MikeC wrote:It does not sound that great up close the sound is dominated by a buzz -- but this is what Larry reported, too. He measured 15 dBA vs 14 dBA for the reference ATI HSF -- installed and running in an enclosed system.
I just got a ASUS EAH6850 V2; I picked this particular model among others on the strength of your review, figuring I might be able to get away with stock cooling for a while. No way. This is one of the worst ~70mm fans I've heard in the post-Panflo era. Admittedly I try to pick quiet parts, so I usually don't get really bad ones.
Memories: I had to purchase a 9500GT in a hurry a while back and the very last of my vintage new-in-bag Panflos got ziptied onto that. :-)
That's what I figured I do in the short term for the EAH6850 V2, although with bigger fans. I took the fan/shroud off the heatsink. Annoyingly, you have to take the heatsink off the card to get to the screws holding the shroud (time to clean up the TIM and then wonder if you got it right....) With the fan/shroud sitting on the desk, I hooked up a variable voltage fan controller. The buzz in mine persists down to stall speed, only getting slightly quieter. It doesn't seem to be proportional to RPMs. At run speeds, putting a finger in it stops the noise. I'll probably repeat this with a PWM controller later for completeness. All fans are different, but after doing this experiment I decide it wouldn't matter how I tuned the BIOS; the most objectionable aspect of my fan's noise probably would have persisted.
MikeC wrote:You have to remember, we do not test the noise on open bench, it is done in our P180 vga test platform system -- the logic is simple, that you never hear the video card outside the case, and it's unrealistic to test to test cooling/noise at load on the open bench -- way too easy.
Since this is SPCR, we're using the best noise-muffling case you can buy. (It's foam lined inside, has no real noise sources other than the video card.) Larry did say the Asus was the quietest overall -- 15 dBA idle + 18 dBA load -- inside the system/enclosure, which is not only well damped but also well cooled.
I don't have the quietest systems I can buy. I'd imagine there are a lot of builds like my desktop machines: Sonata-class enclosures, Yate Loon fans, Green Power drives mounted on stock antivibration mounts. Power supplies from a decent vendor (although Antec has been a real roll of the dice lately.) Fan mounts, fan controllers. I've had a couple of Antec Fusion HTPCs; they have a straight noise path out the right hand side.
The odd computer out for me is a gaming machine. It's not silent; its design is "use large diameter case fans to make everything else quieter". I had pretty good results with an Accelero S1 on a 4850, either with a Scythe zip-tied to it, or passive with external fans flowing over it. When I went Crossfire, the second 4850 had a good enough factory cooler that it wasn't really objectionable, especially when it already had a lot of cool air going over it. Let's put it a different way: the 4850 CF config was far quieter than my PS3 despite dissipating around four times the heat.
MikeC wrote:Maybe all of you guys complaining of high idle speeds need to work on improving cool air intake to the VGA card?
I doubt it would help me much. My 22% was ~1700rpm. This is in a HAF 922; the side is an open mesh with a 200mm fan. Stock 200mm top. With the top and side 200mm fans set for optimal airflow, or both at 11v, the buzz of the EAH6850 at idle was still clearly audible and objectionable.
I actually would get information from runs on an open test bench for two reasons. First, "gamer" cases are getting that way. The way you can kinda-sorta get away with a mesh case is through careful component picking. More importantly, many mid-quiet machines have straight noise paths or otherwise less than stellar noise mitigation. That's true of the Fusion, and I would never put this 6850 in one. Restricted airflow with clear noise path is the hardest test, but I'd still love to see something like "This card produced a midrange frequency buzz at idle in open air. However when we put it in our standard foam-lined labyrinthine air path test stand, the buzz was not audible."
I think the SPCR review of the EAH6850 will lead to a lot of disappointments if the fans people are getting lately are anything like mine.
Anyway I ziptied two cheap-o 800rpm 120mm fans to the ASUS heatsink and that solved the buzz problem. At a 840/1100 overclock at stock volts I get Furmark 80 degrees, memory controller bouncing around 91 degrees. Idle is 48. I was getting 56/65 in one game. I'm not taking this super seriously because the old Accelero S1 is going on there soon, and I'll sink the VRMs then and undervolt etc. The ASUS heatsink probably wants higher static pressure than an Accelero.
Unlike many 6850 PCB layouts, the S1 looks like it will clear the PCI bracket without tin snips on the ASUS card.
Eventually I may get a second 6850, and since it's nice to have matching cards, I do kinda care what kind of performance I can get out of a modded ASUS heatsink. I'll need to check performance with restricted clearance; 120x25mm fans ziptied to sink may be too tall. If I'm going to put custom cooling on immediately, I need to budget another ~$50.