Installed a waterblock (NZXT G10 + H55) on to AMD R9 290, worked fine for an hour and then crashed, now it has heat problems

oppoman

Reputable
Jan 10, 2015
6
0
4,510
Note: I have been monitoring the VRM temps and even after the crash they haven't gone over 85. I do not believe that not yet having VRM heatsinks is the problem.
SPECS:
AMD R9 290 REFERENCE with H55 and G10 bracket
FX 6300 with EVO 212+
8GB RAM
Asrock 970 Extreme3
Windows 8.1
Corsair 300R
I've had two crashes so far, the first one was caused by me wanting to test out the game Distance with multimonitors but then either when it was loading or had just loaded my screens suddenly went black and I had to do a hard reset. Second time happened after I was able to successfully recreate it with the same game (I don't want to try a third time due to being scared I may be harming my system by doing so).
I'm not sure whether it has to do with how my heat is circulating properly , whether or not I used enough/too much thermal paste, or if my G10 isn't screwed in enough. Also when my case is right-side-up the temperature quickly increases, but when it is on its side it will appear to stay at a lower temperature (until I do anything taxing, then it will rise in temperature).
For the hour of it seemingly working fine the max temps for the GPU was around mid 60s after a steady incline. Now if i play a game or benchmark it will shoot up to 60 or 70 depending on which one and rise another 10 and apparently have the possibility of instantly crashing.
 
The H55 (120mm @ 1700 rpm) is way undersized for a GPU pushing 300 watts

http://martinsliquidlab.org/2012/05/01/alphacool-nexxxos-st30-360-radiator/4/

The Alphacool ST 30 series is the closest radiator equivalent to the H55. We see here that at max speed a 120mm x 360mm copper rad (the H55 is aluminum and 120 x 120) can take care of 242 watts....to account for that rad being 3 x 120 to the H555's 1 x 120, we can expect say 81 watts of cooling @ 1800 rpm....lil less for your 1700 rpm ... lets say 76 watts.

Now your card when overclocked gets up to near 300 watt territory .... about 60% of that as a minimum (rest is radiated thru card, backplate, tubing other surfaces) has to be handled by the rad or 180 watts..... so you have 76 watts of cooling and a cooling requirement of 180 watts. Two H55's would give you 152 watts..... if it was 140mm, ya get about 35% more cooling or about 200 watts.

I have twin 780s and overclocked the begeezes outta them to the point where they each draw almost as much power as an overclocked 290. I have 4 x 140mm of radiator on them and they run at about 39C w/ 1200 rpm fans.....44C if I keep fan speed below 850 rpm.

So what can we take away from all this.

1. The R9 series is already very aggressively clocked in the box. I have gotten 7 - 12% OC's on the 290 / 290X versus 25- 28% on nVidia 780 cards. So first thing I'd say is don't be too aggressive. Heat issues aside there are limits to how far ya can overclock these things cause AMD went and did a lot of the overclocking before it was put in the box.

2. You have about half the radiator ya need to cool a 300 watt GFX card to typical water loop temps. 120 x 120 CLC water coolers typically give comparable temps to air coolers.

3. If ya wanna test and compare with others, you'll need to run a standard benchmarking tool. If ya wanna see the biggest load possible on your GPU, run Furmark. Then you will be in a position to compare w/ results which other users have reported.
 

oppoman

Reputable
Jan 10, 2015
6
0
4,510
Even after I turned off my overclock after the crashes, my temperatures now boost up much higher than they were in the beginning and compared to other benchmarks I've seen from similar setups (don't know where, but I remember seeing others with H55s and H50s getting better temps than I currently am). maybe like this one where he has a H50 and is only getting 55-65 C while I could randomly blow upt o 80 C or instantly crash
 

oppoman

Reputable
Jan 10, 2015
6
0
4,510
I redid EVERYTHING in the process of putting it back together, and now it won't even boot correctly! I used 70% alcohol with qtips to wipe off more of some thermal paste and applied some new thermal and it just won't work. I have absolutely NO IDEA how to fix it now. About every 1/5 the time it looks like it is going to boot, it has lines on it and will go black after the windows loading screen. I put in my 270X just to verify it was my 290 causing this and my 270x worked fine.

I blew into the power inputs, i blew into the pci-e lane, I picked different pci-e cables to go into it, I switched around the fan cables, uninstalled drivers while in 270x to see if some weird driver incompatibility, reinstalled them midway through while having the lines all over (this was the only time out of two hours I was able to get past the windows loading) only for it to go black for good making me have to do hard reset. I used DDU to uninstall drivers but now it says that certain AMD folders are corrupt and I cannot delete them myself.