Why are you running Furmark? Try a game instead.
As long as you don't use the exact same settings as the reviewers did you're working blindfolded.

Even though this review clearly shows that Furmark uses less power with this card than when gaming, it could be a matter of settings.
http://www.techpowerup.com/reviews/ASUS ... II/26.html

Just compare your model with other cards. For instance, the 680 uses 62 W more in Furmark.
I really don't know why there's not a 50 W bump (or so) in the 670 review, but before you do anything else:

Try a game.

To everyone who said it's a air flow problem i have a question (I'm not disagreeing with you)

Why when he has the side panel off is he still seeing 82°C SPCR's own review shows MAX is 72°C and all other information i can find says about the same for MAX load temp.
To me it seems it could be poor airflow but im leaning more towards a bad contact between the GPU and the HS.

If its of any use my Windforce 670 when running furmark hits 83°C in a Antec P182 with two Noctua NF-P12@800rpm for intakes (one top mount and one in the 5.25 bays) with the middle drive cage remove for better airflow.
Then just one for exhaust in the rear as i wanted to create positive pressure in the case with the hope that air would flow past the GPU and out the GPU's rear vent.

If it was my 670 i would redo the TIM, or if there would be a chance of that going wrong i would RMA it.

Why voiding the warranty? Just RMA it. There's no guarantee that new TIM will help, and then he's stuck with a possibly bad card.

Yea i agree, but if i think i can get away with redoing the TIM without breaking any "warranty void if removed" stickers, etc whose to know.

I'm probably going to the lowest depths of hell for doing things like that aren't i

Naah, it's your money tho.

What temps do you get when gaming, compared to Furmark?

Playing Far Cry 3 in 1920x1200 with full details got the GPU to more than 80°C and, yet again, it was too loud...

I'll start by comparing my cards to other cards, in some of my friends' cases. If my card is excessively hotter in the same setups, I'll RMA it.

1) SPCR reviewed system had different case fans/airflow, so you can't use apples-apples.
2) idle temps for both this system and Puget were ~40C for the GPU.
3) temps dropped dramatically when the side door was opened implies starved for air. In a system with decent airflow, temps may drop when the door is closed.
4) open door Furmark of low 80's is fine. That's 80C with the only airflow being the GPU fans...and ambient air convection.
5) 2 and 4 imply the cooler is attached ok.
6) 400 rpm case fans -> insignificant CFM.

Thanks for this insight, CA_Steve, that's somehow encouraging

As written above, I did try raising the case fans speed, with little results. If I understand you correctly, my last hope to avoid an RMA is that the M12-PS are crap.

While i agree you cant directly compare different cases/fans wouldn't having the side door off negate much of the differences and air flow problems?

The idle temps are what leads me to believe its a bad contact with the HS, When my old 8800GT started to fail idle temps where fine but when put under any stress the temps spiked and crashed the system. After lots of messing i found the after market HS was no longer making good contact with the GPU core (found out by wiggling the HS)

What also leads me to believe its a problem with contact is how much Torajirou says the fans are ramping up
I do agree 400rpm for case fans is very low. And Furmark is VERY hard on the GPU its not likely you will see real world temps the same.

Hello,
Have you tried running the Macho without a fan ? If it were me, I'd get rid of the fan on the macho, and block all the empty vented pci slots, pipe holes... how much air is coming out of your back fan ? It should be quite a bit given how much heat is generated on full load...
Finally, if nothing else works, I'd angle the input fan down (more of a hack, depends how nice and parallel you want the inside of the case to look like when the case is closed).

Yep, it might be both airflow and a crappy cooler.

So I tried with (non-PWM) Nexus fans at full speed, it wasn't much better than with the M12-PS.

I reapplied the best thermal paste I could easily find (Artic Cooling MX-2) and tweaked the case fans with Fan Xpert 2 to get acceptable results while playing.

Basically, I managed to limit the GPU fans speed around 50% (which is the upper limit to keep them inaudible) and the case fans around 800RPM. All in all, the system is VERY quiet, and even inaudible, since playing involves game music and sounds which largely cover the system's (almost inexistent) noise. GPU temperatures remains stable around 71°C.

I guess that's a good compromise; I'm a bit disappointed not to be able to get even close to my initial expectations, but I should be glad : I got myself a VERY powerful gaming rig, that does what I want it to do, and that does it almost silently

Great to hear that you're doing some progress.

71° C is nice, you may even raise the temp to 80° C, by lowering the GPU fans.

Or even 85C.

How much did this lower the temps alone?

No difference in idle temps for the GPU.

How much at load though ? i have found GPU's do a fair job of keeping a low idle temp's


I have to ask what where your initial expectations ?

I echo what others have said that you could lower the GPU fans to aim between 80°C - 85°C

I am happy though that you got it sorted

Which fans have you kept in the end ?

The Noiseblockers

So just changing the TIM made thing much better?

It's difficult to say, because I have no rigorous test protocol for load tests; Furmark still crashes if I try to keep the fans in the "quiet" zone.

But I guess it did make things a bit better.

My initial expectations were to have an inaudible system at full load (Prime95 + Furmark), much like the Serenity Pro or SPCR's GTX 670 DirectCU II review test system.

Just out of curiosity do your 670 fans under load run faster than the SPCR's review sample of 1890 RPM
No need to test if you CBA.

Depends what you mean with "under load".

Running Furmark, it goes well beyond 1890RPM. Playing games, I can limit to ~1800RPM and the temperatures stay in check.

Small update: I applied AcoustiPack on the door and on both side panels.

I ordered additional AcoustiPack foam blocks to pad the 5"1/4 and floppy drive bays. I'm curious to see the effect on both the noise and airflow (the airflow should theoretically be more "focused", but I wonder if it will have an impact on noise or temperatures).

I would also like to get a my hands on such a fan mount cover, but Puget doesn't ship to Belgium, unfortunately If they could also sell their niftly little acrylic brace for the GTX 670, that would be awesome ! I actually emailed them to ask But I would need to find a "shipping proxy" in the States or in Canada (if anybody is willing to do so, I would be eternally grateful )

I'll post more pics when I'll consider it a finished project !

EDIT : Coincidence or not... I just got Puget's answer: they won't sell it

Since I'm still not 100% happy with the results and I don't have the feeling I did everything I could to make this build as quiet as possible, I'm planning the following final steps :

* Install an acrylic cover on the top fan mount (I finally found one)
* Apply AcoustiPack to the top of the case
* Still waiting for the Acousti foam blocks to pad the drive bays
* Make use of the tin snips I had purchased to mod my Solo years ago and remove the intake and exhaust fan grills (with the obligatory rubber joints)

Once it's done, I *think* that'll be all

Check out my old P182 rig for some ideas.

In particular:

• Using cable ties to support the GPU (2x cable ties, and the metal hook on the roof of the P18x cases)
• Using corriboard or several layers of cardboard to block the top fan vent, with black texta or tape for aesthetics

I agree with you cutting out the fan grills, it doesn't make much difference to noise, but the difference in airflow (at least perceived airflow, with your hand feeling how much air the fan moves) at low fan RPMs is significant IMO.

I wouldn't bother with the 5.25" foam blocks though, as the vents there provide a secondary source of air for the CPU cooler. Put your hand over the vents there and feel for yourself. But as long as you leave at least one 5.25" bay empty, I think you can have the best of both worlds.

Thanks for the tips, your build is very inspiring indeed !

As for the 5.25" foam blocks, I want to try and see what happens exactly.

Isn't that because the review sample comes from the Puget system, with a tweaked BIOS?