PDA

View Full Version : ASUS GTX 580 Matrix - High temperatures



johnnyw
01-03-2012, 04:41 PM
So just got this card today and it seems to work ok but my temperatures seems to be really high compared to what other users and reviews are got :( Most reviews have got temps like low to mid 60:s at stock but im getting now like mid 70:s only after 20mins of bfbc2. Also what kind of voltage this card should have at default? Just taked a look to it with afterburner and according it stock volts are 1.088v which seems really high?

This is after 20mins of bfbc2 case closed:

http://i5.aijaa.com/b/00251/9293115.jpg (http://aijaa.com/002519293115)

And this 20misn bfbc2 with case side panel open:

http://i8.aijaa.com/b/00610/9293116.jpg (http://aijaa.com/006109293116)


I have silverstone TJ08-E case which has 180mm fan blowing cool air directly to gfx card and other components and when difference is so small case closed and open im 99% sure its definately not airflow issue. Also i live at Finland and ambien in house is pretty low, around 22 degrees atm.

http://i5.aijaa.com/b/00829/9293138.jpg (http://aijaa.com/008299293138)


Just really disapointed to spend loads of money to card like this and then have problems immideatly :(

Edit: After 1 hour of bfbc2 peak temp was 76c but mostly it stabilized to 74-75c range. Really would want to hear what kind of temps other are getting with this cards, these temps just dont seem right for cooler this size.

HiVizMan
01-03-2012, 06:45 PM
Hello mate, could you tell me what the ambient temperature is?

And could you post a GPU-z shot of your 3D clocks too.

Thank you.

johnnyw
01-03-2012, 07:30 PM
Allready mentioned in that earlier post that ambient is 22 degrees and gpu clocks are just stock 816/1002. I havent touched to any settings.

These are the settings (totally stock)

http://i5.aijaa.com/b/00917/9294682.jpg (http://aijaa.com/009179294682)

Voltage just seems really high. Just read in finnish forum that all matrix users there had voltage in range beetween 1.038-1.050v so could that be the reason that im getting these ridicilous temps?

HiVizMan
01-03-2012, 08:29 PM
Sorry mate I must have missed the ambient. Long day I guess. :)

Thank you for the screen shot. Could you confirm that your fan is scaling or does it stick on 20%?

Joni
01-03-2012, 08:37 PM
Ye fan does ramp up in load. As you see in that gpu-z pic above it was 35% = 1800rpm when temp was 74c

HiVizMan
01-03-2012, 08:52 PM
Man I had hoped it was something simple like the fan profile was not kicking in. Darn, nothing is easy yeah.

OK I will pull out my Matrix and have a play around and see if I have anything similar. Temps are a bit lower here but the delta should be the same.

Thanks mate.


update: Voltage on my card is the same as yours.

CPU Voltages (nV) 1088
OEX,VDD (mV) 1059
Memory Voltage (mV) 1580

How does that compare with your card?


My temps are 26'c idle and my max during a stress test looped was 54'c and my fan cycle did not budge from 20%.

Looking at the other temps my card was about 30'c cooler than your on the PCB.

One thing that I noticed and is maybe interesting for your problems the reason why your clocks do not seem to revert to the 2D settings? After a few seconds of not being used (stressed) my graphic card reverts to the second 3D 400MHz mode and then down clocks to 50MHz and the default 2D clocks.

Could you have a look and see how long it takes before your card down clocks to 2D settings please.

Joni
01-03-2012, 09:21 PM
I would guess it has something to do with voltage as it seems to be higher than most has. If you just bother test how your temps will be at 1088mV and auto fan.

HiVizMan
01-03-2012, 09:30 PM
See my reply I have done a looped stress test for you.

Joni
01-03-2012, 09:45 PM
My card dont get to 2D clocks at all because i have 120hz monitor, dunno what is the point but nvidia drivers dont downclock to 50Mhz when using 120hz/3d monitor. Your voltages are exatly same as mine but which stress test you looped? I just runned unigine heaven 2.5 benchmark 3x loop and in some reason that dont heat the card even close as much than bfbc2, after 3 loops of heaven 2.5 with extreme tesselation i was having 66c max when with bfbc2 im at 70c in few mins after starting game and then slowly increasing to 75c range.

Anyway i think i must of got lemon and this really sucks hard. Paid insane money off it ( could of got 570 sli with same price almost ) and now if i need to rma it probably have to wait replacement month or even more as this was last at stock :(

HiVizMan
01-03-2012, 10:14 PM
Yeah I used the same stress test as you.

I wonder how much the drivers are making a difference when playing the game. I have not installed the game onto my test bench and really don't want to. I end up playing the game and not working :D

But I am sure you have tried different drivers for the card and that you have the latest patch for the game.

Right now I am calling it a day and will see what the morning brings for inspiration. Sorry I was not able to find a fix, yet :)

Joni
01-03-2012, 10:21 PM
Ye tried 3 latest driver versions allready and dont make any difference. Anyway im off to bed but will take the card out tomorrow and see if there anything obvious like loose screws etc.

johnnyw
01-04-2012, 04:44 AM
Well taked a look to card and didint notice any loose screws or anything obvious wrong in it externally :(

Did some bit more testing and in 5mins of furmark i got toasty 88c with totally stock settings:

http://i7.aijaa.com/t/00360/9295949.t.jpg (http://aijaa.com/003609295949)

Also runned kombustor which gives around same temps (75c) than most stressfull games:

http://i2.aijaa.com/t/00253/9295950.t.jpg (http://aijaa.com/002539295950)


I start to think im out of luck with this :(

_
01-04-2012, 05:41 AM
Can I just ask: Is it stable?
Have you tried making some custom fan profiles in GPU Tweak to make a personal balance between fan noise and temps? The Platinum models are pre-overclocked, so you can't really compare to standard Matrix temps.

johnnyw
01-04-2012, 05:54 AM
Well i havent had problems so far ( cant say really yet as this is only day old )but temps are definately not normal. Also didnt even know there is "matrix" only version, or do you mean 580 DCII? I have only seen matrix platinium and my temps are far off from any reviews ive seen so far. In reviews gaming temps have been high 50:s to low 60:s and im getting mid 70:s so there is definately something strange going on specially when ambient is as low than 21-22 degrees. In last page at this thread there is guy with same voltages that my card has and hes getting like 10-15c lower temps than me.

Im not going to make custom fan curve, then biggest reason for me to buy this card was that it supposed to be very quiet and run cool at same time. First claim still stands but i wouldnt count 75c in normal gaming to be cool. Airflow issues is also counted out now, been running it with case side open and even putted extra 120mm fan to blow cool air direcly to card and it didint make more than few degrees difference to temps compared to panel closed and without extra fan.

johnnyw
01-04-2012, 06:39 AM
I think i found atleast part reason to high temps, taked the card out again and taked bit closer look and noticed that there is a lot excessive thermal paste squirting beetween gpu and cooler, too much paste wont really make any good to heat conductivity :( I guess there is nothing i could do myself to make it work "normally" when this is the case. Well i could repaste it myself but would loose warranty and thats not something im ready to do with premium produckt like this.

http://i1.aijaa.com/t/00642/9296032.t.jpg (http://aijaa.com/006429296032)

http://i9.aijaa.com/t/00358/9296033.t.jpg (http://aijaa.com/003589296033)

_
01-04-2012, 08:07 AM
Wait.. there's something strange going on here.

Your card is listed as a Matrix 580 Platinum in GPU Tweak, yet the card you show above is using a stock DirectCU II cooler?? The Matrix has an LED logo covering most of the side. Could you take a picture of the whole card and also tell me what GPU-Z says it is?

I think someone has taken a Matrix Platinum BIOS and put it on a stock DirectCUII card. This is why you're seeing abnormally high temperatures - true Matrix have hand-picked GTX 580 chips and must more VRM hardware, so its temperatures will be lower.

Where did you buy it?! I need to pass this info onto our Nordic team if it's from a retail shop.

Look at this review: http://www.guru3d.com/article/asus-gtx-580-matrix-review/ see the cooler difference.

johnnyw
01-04-2012, 08:11 AM
It has matrix logo on it as seen in pic at opening post and its definately matrix platinium all the way.. Those pics above is just taked really close to show how much there is that excessive paste. This card is some true lemon, with fan at 100% only passes 3Dmark11 with 890Mhz with stock 1088mV, auto fan only at 880MHz. Most of these cards gets 900Mhz+ with stock voltages at 1038-1050mV :( Surely last time i will buy any premium produckt as there dont seem to be any premium at performance on it compared to vanilla cards. I Think ill just have to start consider RMA as those temps really are far from normal, done bit more research over net and im getting easily 15 degrees higher temps than 99% of owners.

Anyway here is another pic.

http://i10.aijaa.com/b/00505/9296190.jpg (http://aijaa.com/005059296190)

cravinmild
01-05-2012, 01:13 AM
Ive also seen the reviews and my temps on my card were more reflective of what your getting and the same as another that i know who has a matrix. i didnt sweat 80c at all. Id get 87c and not batt an eye at that. The thing is to raise the voltage and oc that card then see your temps. Im going to guess that once you hit 1.150 with 980-1000 core your temps will not be much different than the highs your getting now, perhaps youll hit 90c, still ok temp for a card like this. My card does not sing untill i hit 1.20v...need it to stay stable for benchmarks above a 1000 core. That said, i dont keep my card oc, just for quick runs. Other than that it runs stock.

1.088 is not high voltage, your card will do 1.293-ish...thats high. Shoot for 1.1v and check your temp at idle. It will spike for a second an come back down. If your around 65c start to benchmark it, go for 900core then 930core than 950core. if youve not hit 90c then keep oc the card till its no longer stable. Raise the voltage again to 1.12v if your not stable at 930-950core. That should be a good core speed and if your not hitting 90c then be happy with your oc.

Like you said, you bought a high end premium expensive card so dont be afraid to use it. If it blows then rma, im sure ROG will be understanding, most companys will rma still if youve oc your card...hell they include the oc utility with the card.

_
01-05-2012, 01:31 AM
It has matrix logo on it as seen in pic at opening post and its definately matrix platinium all the way.. Those pics above is just taked really close to show how much there is that excessive paste. This card is some true lemon, with fan at 100% only passes 3Dmark11 with 890Mhz with stock 1088mV, auto fan only at 880MHz. Most of these cards gets 900Mhz+ with stock voltages at 1038-1050mV :( Surely last time i will buy any premium produckt as there dont seem to be any premium at performance on it compared to vanilla cards. I Think ill just have to start consider RMA as those temps really are far from normal, done bit more research over net and im getting easily 15 degrees higher temps than 99% of owners.

Anyway here is another pic.

http://i10.aijaa.com/b/00505/9296190.jpg (http://aijaa.com/005059296190)

Sorry I didnt see the first page. That first picture is on the INSIDE - those slots are the same on the outside of the stock DirectCU II card so I didn't see the logo. The unfortunate situation is that it's stable and working perfectly at its rated specs and no one can guarantee OCing beyond the spec they sell for from batch to batch. I would contact where you bought it from to try and arrange an exchange if you're still unhappy. :)

Cravinmild - In the first year it's mostly not down to us. Different retailers have different policies in those terms and the GTX 580 Matrix's are no longer being made afaik due to limited GTX 580 cores available.

cravinmild
01-05-2012, 03:56 AM
Yes you are right Marshall,

@Johnny i would suggest that you start calling the retailer and get a log going with them regarding the heat issue. Be colorfull if you must. Once youve gotten three/four calls in (with the same person-one with athority to switch the card out) then push that card hard (dont tell them that) and see what happens. If she blows then it was going to anyways at some point in the near future, return the card for your money back or a different venders card. As Marshall has stated, there are no more Matrix cards left so getting another now or in 6 months from now is never going to be an option. If i was you i would want to have a fully working card sooner than later...like after the 3 year "times up" period. Protect your own interests first.

Really sucks if it is a defective card as the Matrix truly is an amazing card and worth of every cent. Its a shame you have not gotten the true expirence. Im not going to push you to do this but if heat is your only issue you could consiter watercooling the card, ive modded mine to a H70 for amazing temps but you seem to feel that the cost for the card you have is not worth it as it is right now so spending another $60 on watercooling may not be your thing but its been mentioned just the same.

johnnyw
01-05-2012, 05:32 AM
I allready talked to retailer yesterday and seems like they dont want to exchange it as its working at stock settings, so ye its just as Marshall stated.

Just feel disapointed, even bit cheated after reading all those reviews with low temps and now getting something totally different myself. Its not only the heat but also when temps are higher fans has to spin faster and obviously creates more noise too.

cravinmild
01-05-2012, 02:07 PM
Well there is always this
http://www.overclock.net/t/1156145/modifying-heat-spreader-of-gtx-560-guide

but im pretty sure rma later may not be an option lol

What you think Marshall, can he return the card if there is a pcb issue, card is still fine after the mod, just has a loose IHS.

I did mine but it was just for kicks, temps were not an issue for me (same as yours acutually) but unlike yours my card oc like a mofo. The op was forced to do this mod as he was having the same issue as your are. Read the whole thing, its very relevent to you.

_
01-05-2012, 03:00 PM
+rep Cravin!

Jonny. Unfortunately it's the roll of the dice on any mass produced product. No one guarantees overclocking and only a temperature range to keep it stable at its sold temps. I know it's an OC card but is it slow in your games? Surely that's the most important thing :) OCing another 50MHz will give you about 3-4FPS tops: benchmarking is important, but gaming not so much.

Since you want to OC it more and you cannot RMA it, why not start modifying it? Clean up the thermal paste, apply something different, check the contact with the core. I wouldn't go to the extent of razer blading as that risk is v.high but you can do some basic TLC :)

johnnyw
01-05-2012, 04:07 PM
Thing is that removing cooler and reapplying paste voids warranty and when i just spend 500€ to this im not really ready to take the gamble that it blews up after few months and then end up having bricked card that i cant even get to warranty anymore. I always thought these matrix series cards are really more like limited range (when DCII models are mass produced ones ), atleast 5870 matrix version was.

Also now when temps are this high arguments in packaging that claims that it runs 20% cooler than reference card dont really stand anymore. I think ill just keep on fighting with retailer as long than possible and try to get this replaced, its just not right imo that you buy expensive enthusiast level produckt and then it dont really function as supposed. I have fairly good airflow in case with that huge fan blowinf cool air directlyto card so id say in "normal case" with few 120mm fans temps would be even higher.

And actually overcloking with these 5xx series cards does give much bigger boost than few fps, like with GTX570 i had before with 60Mhz OC over stock setting i got like 10-12fps boost in bfbc2 and in some older games difference is even bigger. Maybe in some really extensive games like metro 2033 & really high AA and AF that kind of oc only gives few fps, but then its kinda different story when fps is low anwyay so when fps is like 28 or 32 there is huge difference in gameplay smoothness.

Anyway i think i will buy some new bigger case next week with even better airflow/sidefan and see how it affects to temps.

johnnyw
01-07-2012, 02:09 PM
Hmm ok now this is starting to get really strange, installed latest beta drivers (290.53) today and now in some reason my stock voltage is 1075mV instead of 1088mV that i had with 285.27,285.62 & 290.36 drivers. I thought voltages are set to vbios and drivers dont have anything to do about those apart from switching in different perormance modes (2D/3D)?

HiVizMan
01-07-2012, 02:32 PM
Let me download the drivers and have a look see.

Sputnikmk2
01-09-2012, 05:40 AM
Default for mine at 816mhz is 1038mV with the 2.0.2.6 Gpu Tweak. Could it be a per card basis as these are overclocked from factory

johnnyw
01-09-2012, 06:05 AM
Ye i know voltage for all nvidia 5xx series cards differs per gpu die, but the weird thing is that mine had 1088mV with other drivers, but then installed these newest betas and now in some reason voltage is 1075mV

johnnyw
01-24-2012, 06:50 PM
Ok i still dont get this, how drivers can affect to voltage that is applied to card? With these latest nvidia beta drivers (290.53) all programs including asus gpu tweak reports my voltage as 1075mV but with older drivers like whql 285.62 voltage is 1088mV.

Very strange!

johnnyw
01-28-2012, 08:28 AM
Changed case few days ago and it made huge difference to my temps, almost cant believe it. With that silverstone matx case i was getting mid to high 70:s in long gaming, now with this CM storm trooper im getting like high 50:s to low 60:s.

HiVizMan
01-28-2012, 12:19 PM
Yeah my friend air flow patters are a big deal in cooling. Most people just do not understand how important the thermal design aspect behind a case is. Even if you were to compare temps with a side panel open a well designed case that is closed will out perform a not so good one.

That is a huge and I am pleased you are a little happier.

Rockford
02-16-2012, 12:51 AM
The card does not hurt at all by that is projected to stay at 80c for several hours, however, it takes significant (relatively) injury of the constant commute between 50c and 80c during the same time interval.

The level of damage to electronics are added after which the temperature of the circuits can handle before being physically harmed. The injury itself is usually caused by thermal expansion, or rather to the different materials in various circles expand and basically rips component. A typical example here is a capacitor in which the content expands more than the metal casing at the same temperature, causing the capacitor to the final deformation (swelling) to explode in the extreme case where the casing can no longer stick together lumen.

The same applies of course ICs. Metals and plastics are expanding greatly. Most good plastic expands 4-5 times more than copper, poor as much as 18-20 times more, while the silicon expands a quarter as much as copper. Although silicon is expanding, however, also places an upper bound on the size of a GPU can become before it harms itself.

One must also distinguish between fast warm-ups / fast cool-downs from the slow ones, when a slow temperature change of course, implies a slow (read less harmful) expansion. Here, the cooler itself less important than the heat sink when the flange slow down temperature increases in the circuit. Small flange and large fan is the most harmful of all.

What hurt the electronics are not heat itself (as long as you stay below the injury level), it is repeated rapid temperature fluctuations that lead to expansion and reduction of circuits, solder and PCB - as they move relative to each other when they get hot or cooled. It required not to come up in impressive temperatures to measure this, a 60-degree temperature difference (25-85c is not exactly unusual) gives an expansion of the video card PCB (by 236 mm) with just over 0.3 mm.

That may not sound much, but considering the amount of heat solutions or cooling a graphics card is going through in their lives, and given that the precision of the solder is higher than said 0,3 mm ... well, you now understand why graphics cards hopefully eventually "break of themselves."

Too aggressive cooling, thus damaging the cards more than anything else, and that's the main reason that the cards are usually asked to stay on an even, rather than low temperature.

Or something like that.......;)