680m Throttling in game

hi guys,

I have a problem with my 680m, dell oem version, no oc no ov, the problem started today, when I'm playing any game and the temperature reaches 78 º begins Throttling.

I have not made ​​any changes, just installed a couple of games, I did system restore if it was and it did not work, I had not happened before.

Any help?


THx

m17x R3

 

Sounds like you might need a repaste. Check that the fan and heatsink are not clogged with dust bunnnies.

I guess you got the card yourself (R3 wasn't sold with the 680)? What heatsink/x-bracket did you use?

At that temp mine goes into the overheat throttle. You can use HWinfo to override it sometimes but not recommended, best to keep it cooler.

 

Hi,

yes install it myself, using the x-braket and proper heatsink.

you have the same problem at 78 degrees?

THX

 

Yes but I never let it get there, and without an OC it never will. Runs around mid 60's to high but never over 68C untill I OC it.

If you got the dell heatsink and x-bracket you should be good but I'd say to take it apart and have a look against the pics the 3rd party suppliers have for the pad placement. It really sounds like you just haven't got it right yet. maybe a misplaced pad is interfering with the core contact.

As long as it was a dell supplied heatsink for the 680 there should not be any tape that causes this problem with mix-n-match configs and it comes pre pasted and padded. The correct x-bracket is very important.

 

get a radeon gpu

 

sounds like its running a bit hot tbh. Mine doesnt go above 65 usually. The fans kick in by then. Mines completely stock (even stock thermal pads etc). Check yuor setup sounds like yu might need a repaste or the heatsink/xbracket tightening. The fans run full speedwhen it throttles?

 

Yes, fans works full.

Im going to repaste witch ic diamond 7 and put this pads

100mm 100mm 1mm Thermal Pad Heatsink Fr VGA GPU Chipset | eBay


Can it help me??

I have nvidia x-braket for upgrademonkey and correctHeat sink i think


Thx for u helps.

 

I'm sure the problem is the heatsink contact but it's can be tricky. Contact upgrademonkey for a pad placement map and ask if there different thickness's of pad (mine hat two diferent thicknesses AND some thermal putty over the VR's.

The x-bracket must be one supplied with the card. The height of the posts is specific to the core thickness and changes between models.

The heatsink does not have to be one designed for the card as the MMX standard specifies the layout of the chips and maximum thickness's - unfortunately minor differences can make it difficult to get good contact though. If the heatsink has tape around the GPU core area remove this - it will interfere with the core contact and could well be (in itself) the cause of your overheat?

 

No it's not running hot.
R4's have a special heatsink for 100W cards.
The R3 does not, hence it will get hotter.

Dell cards should not throttle at 78 degrees, that's a normal operating temp.
Throtteling should occur as a failsafe mechanism, that is triggered a 90+ degrees celcius to prevent damage to die or melting

 

Anything above 70C is hot on these cards. The 78C throttle is hard wired and drops the clock to 135 on a 680m.

It is out-of-spec at temps above 78C. See this thread where Dell arranged a conference call (that I was on) to explain why all out cards were throttling

http://forum.notebookreview.com/ali...x-580m-78c-throttling-see-post-191-later.html

Edit: thought I'd add a conclusion since it's a mega read! Basically, working as designed. Dell could raise the throttle temp but will not since that would require the whole machine to go through certification again.

If you get the 78 throttle at stock clocks then the issue is with the pads/paste/heatsink - and for many of us it was just a poor factory assembly. A replacement was made available for anyone calling in for this problem.

Many R3 owners have upgraded to a 680m, and with careful attention to the heatsink, have had no problems. The 680 runs cooler than my 675 OR the 580m I had before from what I see.

 

please don't use these pads, have you even looked at the thermal conductivity? it's rated lower then 2w....
Get some Fujipoly pads (less compressable but better then conductivity then TIM (wich is around 8~9w)

 

Strange, but doesn't this just concern the 580m and not the 680m?
My card doesn't start to throttle around 90 degrees.

My R2 can't cool the 680m fast enough (75W cooler) so it will run passed 80 almost everytime.
It does not affect performance or life span btw...just Nvidia being Nvidia

 

I didn't know if the 680 had the same 'feature' but since it is happening assumed that it is still there. Maybe machine/card combinations have a relevance??
All the 680m upgrade threads I followed say it runs around 65C-68C in the R3.
Are you using HWinfo? That can sometimes stop the 78 throttle - they even released a specific version with a startup switch to stop the 78 throttle for desktops.

 

lol, you can't compare my 680m to a stock one
nothing is stock about it, and like i said, i don't have problems with it

I am the record holder for the R2 benchmarks :thumbsup: lets just say i know what i am doing

 

Cool, but I assume your are using hwinfo to control the fans? I found that when I tried it the 78C throttle disappeared. This guy is totally stock with the clocks and still hits 78C which it shouldn't

 

ow sorry, i must have read passed that.
Yes, offcourse i am using HWinfo64 to control the GPU fans.
It has a volt/bios mod so i need to control the fans to spin up way before the designed thermal throttle design
The heat (exctrapolation) build ups faster because the voltage is higher then stock.....and it still idle's @34 degrees

But lets focus on antriach issue and not my show boating
perhaps a bios mod might fix this issue....but then offcourse there is a risk of bricking it...

 

The bios guys over in techinferno tried to find the throttle and failed. HWinfo may allow it to go hotter but the best thing (IMO) is to stop it getting there in the first place.

Done a lot of modding myself, last build, 2 x GTX 295's in quad SLI + 9600 dedicated to physx. Both 295's, processor, NB/SB and VR's all with water blocks. Running hugely out of scope voltages all over the place. Those cards run at 90C-100C always while benching even on water

 

I hear that, nothing but higher than the max

My current desktop
Your temps are insane though....only guess is that you are not covering with a FC waterblock?
I have completly worked my 800 up with a dremel.
It now holds 720mm of rad surface and running (not at the moment) 2 seperate D5 pumps.
680 + 2600K @ max (and beyond hehe)

 

Sweet!!! They were full card waterblocks designed for the 295. Had to sell it for (believe it or not) medical reasons! I got this machine to play my games at a decent rate and stay away from any voltage mods.

It got stupld, 5 diferent temperature monitoring points, fans and pump controlled by manually applied temp sensors. Only stat was that I had a 2.2 i7 benched at 5.1 and stable at 4.7. Made 57 worldwide in class on 3D mark at the time I finished it. I even lapped all the blocks. I miss that machine (but not spending 48 hours straight tinkering with it!)

Prob time to stop the thread hijack - Sorry OP

 

Thanks guys for your help, but the truth is I'm a bit lost and scared.

In normal use (internet browsing and watching videos) temperature is around 57 degrees.

can u suggest what thermal paste and thermal pads i might buy to try to improve this situation?

and where I can buy them?? I live in Spain and is a bit difficult to find.

how and where should place them thermal pads??


Thank you for your help.

 

For TIM you can use a great number of things.
Most common is Arctic Cooling mx-4.
Has a good thermal conductivy and a low thermal resistance.
This compound is also easy to spread but has a high "setting" time.
As long as the TIM is NOT electric conductive you should be fine.

Zalman has some nice thermal paste solutions but MX-4 would do the job just fine

Well you can download a default thermal pad placing template from Eurocom.
Regarding thermal pads, just look for one with the best thermal conductivity (7~12w) and that are still somewhat compresable.

I myself am looking at Fujipoli, that is rumored to be the best but is also a hard compound by nature so it'sa bit of a trade off.

Always try to find something that has something with a thickness of either 0,5 or 1,0mm so you can stackup.

Using advanced search on this forum should get you some good results regarding thermal pads.

Btw,
57degrees for just browsing seems way too hot since your igpu should do that work. It would indeed seem that your gpu is not properly set

For reference; my R2 does not have a igpu so i am using the 680m 24/7 and browsing leaves my gpu running @39 degrees celcius.

Sent from my iPhone.

 

OK, Im going to try it.

which of these two would be a good option?

Fujipoly Extreme System Builder Thermal Pad - 1/4 Sheet - 150 x 100 x 0.5 - Thermal Conductivity 11.0 W/mK - FrozenCPU.com

or

Fujipoly Extreme System Builder Thermal Pad - 1/4 Sheet - 150 x 100 x 1.0 - Thermal Conductivity 11.0 W/mK - FrozenCPU.com


THX

 

The top one will do nicely

 

thanks, I'll go for it.

another question

which is the correct x-braket for gtx 680m?


thx

 

Honest,... I have no idea. Send Eurocom an email with your specs and let them tell/sell you one.

 

Yeah, it's the height of the posts that is most important. Sets the pressure of the heatsink, so just because one fits it may be for a different thickness core and not apply enough contact. All cards should come with an x-bracket when purchased.

 

but which of the two is correct?

I need to know that I have the biggest and need to know if it is erroneous to claim upgrademonkey.

which have ye?


thx

 

The one that was supplied with mine was to high, it never even made contact with the Die

 

Hi guys, what is the max temperature u have running 3Dmark11 PERFORMANCE????

i repasted with mx-4 and i obtained 68 degrees, is still too high?


thx

 

68C IMO, is still very cool. I have never seen mine go above 66C in any performance test or hours of straight gaming.