New 480 SOC Not Stable At SOC Clocks ?

[Buckster]

new 480 SOC owner - lovely card - nice and quiet

but ran Heaven and artefacts immediately apparent ran OCCT - 500 errors within 20 seconds

tried 3 different driver varients - made no difference

dropped clocks to 770 - and it then managed OCCT for 4 mins with no errors, then had 1 error

dropped clocks to 760 - and finally stable - 1100 seconds of OCCT without an error

so obviously not stable at SOC speeds.  How are these cards tested in the factory ? individually or samples out of batches ?

frustrated - any clues on anything I may be doing wrong ?

cheers

[Dark Mantis]

Hi and welcome to the Gigabyte Forum.

You don't say what other hardware you have especially the PSU , motherboard etc which makes a conclusion harder to come to.

Personally I am not a great fan of the factory overclocked cards as overclocking is an art and as such not every chip will OC to the same extent as the next. In saying that however the factory OC is not huge and normally most chips will accept it with no problem. I thionk you might have just been unlucky and found one of those stubborn chips.

You have proved that the GPU is sound by turning the clock back and then testing it. It would just seem to be a case of it not running happily at the OC speed.

I think that you could try and update the firmware on the card if it isn't the latest  or RMA the card for another one.

[LanderTaker]

Hi,

I have problems with this gpu with the soc bios enabled.
MB: p5b-de
PSU: gygabyte odin 850W
Pro: q6600 overclocked to 3000mhz

When I play crisis 1 or alien vs predator with high or very hilg settings (not tried lower ) the game goes smoothly, but after a pair of minutes the game freezes.

The windows 7 pro is not affected so I can get out of the game and kill the process normally...

I usually see when returning to desktop some small windows talking something like "window manager stoped working".

Please, do you have any ¡dea?

Other question: About firmware, how do you know if you have the lastest firmware version and where could I get the lastest one?


Thanks you very much!

[Dark Mantis]

Hi

If you download GPU-Z and then use it to find out what your current firmware(BIOS) is then you can check it against the website files and see if there is a newer version for download/installation.

How long have you been using the card ?

What temperture is it running at under stress ?

[LanderTaker]

If you download GPU-Z and then use it to find out what your current firmware(BIOS) is then you can check it against the website files and see if there is a newer version for download/installation.

On gigabyte web page?

The bios version of my card is 70.00.21.00.00

How long have you been using the card ?
What temperture is it running at under stress ?

Hmmm... like a week. It is new.
Temperature: 81-83ºC runing crysis 1 at max.
PSB got to 56ºC
The cores got up to 60ºC at the hotter core.
"Crysis have stopped working" after, maybe, only 5 minutes of play.

(I think that alien vs predator lasted like 3 minutes or so before stopping!!)


With kombustor, extreme burning, I have seen the GPU 94ºC after 4 minutes. maintained at the range 94-96ºC after 6 to 8 minutes. Reached 97º after 15 minutes (like max, just after it it went down again). PCB up to 64ºC. CPU up to 60ºC in the hotter core.

MY OS is windows 7 pro.

Thanks you!

[Dark Mantis]

With kombustor, extreme burning, I have seen the GPU 94ºC after 4 minutes. maintained at the range 94-96ºC after 6 to 8 minutes. Reached 97º after 15 minutes (like max, just after it it went down again). PCB up to 64ºC. CPU up to 60ºC in the hotter core.

I am not so familiar with nVidia cards but it would seem to me to be a bit too hot for comfort at those temperatures. Have you got all 8 of the power cables connected to the card ?

Does the fan appear to run smoothly ?

What is airflow in the case like ?

[bytheway_r]

Unfortunately these temps are about right if your ambient temps are 25C+ or case airflow isn't all that great.

Is your CPU overclock stable? I'd check CPU at stock, GPU at lower / reference clocks and eventually in another system. Myself, I'm against SOC cards as they have too aggressive clocks. Because of that you'll have a card that's not capable of working at these factory settings every now and then.

[LanderTaker]

I am not so familiar with nVidia cards but it would seem to me to be a bit too hot for comfort at those temperatures. Have you got all 8 of the power cables connected to the card ?

Does the fan appear to run smoothly ?

What is airflow in the case like ?

Hmmm, I have been looking reviews and forums and appears to be common to reach 95ºC or 97ºC... By example http://www.tomshardware.com/reviews/geforce-gtx-480,2585-15.html says 97ºC.

Yes, the 2 power cables: One with 8 "pins" and the other with 6.

Hmmm... I have not observed the gpu fans (I have only heared them!!!) I will look them later when I arrive at home.


Ehhhmmm,,, Air is transparent

Now serius: About the airflow. (by memory):
Alluminium case. 3 years old Nzxt lexa. (have not inner lights) (and I do not recommend it, scarce of space...)
One 12" fans in the front blowing air to the inside. (I think that only one, almost sure)
One 12" fan in the openable panel, just targeting the GPU (aproximately!).
One top 8" output fan.
A 12" at the back, at the CPU height, getting out air.

[Dark Mantis]

I was impressed when I first read your reply, and not by that pathetic joke ( sounds like one of mine) but by the size of your fans. I thought they were huge then I realised that you meant cm not inches! 

[LanderTaker]

Unfortunately these temps are about right if your ambient temps are 25C+ or case airflow isn't all that great.

Is your CPU overclock stable? I'd check CPU at stock, GPU at lower / reference clocks and eventually in another system. Myself, I'm against SOC cards as they have too aggressive clocks. Because of that you'll have a card that's not capable of working at these factory settings every now and then.

I have checked the CPU like 12 hours without any anomaly. I do not remember the program used. Was not cpu prime . The program checked cpu and mem accesses.

Ok, I will try all components with stock timings this afternoom-night (2 or 3 hours from now). If all works well with stock settings I will try to determine which component is causing failures when overclocked.

((( I am guessing that it may be the CPU because before applying the OC settings to the GPU I had already detected a freeze with alien vs predator (and had not tested crysis enought to see any anomaly). )))



tyyyyy!

[LanderTaker]

I was impressed when I first read your reply, and not by that pathetic joke ( sounds like one of mine) but by the size of your fans. I thought they were huge then I realised that you meant cm not inches! 

The joke was an easy one


ouch! hahaha. Yes, cms, cms

[Dark Mantis]

Please let us know how you get on. 

[LanderTaker]

weir:

crisis 1: It appear that without oc I can play indefinitely...

after, I activated only gpu OC...

crisis1: played some time (5 minutes? maybe more) and freezed. The binary got corrupted from this point and It hangs just before the game show the saved games

Then i deactivated all oc again... (alwais turning off and on so the settings are used correctly )

crisis1 continues hanging up at load game windows before showing games.

Tried alien vs predator DX11 and... 1 minute and freezed. :/ (supposely no OC at all!!).

It is rare, I with the button activate the OC (red light) or deactivate it (blue light), but the afterburner shows the same info: core 820, shader 1641 and memory 1900.     ()

reinstalling crisis 1...

[Dark Mantis]

There seems to be a problem with some of these cards that are factory overclocked as it might be a little to aggressive. My advice would be to take the card back to stock settings instead for now until they bring out a new BIOS for it.

[bytheway_r]

Set all the values to what a reference card would run ( clocks and voltage ) and try if it runs ok. If not, increase the voltage a bit and check again.

Another option could be faulty RAM ( not VRAM ). Run memtest and see if it shows any errors.

If the card's still not working properly after this, RMA it. Eventually check with some other card to see if it's not something else dying.