GPU heating up? No indicated temp above 40c

Thanks for posting. I run water cooling and the water is cooled down to 30 to 40 F. The indicated temps are always below 40c. But there are areas on the cards that do not have cooling. I read that the area that controls the voltage can get hot and that is on the back side of the card. I just ordered some heat sinks for that area. For all three cards.

My temps for card (1) is 34C, Card (2) 40C, and card (3) is 32C.

Sure, I will download some other stress tests and report back. I will get back with that information on Sunday.

I forgot to ask 1LiquidPC and Davidarad02. Any particular tests / software you would like me to test with?

Hi 1liquidPC,

Thanks for posting. I will run some other stress test and report back on Sunday. I know I already mentioned with another reply to Davidarad02 but "I read that the area that controls the voltage can get hot and that is on the back side of the card. I just ordered some heat sinks for that area. For all three cards."

P.S. to 1LiquidPC and Davidarad02,

I found my gpu in slot #1, on the back side of the card where the voltage control takes place gets hot while playing BF4. I will take a optic temperature reading of that area on Monday. I do not have the tool to read the temp at home right now. The other two cards don't seem to have the same area getting nearly as hot. The card inn slot one is in a pci-16 slot. The second card also is in a pci-e 16 slot, this card is second in the race for heat on the back side of the card. The card that is running pci-e 8 slot is barely warm. That is all on the back side of the cards where the voltage is controlled.

I forgot to ask 1LiquidPC and Davidarad02. Any particular tests / software you would like me to test with?

Thanks for the complement on my rig 1LiquidPC. I just built with in the month, about.

Ok, I found myself with nothing to do so I did some bench marks. These are the free versions of the software.

Performance Test 8: 6918.2

3d Mark 1.4.828
* Fire Strike: 22924
* Skydiver: 46344
* Cloud Gate: 33893
* Ice Storm: 215963
*Fire Strike score compare/ Better than 99% (The software just has this listed).
This program was only using one of my monitors though. It looks like I need to buy the program to have all three monitors being used for the test.

During the tests my gpu temps really did not bat an eye. GPU1: 35C, GPU2 41C, and GPU3 35C.

GPU settings:
Core voltage: +63
Power Limit: +50%
Core Clock: 1150
Memore Clock: 1400
Aux Voltage: +94

UPDATE: I just ran OCCT 4.4.1 and I have some errors. One sec and let me up the voltage.

New settings:

GPU settings:
Core voltage: +81
Power Limit: +50%
Core Clock: 1150
Memore Clock: 1400
Aux Voltage: +94

3d Mark 1.4.828
* Fire Strike: 22654
* Skydiver: 46085
* Cloud Gate: 33908
* Ice Storm: 216261

GPU temps:
GPU1: 30C max
GPU2: 40C max
GPU3: 31C max

OCCT 4.4.1
Short run but with no errors.
I will run it for an hr. or more tonight.

The cards get an indicated temp higher when I play BF4.

Is there any other tests you would like me to run. Any good stress tests you would think I should run to get the cards warmed up like BF4 or better?

Thanks. Kombustor will not run...

Unigine Heaven Benchmark works. I will run this for an extended period of time and report back with the results.

After about two hrs. of running:
Waiting for results.

Scores:
Render: Direct 3D11
Preset: Custom
Qulity: Ultra
Tessellation: Extreme
Full screen: 5760 X 1080


GPU: Stock
FPS: 63.4
Score: 1598
Observation: I was not running MSI Afterburner to monitor the cards.

GPU: 1130
FPS: 68.4
Score: 1723
All three GPU cards held there clock speeds but they were not all used 100%. The use went up and down but not as bad as the GPU 1150.

GPU: 1150
FPS: 69.9
Score: 1762
All three GPU cards held there clock speeds but they were not all used 100% All the same but the use went up and down during the testing more than GPU 1130.

GPU: 1160
FPS: 70
Score: 1770
All three GPU cards held there clock speeds but they were not all used 100% All the same but the use went up and down during the testing more than GPU 1130 and 1150.

Maybe the reason why the GPU % use is not used all the way because the test is just using them as much as it needs to be. Any thoughts on this situation?


Note: I found out I need to set the MSI Afterburner settings first then go into AMD Catalyst Control Center and turn on the "Enable Graphics OverDrive" and then set the Power limit settings to 50% for all three cards. Other wise, if I do not. When I go back into MSI Afterburner and look at my power settings they are back to 0%. If I do adjust the OverDrive then when I go back into the MSI Afterburner the power is still at 50%.

Observation: I heard a squeal when the tests were under full stress. I located the squeal inside my power supply. The fan for the power supply is barely moving any air.

Thought: A new MSI Afterburner is out so I upgraded to it. Using this software I noticed that my GPU voltage is varying allot, to my opinion. When I am not stressing the cards the GPU voltage is 1266mv. But when under load the voltage is jumping as low as 1172 mv. It does not stay at a set voltage. It is jumping between 1230mv and 1172 mv. I wonder if I am having a power supply problem.

The back side of my Gpu slot 1 card back side is reaching 135F. No really hot to my understanding. Since I have the heat sinks coming already I will install them.

Trying on Tuesday or Wednesday: install the heat sinks for those areas. Install a fan to force allot more air through the power supply. Finally report back with my results.

Could to high of voltage mess things up? Besides over heating. As long as I keep the temperatures down. During the stress test and bench mark testing the higher the voltage the more stable it gets. But bf4 so far anything above +44 it gets ghosts (center monitor only) normally after a short period of time and if I really crank up the voltage then I immediately get ghosts (center monitor only) and sometimes the center monitor goes black.

About the graphics cards cooling blocks (Koolance, VID-AR290X Water Block, Rev.1.1 (AMD Radeon R9 290, 290X). I really don't know. They do cover the complete front side of the card. But I am not sure if they even touch the portions where the VRM locations are. I measured with an optical sensor the back side of the card where the VRM locations are. It was reacing 135F. I believe the VRM's are on the back side, aren't the? Not really sure.

I did not know GPU-Z has that ability. This is great news. I will run that test tonight. Yesterday I was going to post my results after installing the extra heat sinks. Well, they just arrived today and missing parts. Bad supplier.

Thanks for this information. I am anxious to go test.

I just checked my order history with Koolance. It does say it is a full face water block. Does it cool the VRM section? I still do not know. It does not say anything about that.

I totally believe you. I just looked at the foot print of the contact points and sure enough. It does touch the VRMS locations.

I am going to run GPU-Z as suggested tonight to find out what it has to say about the VRMs temp.

It is great to have so much help.

Thank you.

It is a mess right now but I took a quick snap shot and uploaded a quick shot I just took. The large white tubes are my manifolds for the input/output for car radiator fluid. This id. of the manifolds are 1.5" dia. The black pipes are the 1" id. supply and return lines. I have a car radiator outside in the garage with a box fax blowing through it. Right now it is below freezing so I use that to chill my radiator fluid. I have a 5 gallon reservoir tank out side as well. The fan pressed against the power supply is me, right at this moment forcing more air through it, just to see if there is an effect.

The highest reading out of all the VRM's from all the cards was 50C. The others are around 40C



A couple people suggested that I might not have enough power for the computer with this power supply. They feel I am running about 1250 to 1300. Maybe I should buy a 1500 watt power supply. I have read until my eyes are bleeding and so far I really have not learned enough to choose which power supply I should get. There is allot to this, I never new.

Good thought, but sadly I did. Every once in awhile I will pull the water blocks off and take them apart to clean. That was on the computer set up before this one. This one I did check the water blocks before installing them but it has only been a short while being used. The computer I came from is in the photo I use for my icon. I had everything cooled including everything on the motherboard. I still have all the water blocks for that but so far the heat sinks on this new board are maybe getting a little warm. I am sure the water pressure is high because I am using a high pressure, high flow 800+ gph, pump you would use to aerate a pond. It makes my garden hose water pressure feel like a drinking fountain pressure. If I take the return hose going to the reservoir and lift if up it is like looking at three garden hoses blasting water on full. When I squeeze the input side hose to the water blocks they are like rock solid when the pump is on and soft when it is not.

After doing all the tests that were recomended and never finding any heat problems. I would have to agree.

"GPU heating up?" Nope. I have used all the stress tests, watched the indicated temps, and measured any physical surface area of all three cards with an optical temp meter and found no indication of high temps.

One or more of the cards might not be able to handle the extra voltage but I have not proved that one yet. The question for this post was about the temp. Thank you for helping me prove to myself that the temp is not the problem but maybe the power source is. Or maybe not and I am maxed out on my system. I have another post about a power supply question:
http://www.tomshardware.com/answers/id-2496581/power-supply-system.html

Thank you and everyone for their help.

Humm, Thanks for the advice. I was leaning towards the EVGA 1600 G2. Because the price is $110 less expensive and it looks like it has more than enough power to prevent any limitations my computer might ask of the power supply. If I read correctly it has 5Amps more on the 12v rail than the Corsair AX1500i. I really do like the ability "while monitoring power draw in real-time using the Corsair-Link software" It would be great now and builds I make when this one gets outdated. I believe the Corsair AX1500i has two 12v rails, correct? The EVGA 1600 G2 has one, I believe. If I have two rails don't I need to be concerned which locations I plug the cards into? to avoid plugging all three cards on the same rail?

I just read the EVGA 1600 G2 has a 10 yr. warranty and the Corsair AX 1500i has a 6 year warranty. Really both are way beyond what I need and really have no influence for which of the two amps to buy.

The EVGA 11600 G@ has 1600 watts and the Corsair AX 1500i has 1500 watts. That is leaning on me. But I really believe you guys and I don't think I need or will ever use the 1600 watts and 1500 watts should be enough for now and the long term. What do you think.

Thanks for you opinion and advice. It is greatly appreciated.

api's ? I will have to find out what that means first.

Just in case if that is referencing Mantle or Direct x than yes. When I try Mantle I get an instant crash even with no overclocking with the cpu, gpu's, and ram. Nothing else to overclock except the clock but I did that once and it worked. I went back to play with that again and that never worked again. I am stuck at 100.

When I was running my last mother board with two R9 290x cards I had no problem running Mantle.
Mother board: Asus Rampage 3 Formula
Ram: Corsair Dominator 6gb
Power supply: Corsair 850
CPU: i7 950

I still have the parts here. Anyone want to buy them? (plug) Except the power supply.

Well, it was a good question and who knows. It might have been a problem I did not try. Thanks for asking though.

api's ? I will have to find out what that means first.

Just in case if that is referencing Mantle or Direct x than yes. When I try Mantle I get an instant crash even with no overclocking with the cpu, gpu's, and ram. Nothing else to overclock except the clock but I did that once and it worked. I went back to play with that again and that never worked again. I am stuck at 100.

When I was running my last mother board with two R9 290x cards I had no problem running Mantle.
Mother board: Asus Rampage 3 Formula
Ram: Corsair Dominator 6gb
Power supply: Corsair 850
CPU: i7 950

I still have the parts here. Anyone want to buy them? (plug) Except the power supply.