PC Gamers....

[roliath]

Were those temps in furmark with the side panel off or on? What do your Vreg temps look like? (880mhz?)
Also don't go by those temps in Furmark, ATI throttles in furmark. Ever since the 4870's.

[Lason]

Those temps are with the side panel on although I dont see much change with it off either. How can I check the Vreg temps? Also those temps are not furmark's, they are GPU-Z, I setup 4 of them to measure each chip and check the max temp after a run.

Right now I have the top card running 940/1400 clocks and the bottom card running 950/1400. Any higher and I cant finish benchmarks. The temps have never been over 71.5* during any benchmark. Top card usually runs about 10* hotter.

[Lason]

ok nevermind, never noticed the Vreg temps on GPU-Z. Here ya go, ran furmark benchmark at 1920x1080 and then took another snip after running Heaven DX11 benchmark at 1920x1080 which I got 188 FPS average.

[Lason]

Are the Vreg temps the memory modules? That 93* scared me so I turned them all down to 1300 and ran 3dMark11 since Heaven11 was acting up and wouldnt run the cards at over 40% load for some reason. That same GPU ran at 83* max now. Score was 15890 btw.

[Lason]

Oh and my windows score was a 7.7 then I overclocked the 2700k to 4.9@1.35v and set my ram back to 1866mhz with the right clock timings and OC'd the GPU's and got my score up to a 7.8 with everything maxing out at 7.9 except calculations per second. Maybe try going higher with the CPU clock?

[roliath]

The vreg temps come from this here (circled), on your card they will be much higher than single gpu cards due to have two banks of them (for both onboard gpu's).



93c is hot! But they are rated for much much more. These temps are high even on watercooled setups depending on the block used.

Since the cards are already oc'd powertune is most likely disabled or the +20% switch is added so they won't throttle down during Furmark because at stock speeds and volts they do. Its built in as a failsafe measure from AMD. Also you might not pick up anything extra with the memory oc, try to run them with mem at stock and oc'd and see how big a diff there is.

I wouldn't bother about the WEI score to be honest, I had an array faster than 1GB read/write and it listed it as 7.8 (with no other drives in the system at the time).

Have you beat the BF3 SP campaign? Or have you passed the Jet level? Alot of people have been using that as the BF3 timedemo (I think PC Gamer came up with the idea) of starting the benchmark with the AI says "NOW" and stopping it after you've put your helmet on and looked up. Use fraps and record min/max/avg. I'd love to see what you get on all three screens.

You get Skyrim?

[Lason]

You got me scared about that 93*, I think Im going to pull that card apart and check the thermal pads on those and see how good of contact they are making. I hear alot of the water cooling guys also having issues and it turns out to be not making enough contact on the thermal pads alot of times. The Haf X case has that huge fan blowing right on those heat sinks so I think they should be cooler than that.

I havent gotten that far on BF3 but Ill try to get there tonight and get a fraps vid up. No, I did not get skyrim.

[roliath]

Do a search and see what others are getting, 93* is on the hot side. But its not uncommon since those aftermarket heatsinks don't generally make great contact with those. Just don't apply thermal paste to them, use thermal tape. Ideally, you should be in the 70's.

Nice clock on the CPU btw, what are the load temps? Shouldn't be that high since you're at 1.35vcore. The same heatsink keeps my hexacore to around 50c load and idle is usually around 25~30c.

[Lason]

temp on the cpu is 65-70* after 15 minutes of prime95 per realtemp.

Using IC Diamond on the CPU and all the GPU's.

[roliath]

temp on the cpu is 65-70* after 15 minutes of prime95 per realtemp.

Using IC Diamond on the CPU and all the GPU's.

A little hot, but not bad. Those higher temps might be due to the gpu's heatsink tossing the heat back in the case, but they are within line. I remember when I first seated mine, they were at like 90c. I had a terrible job seating it since it was riding up right on the vreg heatsink which I had to modify to get it to sit properly.

[Lason]

I went back to that kitguru review to see what his Vreg's were showing and his were 100*+!!! This screen shot shows them at 110.5*!

http://www.kitguru.net/components/graphic-cards/zardon/arctic-cooling-accelero-twin-turbo-6990-cooler-review/

Im still going to pull mine apart and see if I can get better contact, I may just be stuck because these are the Vreg's in the top card in the back by the mobo so it makes sense they are the hottest.

[roliath]

Yeah you see his mem is at stock clocks, you may actually be losing fps with your mem that high. Vreg shutdown temp is at 115c, so keep that in mind for the "wall."

[CJ]

93C is crazy.. I use to run my GTX OC'ed with water and I had a 14C delta, from 45C to 59C. I think pulling that sink off and lapping both surfaces and using a tiny bit of arctic silver might pay off.

[Lason]

yeah Im going to pull it apart tonight and see how good the impressions are on the thermal pads. I just hope I have enough IC Diamond to repaste.

If its not contacting very well I may try stacking the thermal pads from the stock cooler on the ones on there now. I read about someone doing that on a watercooling setup and it helped alot.

[Tx Redneck]

Might I suggest making some shims from copper sheet scrap and a good Tim instead of the pads. It's the saving grace of the HP DV-9 series lappy and their overheating issue.