Edit: Something is broken, and I think its heat related!

InternetDuder

Member
Jun 12, 2005
47
0
0
Hi.

My video card:
http://www.newegg.com/Product/Product.asp?Item=N82E16814122209
has an overheating problem... I haven't overclocked it at all, the case has an 80mm and a 120mm fan both blowing towards the video card. The card's fans are working properly, and the surface of the heatsink is close enough to the readout temp to assume proper contact. This thing is just too hot!!

Whenever I'm playing a game, I get the ol' stutter once every 5 minutes -> stutter once/minute with crappy fps -> blue screen. I don't like this. I now have a table fan pointed at my open case on high (it is idling at 52' WITH that). With the fan off, and 5 minutes of playing HL2, it reached 67`. To contrast, the CPU has been running prime95 for an hour now, and has not yet passed 40`. What could be doing this??

CPU: AMD 64 3200+ Venice
Mobo: DFI Lanparty UT nf4 Ultra-D

The clock settings are at 350/1000 right now. I think it is a problem with the fan staying on low speed for some reason.
 

Deadalus

Member
Jun 15, 2001
129
0
0
Originally posted by: hatim
MY 6800GT idles at 60C....

52C is supposed to be great for 6800GTS

At load mine goes to 78C

My 6800GT has the same temps as hatim, it was running even hotter before I installed a Zalman VF700 cooler. Never had any problems with it running at those temps...

Perhaps the cause of the suttering lies somewhere else? Do you have another video card you can try?
 

Operandi

Diamond Member
Oct 9, 1999
5,508
0
0
That's a normal temp for high powered card like that 6800GT, your instability must caused by something else. Typically signs of an overheating card are graphics glitches and artifacts not stuttering and poor performance.
 

Continuity27

Senior member
May 26, 2005
516
0
0
My two idle at 67C and at maximum load hit 99C! :Q

Which is odd considering I get no instability or artifacting at all... everything I throw at them comes back perfect, despite being hellishly hot.

They were set to throttle at 135C... and at least I have my warranty! ;)

Oddly though, they used to be a lot cooler, my room didn't change temperatures that much, but I did change my motherboard. Is it possible I'm getting the wrong temperature readings?
 

InternetDuder

Member
Jun 12, 2005
47
0
0
I have another video card I can try, but I don't see how the stuttering can be anything but temp. I'm surprised these temps are normal, they seem so high. It is strange that at those temps, it stutters, and then when i put a fan blowing on the case, it runs fine. Ive tested with and without the fan a few times, and it is just how I would expect an overheating card to act. I'm gonna try my Geforce 3 in it, and see what happens. Looks like the card is just poo :(
Thanks
 

InternetDuder

Member
Jun 12, 2005
47
0
0
A little more testing with the 6800:
first stutter: 68'
stuttering is <1min apart: 71'
crash: 72'

Took about 20 minutes of playing to crash. During the last 5 minutes, there were some artifacts, but that is probably unrelated. Testing the GF3 now.
 

Continuity27

Senior member
May 26, 2005
516
0
0
It could be heat related still. Another possibility is that your graphics card is blowing all that hot air and maybe it's affecting the chipset, or other misc parts on your motherboard, that may be what's causing the stuttering. Your video card should be able to handle those temperatures, but you must also think of the components around the video card.
 

InternetDuder

Member
Jun 12, 2005
47
0
0
upon closer inspection, i have no AGP slot. I don't think its the chipset, it has a pretty good hsf on it, and the card isn't really positioned to blow air on it. And the CPU is stock clocked with a very good swiftech HSF, with good temp readout. Plus, if the chipset went, I think i'd have more weird stuff than the standard stutter stutter fps crash.
 

InternetDuder

Member
Jun 12, 2005
47
0
0
Okay, Continuity27, I think you may be right. When I was installing this video card, it bumped the chipset's HSF, which is located unforunately right under the video card. It right now is blocking air, but I didn't think it would really be a problem. I may have jostled it a bit enough to ruin contact, though.

When I overclocked the FSB by 5mhz, it did get a rare stutter, even with the table fan blowing at the computer. Without the fan, and overclocked, it got the stutters much faster than before. I just reapplied the heatsink, and am about to do a test run.
 

InternetDuder

Member
Jun 12, 2005
47
0
0
Didn't seem to affect it much.

I think the problem may be:

http://images10.newegg.com/productimage/13-136-152-02.JPG

in here. With the card in the right-hand PCIE slot, it blocks some airflow to the chipset. The PSU's power connector for the card doesn't reach the left slot, and the adaptor I have (2 4-pins in, pcie out) can't work, because I only have one free 4-pin connector, which is frustrating. Before I go out and buy some Y-cables or extensions or something, will a single card work in the lower PCIE slot? I know it's made for SLI, so it might have a primary/secondary hoo-hah.

Thanks.
 

Operandi

Diamond Member
Oct 9, 1999
5,508
0
0
Originally posted by: InternetDuder
Didn't seem to affect it much.

I think the problem may be:

http://images10.newegg.com/productimage/13-136-152-02.JPG

in here. With the card in the right-hand PCIE slot, it blocks some airflow to the chipset. The PSU's power connector for the card doesn't reach the left slot, and the adaptor I have (2 4-pins in, pcie out) can't work, because I only have one free 4-pin connector, which is frustrating. Before I go out and buy some Y-cables or extensions or something, will a single card work in the lower PCIE slot? I know it's made for SLI, so it might have a primary/secondary hoo-hah.

Thanks.

The chipset doesn?t need any air flow it already has its own fan.

I would stop running tests with games for now and install prime95 to test the CPU and RAM and all sub systems. Run each test for an hour or so and see if you encounter any errors.
 

Operandi

Diamond Member
Oct 9, 1999
5,508
0
0
Originally posted by: InternetDuder
I ran prime95 for like a half hour already with no errors, what other tests are you referring to?

Prime can run different tests geared to stress various parts of your system, you should see small description next to each one. One tests primarily the CPU, while one test mostly RAM and I think the third is combination.
 

InternetDuder

Member
Jun 12, 2005
47
0
0
Ok. After talking to a friend, I'm gonna run prime95 one night, memtest86 the next, and see if either have any instability. If prime breaks, I blame CPU. If memtest breaks, I blame chipset or ram. If neither break, I blame video card. All will be running ~20 hours, so I think its a reliable sample. Any final input would be appreciated
 

Operandi

Diamond Member
Oct 9, 1999
5,508
0
0
Originally posted by: InternetDuder
Ok. After talking to a friend, I'm gonna run prime95 one night, memtest86 the next, and see if either have any instability. If prime breaks, I blame CPU. If memtest breaks, I blame chipset or ram. If neither break, I blame video card. All will be running ~20 hours, so I think its a reliable sample. Any final input would be appreciated

You could try something like an equivalent of ATi Tool to test your card. I'm sure there is something similar for nVidia cards.
 

InternetDuder

Member
Jun 12, 2005
47
0
0
I don't think I need to... prime95 failed. The CPU temp was recorded somewhere around 40C. Think it has bad contact?
 

Elcs

Diamond Member
Apr 27, 2002
6,278
6
81
Originally posted by: InternetDuder
I don't think I need to... prime95 failed. The CPU temp was recorded somewhere around 40C. Think it has bad contact?

Its possible but unlikely. My XP 1700+ would reset itself if i put any stress on it, cpu temps read around 46-47C... It turned out that my Heatsink wasnt getting a good contact with the cpu.

I think a process of elimination needs to be carried out here. Eliminate 1 problem component at a time at stock speeds and voltages. Making notes on everything you can.

I cant particularly find anything else to try beyond what has been suggested.
 

InternetDuder

Member
Jun 12, 2005
47
0
0
Everythign is already running stock voltage and clock speed. I can't test other systems, nobody else in town has a AMD 64, or a 939 motherboard. After work I'm going to recontact the CPU. But for now, since prime95 DID fail, that DOES mean a problem with the CPU, correct? Or could it mean ram? I Have 2 ram chips, so I could test them individually. I don't have the error with me, but it was something like
"Expected 0.4, got 0.5!"

At least I can count out the video card... maybe I should make another topic. (edit: changed topic title)

I seriously still believe it's heat related though, because the only time I get stuttering in games is when the fan is not blowing on the motherboard. I should try running prime95 again with the fan blowing on it.
 

CrispyFried

Golden Member
May 3, 2005
1,122
0
0
40 c on the cpu doest seem bad. The temp is read between the socket and cpu (?) not the heatsink. So if the hs had bad contact it should read higher.

Prime failing could be anything from cpu to mem to something on the mobo. But it points more to cpu or mem than anything else. Definately pull and reseat the cpu hs with new thermal compound.

Can you run rthdrible for a while? That would pretty much isolate the vid card (although it seems youve already done a good job ruling that out). Run it alone, then with prime.

You can also try pulling and reseating the RAM and other cards, thermal expansion may be a factor but usually that results in lockups.

What case do you have? There are also slot coolers that have "extentions" that can put outside air where you want it, you could have it directing air right at the northbridge. Im sure that vid card blows some seriously hot air right at the northbridge.
 

InternetDuder

Member
Jun 12, 2005
47
0
0
I'll try that. And are you sure that's where the sensor is? I know it's there on the chips that have a space in the center with no pins, but in this socket 939, theres no area of the bottom of the chip that doesn't have pins.