the front to suck in air. This case doesn't have a sufficient air intake in the front.

You can use GPU-Z and CPU-Z to monitor temperatures of those individual items. Or you can download one of many other temp monitoring programs to watch both at once. HWMonitor is a good one (made by same folks as CPU-Z) that will allow you to see temperatures of the CPU, GPU, Motherboard, hard drives, etc. All at once even!

Have him us the HWMonitor program while gaming. He can close out the game after 5-10 minutes and look at HWMonitor. That program will show you what the "Max" temperature of each component during the time it was running. Give us those temperatures and we can point out which item was the likely problem. Screenshots help too!
 

chuckberg

Distinguished
Jan 10, 2011
22
0
18,510



Thanks for the info!! He did make sure the CPU cooler was clean, along with everything inside the case, but I think you are right about the cooling capacity of the case. Funny part is the problem is almost instantanious. He tried to run the built in Windows 7 Assesment program and it too makes the machine beep just benchmarking the graphics. It did that on both the GTX4600 and the GTX4700 cards. He did download and run a program that shows the machine is only allowing the heat on the GPU to get to 60 degrees celcius, and we are hoping there is a program that allows you to adjust that setting. Nothing in the Bios or on any included software from HP will let us do that. From Nvidia the GTX4700 can run up to 105 deg. celcius, so we really need to change that setting on his system. Do you know of any software that lets you change the system setting for overheat limits??

Chuck
 
You cannot override the temperature limits. In fact doing so would be incredibly stupid. The GTX 470 limits itself to 105C because anything hotter than that will destroy the card. (Keep in mind, 105C = 221 degrees Fahrenheit, where as 60C = 140F)

Now, if the software program (which you haven't named) shows the GTX 460 only reached 60C then that's not your problem. You likely have something else causing the shut down. If the GPU is only reaching 60C you might want to monitor your CPU temperature.

If that temperature is safe, you may very well have some other kind of hardware problem. Could even be the power supply causing a problem once the CPU & GPU are trying to pull a large amount of power through it. Make sure the GTX 470 has all the power connectors attached to it. It requires TWO (2) 6-pin PCI-E power connections from your power supply. It needs a hefty amount of electricity coming in. If he has a Corsair 850W power supply it will more than likely have the required power connectors for the video card.
 

chuckberg

Distinguished
Jan 10, 2011
22
0
18,510





I agree that tampering with the settings would be stupid, but if the limit is set incorrectly for an onboard video card then there should be a way to tell the system what the new boards specs are. I have emailed my friend to get the name of the software he is using so we all are on the same page. I was pretty sure the software told him what the max range was not what the actual temp was, but I will confirm that with him. I will get back to you later. Thanks again for all of your help!!

Chuck
 
Download HWMointor,it should list everything from the temps to fan speeds to voltage's.

Has you or your friend done any Overclocking to the GPU or CPU?If their are beeps coming from the mobo then that would lead me to think it's the CPU overheating not the GPU.

It is very possible that the GTX460/470 is overheating the CPU because of how much heat it is giving off(they are such large cards that if you don't have good airflow it will bake everything inside the case).Espically because it's an HP case and a very small Mid Tower case at that with little to no airflow.

I reccomend going and checking out those temps in HWMonitor, i doubt it's your GPU so watch the temps of the CPU right before it crashes and report back.

EDIY:Just looked up your CPU and the max temp is 71c so make sure it doesn't go above that.
 

chuckberg

Distinguished
Jan 10, 2011
22
0
18,510
I will have him do that. I know everything is stock, no overclocking, unless PNY out of the box is slightly tweeked. I still am troubled by the fact that the overheat happens in 10 seconds during video testing and simple gaming. None the less I will have him download and run HWMontior and report the results.

Chuck
 
Yeah sounds like CPU overheating to me as well. If a GPU overheats it won't beep it'll just BSOD and restart (usually).

Also whatever he's seeing about 60C max doesn't matter. The card isn't going to cause any issues at 60C, more than likely that is simply the point where the fan speed goes up. Considering having the side open and a fan blowing in helps the situation, then definitely it's a case air flow issue and probably some other component like that CPU overheating.

HWMonitor, as mentioned, is a great program which will show everything you need right now - core temps, gpu temps, mobo temps... it'll read every sensor. You can also see if the PSU voltages are what they should be.

I have a Corsair 750TX and it's a great PSU, unless you somehow got very unlucky with it I'd say your 850 is fine. Maybe something got plugged in wrong?
 

chuckberg

Distinguished
Jan 10, 2011
22
0
18,510
You know what...That is a good thought! The problem keeps happening during video testing, but the heat generated could be pushing the CPU out of range...Hmm....And I am certain that it is a stock CPU cooler. I will confirm the HWMonitor results asap. Thanks guys!!


Chuck
 


Exactly what I've been suspecting. That HP case doesn't appear to even have a front air intake at all (at least not a real one). it's just a bake box. So the CPU is cooling itself with whatever air is pulled into the case by air pressure from outside (and the negative pressure from exhaust fan). The GPU is probably cooking the processor.

Should find him an inexpensive gaming case to put all his equipment inside of. LOL Course he'll have no warranty what so ever though. Granted HP probably won't warranty it now anyhow since the case has been opened.
 
If it does appear to be the CPU thats overheating i would go ahead and return it right away and get a new one before anything else happens.

I would also reccomend getting a new case even if it isn't the CPU overheating.If it's just a heat issue in general,getting a new case is just the better option.
 


? You're telling him to return the HP built computer because the video card is making the CPU overheat? Or did I misunderstand?
 


Well if it's shutting down to a CPU related overheat the chip could be damaged from it and shorting it's life span.I'm unclear if overheating would start to cause other problems as well.

Just put the orignal PSU that it came with and send it back for a whole new one,or contact AMD and see if they can replace your chip.
 


Doubt it on both counts:

1) AMD won't warranty a CPU sold by a PC distributor. That's the PC distributor's job.

2) HP probably won't accept a return, let alone warranty, on the PC because the case has been opened. They usually put some kind of sticker over the side panel so you have to break the sticker to open the case. That way they know it's been tampered with internally.

If you tell them you want a replacement because it has overheated, they're going to automatically talk about whether you've OverClocked, etc. It might be an uphill battle to get HP to replace the whole PC, let alone just the CPU.

That would be my guess anyhow. But who knows, HP could prove me wrong.
 

Mark _01

Distinguished
Jan 10, 2011
1
0
18,510
First, thanks Chuck for starting this blog for me.

To start - MY temps in C at idle with only explorer and word running no fan, closed case.
NVIDIA GEFORCE GTX 470 GPU - 47c (Speedfan)
CPU TEMP - 46c

(HWmonitor below)
NVIDIA GEFORCE GTX 470 power - .88V Value - 1V Max

TMPIN0 37 value - 38max
TMPIN2 20 value - 26max
AMD Phenom II x4 820
Core#0 25 value - 30 max
Core#1 "" "" - ""
Core#2 "" "" - "" ""
Core#4 "" "" - "" ""
Processor power - 32.80 value - 95.40 max

After running Stone Giant for 10 seconds and hitting the buzzer (again closed with no fan) Max temps

CPU temp - 63

TMPIN0 37 value - 40max
TMPIN2 20 value - 27max
AMD Phenom II x4 820
Core#0 25 value - 32 max
Core#1 "" "" - ""
Core#2 "" "" - "" ""
Core#4 "" "" - "" ""
Processor power - 32.80 value - 95.40 max

now with fan on and case open at idle

NVIDIA GEFORCE GTX 470 GPU - 39 (Speedfan)
CPU TEMP - 36c

(HWmonitor below)
NVIDIA GEFORCE GTX 470 power - .88V Value - .88 Max

TMPIN0 30 value - 30max
TMPIN2 20 value - 20max
AMD Phenom II x4 820
Core#0 25 value - 32 max
Core#1 "" "" - ""
Core#2 "" "" - "" ""
Core#4 "" "" - "" ""
Processor power - 32.80 value - 95.40 max

After running Stone Giant for about 10 seconds and hitting the buzzer open with fan Max temps

CPU temp - 48 at buzz (note that it shot up in 5 secs hitting all the alarms)

TMPIN0 30 value - 31max
TMPIN2 19 value - 27max
AMD Phenom II x4 820
Core#0 25 value - 32 max
Core#1 "" "" - ""
Core#2 "" "" - "" ""
Core#4 "" "" - "" ""
Processor power - 32.80 value - 95.40 max

Question, is it possible that it is viewing how fast the temp goes up expecting that it will fail?









 
Can you plz provide screenshots of the HW Monitor at idle and right before you think it will crash.It is way easier to decifer what is actually going on if i can see it with my own eyes.~Please.

Just something to try out,it shouldn't really matter much but, go into your BIOS and turn off AMDKool&Quiet.Unless your really really trying to save money then you won't need it on.
 

chuckberg

Distinguished
Jan 10, 2011
22
0
18,510
Well we basically have it narrowed down to the CPU overheating. While testing with case closed we saw the CPU temp spike up to 73 degrees Celcius and the machine started the beeps. Not sure if this is a AMD Phenom issue but we keep thinking the CPU should have no major load increase with a better video card, and the GTX 470 video card ejects the vented heat out of the rear of the case, but for now we are working directly on lowering CPU temps. The stock cooler is being replaced by a Spin Q Black widow cooler and once we get that installed and tested we will update you folks with the results. It is a large ah heck for sure, as shown below, but hopefully it quickly pulls the heat from processor. As far as a screenshot he keeps trying to post one but for some reason the site won't let him post. We are working on it. Thanks again for all of your insight!!!

Chuck

1252793_5465061_290.jpg



 

chuckberg

Distinguished
Jan 10, 2011
22
0
18,510
Well guys, more bad news..... The new heat sink had no effect on the overheat problem. The fins on it were so large he couldn't even put the cover back on the case. So with the side panel of the case off and the new heatsink on he still had the over heat alarms when doing graphic intensive work. We did notice some posts online about problems with the motherboard sharing video memory, and since the new card has DDR5 memory and the shared memory is DDR3 we think the issue is hidden in there somewhere. There is no option in the bios to disable shared memory. For now our next attempt this evening is to swap out an existing GTX260 card I have into his case and see what happens there. The GTX260 used DDR3 and is a much cooler card so something in that combo should work. We will keep you posted. We will also look into upgrading the Northbridge software for the mobo and see if there is somethng possible there. Wish us luck!!

Chuck
 
Screenshots of HWMonitor would still be very interesting to see. None of this really makes much sense to me. I'm not so sure this is an overheating issue, at least not the CPU.

I wonder if there's an overheat on the motherboard itself (North/Southbridge for example).