Safe gpu vrm temperature

Superlp12

Distinguished
Feb 18, 2013
261
0
18,810
Is there a safe temperature limit for vrm temperature of a gpu? My current over clocked 280x's vrm temperature hit 81C when the gpu core temperature reach 62C. I have have bump the core voltage to 1.3v from the default 1.256 and the vram voltage from the default 1.5v to 1.588v.
 

gernstsmit

Distinguished
Feb 16, 2010
36
4
18,535
Hi there guys, I also have a question regarding VRM temperatures.

I recently bought an Asus R9-290X OC card. When I started running it with GTA V I noticed that I got some stutter / dropped frames while driving in game after playing for a while. I checked the GPU and VRM temperatures while playing and noticed that the VRM spikes up to 110 °C. I found this quite alarming and proceeded to run a stress test with Furmark. During the stress test the VRM temperature started climbing rapidly and I had to stop the stress test manually when the VRM reached 120°C for fear of damaging the card.

I sent the card back to the supplier and asked them to check it out. They let me know that other cards of the same model showed the same tendency and thus there was no factory fault with the card. I read online that this specific card's cooler was actually designed for a different GPU and copied, as is, to this card. Therefore the cooling issues. Apparently the Saphire R9-290X cards have a better designed cooler, Unfortunately I did not know this before buying the card. I also read that, at maximum performance, the GPU draws 420W of power. That was quite a shock to me and explained the high temperatures somewhat.

My case is a Coolermaster Cosmos SE. That is a full tower (some describe it as somewhere between a mid an full tower in size) and it has, in my opinion very good cooling. I have 2 x 120mm fans in front, as well as 2 at the top and one at the back. So I didn't think anything was wrong with the case layout.

I thought about many ways to address the problem including modifying the GPU cooler with copper heat spreaders between the components and the cooler and also considered water cooling. My problem with water cooling is that, as far as I can see, the water coolers only cool the GPU itself and not the VRM. Thus it would not help.

I bought a "spot" cooling fan and tested various locations around the card, in the case, to find where it would have the greatest effect. In the end I found the best location to be right above the card, blowing over the "rear" as well as the "front" of the card. Then I took my windowed side panel out and removed the plastic part of the panel. I sent it to a "waterjet cutting" company to cut a profile for a 140mm fan in the window part.

Now I have a 140mm fan in my side panel that blows fresh air directly onto the graphics card from outside the case. This seems to work well enough.

During game play my VRM rarely exceed 90°C in temperature and during a stress test it hovers just under 120°C maximum. Also the problem of stutter in my gaming has all but disappeared.

I would like to know if anyone else has had a similar experience and how they have resolved it. Is there a better solution.

Note that the GPU itself never went as high in temperature and as a rule of thumb stays around 10 to 15°C cooler than the VRM. Also I have two VRM temperature sensors on the card and only the one sensor reported these high temperatures. The other VRM sensor reports values lower than the GPU temperature.
 

Mike_149

Reputable
Jan 17, 2016
4
0
4,510
Are you running the r9 290x matrix ? If so this card and the OC card rub very very high all round made for extreme overclocker with watercooling and ln2 I'm running 290x matrix on watercooling and the block does cover the vrm I never go above 70 and I've overclocked mine to hell !!! ( suck it titan uses) haha it get expensive because u need custom loop with the block from Italy costs around 250 quid to watercool the card then 100 per card after that it's worth doing. People complain on reviews about the card when it's stock and never watercooling it but once it is water-cooled OMG overclock levels are insane better than gtx 970 all day long and the titan you have to know what your doing with these cards

Thanks mike
 

gernstsmit

Distinguished
Feb 16, 2010
36
4
18,535
After digging further and taking the cooler assembly apart, thus voiding the warrenty on the card :-( , I found the source of the problem. The problem was actually with the design of the VRM heatsink that sits below the "cooler". This was very flimsy and poorly secured. As a direct result the heatsink did not touch some of the VRM components at all, so they had no cooling whatsoever. Having some experience in mechanical design helped me to come up wit a solution. I had 2 aluminium spacers lasercut the fits under the heatsink where it is secured to the card. This did two things: 1) It lowered the entire heatsink by a few tenths of a millimeter 2) It helped "level" out the heatsink and prevent it from bending. Also I replaced the gap pad that came with the card with a better quality, beneath the heatsink. This was the final solution. My VRM now has full contact with the heat spreader and the VRM temperature now stays equal or below that of the GPU.

http://tfd.atmailcloud.com/index.php/mail/shared/dl/f/5b0b1c64564abc5abaef2f1907605dc9/

d8b01819d87082fe8e5b67eaecc12431

http://tfd.atmailcloud.com/index.php/mail/shared/dl/f/2ba690a037cdba71acf064114323beab/