Sign in with
Sign up | Sign in
Your question

Could this be a problem? lol

Last response: in Graphics & Displays
Share
July 9, 2006 12:28:12 PM

[http://homepage.ntlworld.com/daniel_hill/6800GT.JPG]

This caused me a huge drop in performance when i was playing hl2 as it reduce the clocks to prevent damage, but nothing was wrong with the fan, the ambient was fine and touching the card felt just warm.

Obviously this is a sensor problem as the card is working fine and the temp has reset itself since.

Has anyone else had any issues like this before.

i have emailed xfx tech support to see what they say.

More about : problem lol

July 9, 2006 12:54:39 PM

Yeah, either the drivers have gone south and are reporting an invalid temp or, more likely, something in the cooling solution has died (likely the fan). I'd suggest getting a separate fan (either a spare computer fan or even a floorstanding one, whatever), open the case, point the fan on the card, boot the system and see if the on-card fan is spinning.

If not, shut down the system and now you know what's going on.

If it is, something else strange is going on, open the nvidia panel and watch the temps.
July 9, 2006 1:11:28 PM

Quote:
Yeah, either the drivers have gone south and are reporting an invalid temp or, more likely, something in the cooling solution has died (likely the fan). I'd suggest getting a separate fan (either a spare computer fan or even a floorstanding one, whatever), open the case, point the fan on the card, boot the system and see if the on-card fan is spinning.

If not, shut down the system and now you know what's going on.

If it is, something else strange is going on, open the nvidia panel and watch the temps.


You didn't even look at the pic did you?
Related resources
July 9, 2006 1:13:05 PM

Quote:
[http://homepage.ntlworld.com/daniel_hill/6800GT.JPG]

This caused me a huge drop in performance when i was playing hl2 as it reduce the clocks to prevent damage, but nothing was wrong with the fan, the ambient was fine and touching the card felt just warm.

Obviously this is a sensor problem as the card is working fine and the temp has reset itself since.

Has anyone else had any issues like this before.

i have emailed xfx tech support to see what they say.



These two should look interesting for you, I know of no software way to disable the temp sensor.

http://www.mvktech.net/content/view/2069/37/1/0/

http://www.mvktech.net/index.php?option=com_remository&...
July 9, 2006 1:19:53 PM

Quote:
Yeah, either the drivers have gone south and are reporting an invalid temp or, more likely, something in the cooling solution has died (likely the fan). I'd suggest getting a separate fan (either a spare computer fan or even a floorstanding one, whatever), open the case, point the fan on the card, boot the system and see if the on-card fan is spinning.

If not, shut down the system and now you know what's going on.

If it is, something else strange is going on, open the nvidia panel and watch the temps.


You didn't even look at the pic did you?

Yeah I did, but I didn't read the rest of the quote (specifically concerning the only being warm to the touch). Sensor problem. I'd wait to see what XFX does, since they have a lifetime warranty and hacking the BIOS will void that (not to say you can't plead ignorance, but I'd just wait)
July 9, 2006 1:28:10 PM

Thanks for your response.

I did actually reinstall the drivers as i thought this was the case, but it did nothing.

Speedfan also reported something stupid like -58 deg c

like i said its back to normal now and i have unticked the option to warn when temps go to high to prevent the clocks going down.

the only problem with this is what happens if the fan does die in a few months time, the earliest i'll know about it is when the gpu melts itself.

this is obviously not ideal.
July 9, 2006 1:36:33 PM

interesting!

i'll bear that in mind should XFX lose interest.

actually perhaps you can shed some light on this issue, or maybe this bios hack could sort this

i bought the card in October 2005, installation went fine but enabling coolbits revealed the default clocks to be 350/900

a 6800 gt stock should be 350/1000!

i emailed xfx about it and they could not offer an explanation and encouraged me to use coolbits to return it to 350/1000.

i did so and have not had any problem at all.

could the bios hack be used to set the default speeds?
July 9, 2006 1:47:04 PM

Yes the NiBiTor is actually primarily meant to be used to overclock the card.
(Software OC can be iffy and tell you it doesn't work, when it does)
July 9, 2006 1:52:36 PM

Also, the tickbox in the default control panel doesn't stop it from downclocking, it just doesn't tell you it did so.

Second, your card will lockup the computer long before it has melted if the fan fails (well, assuming you're there to turn it off *cough*)
July 10, 2006 8:21:57 AM

asdasd123123 Wrote:
[Also, the tickbox in the default control panel doesn't stop it from downclocking, it just doesn't tell you it did so.

Second, your card will lockup the computer long before it has melted if the fan fails (well, assuming you're there to turn it off *cough*)]

I suppose that makes sense. Is the speed control done via software or is it chip level kinda like Intels throttling or speedstep.

Any ideas on why the default speeds are not to nvidia spec
July 10, 2006 1:55:54 PM

Quote:
asdasd123123 Wrote:
[Also, the tickbox in the default control panel doesn't stop it from downclocking, it just doesn't tell you it did so.

Second, your card will lockup the computer long before it has melted if the fan fails (well, assuming you're there to turn it off *cough*)]

I suppose that makes sense. Is the speed control done via software or is it chip level kinda like Intels throttling or speedstep.

Any ideas on why the default speeds are not to nvidia spec


I think it's hardware, probably why there's no obvious software override.
Ahh, I just figured out why your card is downclocked, it's an AGP card with an HSI right? (you can see the HSI by an extra heatsink or chip near the port)

AGP cards are downclocked for no good reason but to look bad in benchmarks!
July 10, 2006 5:22:21 PM

yes it is AGP.
i remember reading on toms when nvidia released the 6 series that they based the design on pci express but used a bridge chip for agp.

it's fine running stock 350/1000 but in your honest opinion and given that every card is different, what speeds do you think i would get out of it if i decided to overclock.

the nvidia control panel seems to think 380mhz for the core is ok, the memory says 980 but i think that is being limited by the default clock so thats probably not a true test. maybe 80 above what it should be, say 1080mhz?
are they likely to go much more than this.
July 10, 2006 5:33:37 PM

Quote:
yes it is AGP.
i remember reading on toms when nvidia released the 6 series that they based the design on pci express but used a bridge chip for agp.

it's fine running stock 350/1000 but in your honest opinion and given that every card is different, what speeds do you think i would get out of it if i decided to overclock.

the nvidia control panel seems to think 380mhz for the core is ok, the memory says 980 but i think that is being limited by the default clock so thats probably not a true test. maybe 80 above what it should be, say 1080mhz?
are they likely to go much more than this.


Well, if you have a second generation gpu, you ought to be able to get perhaps as much as 100mhz more, if first gen you'll need at least water cooling for the same.

Although 400 core seems plausible with air, assuming it's decent.

Rams are locked in entirety to what they're made to reach, if your card has 500mhz rams, they won't go more than a few mhz above, but if you're lucky, they spat on a pair of 550-600mhz modules on there, some makers do for to me unknown reasons.

Try to set rams to 1100, and save, start a game let run for a while, and if you get any form of noise you can't go higher.

The core is a bit more tricky, some errors don't become apparent until the errors are beyond reversal (you can't alt-f4 fast enough and you crash)

Try with lengthy 33mhz increment test, and see what happens. 3D Mark is fairly unstable, so it'll crash with even small errors, wich might actually be good for this purpose.
July 10, 2006 6:34:09 PM

Quote:
Yeah, either the drivers have gone south and are reporting an invalid temp or, more likely, something in the cooling solution has died (likely the fan). I'd suggest getting a separate fan (either a spare computer fan or even a floorstanding one, whatever), open the case, point the fan on the card, boot the system and see if the on-card fan is spinning.

If not, shut down the system and now you know what's going on.

If it is, something else strange is going on, open the nvidia panel and watch the temps.


You didn't even look at the pic did you?

ROFLMAO!! Yeah, I would just say your sensor has gone out. Maybe you can get them to give you a gs or something.
July 10, 2006 6:47:56 PM

256c seems a little hot. 8) Maybe you need some more cooling as your case is 44c.
July 10, 2006 7:32:04 PM

Actually 44 degrees was the ambient temp around the card, and is considerd pretty normal, the case temp was something like 30 which is way under the intel spec of 38.

i have 2 antec 120mm case fans doing a good job keeping the case cool.
one of them is controlled by the northbridge fan controller (i unplugged the active northbridge fan and plugged the 120mm in to that, so i can control the 120mm fan speed depending on apps).

http://homepage.ntlworld.com/daniel_hill/speedfan.JPG

remote = gpu
local = ambient around the card
hd0 = hdd
pmw = power management (MOSFETS under OTES)
cpu = self explanatory (C0 rev and always hot, can't overclock. abit sensor could be overestimating this though)
motherboard = case temp

the reported 256 degree gpu temp was actually a blip as speedfan and nvidia both report normal temps now. it was quite funny though!

also take a look at my case http://homepage.ntlworld.com/daniel_hill/P1010001.JPG
July 10, 2006 8:40:48 PM

i know i know, theres nothing i can really do about the wiring.
thats one bad thing about the antec psu is the way its been wired.
they are all interleaved! you cannot seperate each set of wires for each molex
still, nevermind you wouldn't want it looking like dell now would you!!!!
July 10, 2006 8:43:08 PM




WTF that card shouldnt be alive

8O 8O 8O 8O 8O 8O 8O 8O 8O 8O 8O 8O 8O 8O 8O
!