HD 7850 reviews are starting to come out. Thought I'd start a post to collect the interesting (quiet) ones. 130W TDP.

XFX Black edition @ HT4U (in googlish). I am pleasantly surprised - XFX tends to run noisier. Really quiet and lifetime warranty. woot. Another review at Kitguru. Apples and oranges acoustic setup to HT4U, but a nice datapoint for other Kitguru reviews.
hardwareluxx.de
Tweaktown
Hexus

Sapphire HD7850 Overclock Edition @ Kitguru.
Hardwareluxx.de
Hardware-mag.de

Sapphire HD7850 Dual-X 1GB @ HT4U

MSI R7850 Power Edition with Twin Frozr 4 @ tbreak.
Techpowerup
Guru3D
VR-Zone


PowerColor HD 7850 PCS+ 2 GB @ techpowerup
HT4U
Hexus
X-Bit Labs

Powercolor SCS3 HD 7850 (passive) @ Hexus
Techpowerup
HardwareLuxx
eteknix with fans attached

HIS HD 7850 2GB Overclocked @ Tweaktown

HIS HD 7850 IceQX @ Guru3D

HIS HD 7850 iPower IceQ Turbo @Guru3D

Club 3D HD 7850 Royal Queen @ Techpowerup
Guru3D

Hardware.info roundup includes Asus, Gigabyte, Sapphire, and XFX.

<content added to 1st post>

MSI R7850 Power Edition with Twin Frozr 4 looks interesting with 54C load temp and only 1150rpm vs Asus Direct CU @ 69C. Review at tbreak. Should have the potential for sub-800rpm while gaming.

Powercolor added.

If Kitguru is to believed, Nvidia is trying to reduce their overstock of the GTX 570..which will delay the GTX 660 intro until August. So, no Nvidia Kepler competition to drive down 7850 prices until August.

Hardware.info roundup added

HIS Overclock and HIS IceqX added.

Club 3D Royal Queen added.

Two more Powercolor reviews and one MSI PE review added.

Another Club 3D review added.

I love these reviews, "Often large fans can cool very effectively and produce less noise, but Gigabyte's card creates 68.4 dB(A) of noise at a distance of 10 cm." (from the Hardware.info roundup).

Here's another for the Gigabyte: http://www.vortez.net/articles_pages/gigabyte_hd7850_oc_windforce_review,1.html

The Vortez take on the Gigabyte cooling: "It [competes with the 570] while being the coolest card on test along with being one of the most power efficient and for the most part, it was also the quietest. I say the most part because while in 'normal' operation, the cooler was extremely quiet. Even when overclocked the cooler didn't really go much over what can be best described as a murmur. However when the fan was set to 100% it was clear their [sic] were issues with the fans. They made a very loud 'rattle' which only subsided when the fans were set below the 70% threshold."

PS. Found it through the Techpowerup list: http://www.techpowerup.com/reviewdb/Graphics-Cards/ATI/HD-7850/

Yeah - there is quite a lot of variation between reviewers on what is "quiet". I'm only listing reviews that take a stab at decibel measurements...while the methodology varies, at least you can get a sense of how cards compare.

That's a good criterion to pick with. At least they're TRYING, if they got some actual decibel measurements in there and not just a haphazard estimate.

The XFX looks to be a bit of a dark/black horse, shame it doesn't seem to be available in Finland.

This is OT, but might be important for someone in the market for a 7850: the recent price drops may have brought some 7870s down into old 7850 price figures. The 7870 is basically a little beefier 7850, has a little more of everything, summarily some 45% more TFLOPS. So, if you had the money for a 7850 before, you might now buy a 7870 for that - the 79xx cards are still not coming down remarkably.

The added grunt comes with added heft of course: much longer cards, and the power consumption is higher (essentially increasing heat output; not so much it would drastically affect cooling, but may increase user discomfort).

Do you know how the "use discomfort" of the HD7850 compares to the HD5750?

The difference itself may not be much, but if you're working in a thermally inadequate environment (high stress, low cooling), every watt counts.

I have used an HD 5770, followed by a GTX 460, and there the difference could be felt. Pulling some reference figures from Techpowerup charts (averages, peaks), the 5770 was 88 W on average and peaked at 93 W (quite close to the 7850 in that regard; 87 and 96 W respectively), and the 460 had a 122 W average and 130 W peak consumption. The 5770 was tolerable to work and even game with, but the 460 was insufferable, fast approaching torture in the summer - it was an overclocked card too, and a reliable review put the consumption figures closer to 140 W. So, a calculated difference of 34 W on average and 37 W at peak was enough to be felt, and a card consuming well over 100 W felt intolerable in my study/workspace in the summer.

As regards to the 5750 and 7850, I would presume the 7850 to be "tolerable" (as above) compared to the 5750, as it does not break 100 W even at peak load, but the difference here is 22 W on average and 26 W at peak, so I think you'll certainly feel it.

My case was the worst case scenario: no airconditioning or mechanical ventilation in the study, so the thermal environment was not only stressed but had poor cooling capacity. I also drove the card hard in gaming sessions and with continuous daily use (HTPC as well as workstation and gaming rig), exacerbating the effect. If you have a cooler environment, you may well get away with a slightly warmer card - but you have been warned!

PS. I have used an Intel iGPU this summer and the difference is remarkable, even to the HD 5770. The study is completely cool, even in the worst summer heat, which has been excellent for working at home and good times with games and movies - both of which the HD 4000 clears admirably, even if not perfectly. Recently played through Spec Ops: The Line even. Only really need the FPS for competitive online gaming.

7850 maximum power consumption is 101W and 7870 144W, this is felt when playing very GPU heating games (e.g. witcher 2 - other games have lower temps). Since I have 2 cats and can't remove any dust filters or fan grills, I would need new case and larger fans. Larger fan allow much more air to pass silently through the dust filters and grills, when turned down to have no airflow noise.

Currently I have only 14cm case fans, but 7870 would require maybe 180mm Air Penetrator to top + 140mm rear fan to remove heat in a cat-safe case. For higher end 150-200W graphics cards Shinobi XL with 230mm case fans would be necessary.

If you are referring to the Techpowerup Maximum chart, then that is not accurate. You are right in saying load does vary by game, but those figures do not represent that variation.


The Peak draw represents heavy gaming load draw with Crysis 2 at 1920x1200 and Extreme settings, single highest measurement. The parts I've highlighted explain the main differences in the measurements: the low resolution with no image enhancement is used to produce extreme FPS in an extremely taxing, constant rendering process, and the card is tampered with to lower or remove safeguards and optimisation that would normally be in place. This is the sort of use you'd expect in a rendering farm application. It is not representative of gaming load that varies and is self-adjusting due to heavier tasks limiting the FPS the card can even render, and the lighter parts hardly being as taxing.

As the article states, the Average figure is what one can typically expect in gaming use (averaged from 12 runs of Crysis 2), and the Peak is what a heavy game load looks like - momentarily. The Maximum is something else, mostly useful for seeing where the card's limitations lie I would say; we can see the extremes of the capacity the 7870 has that the 7850 lacks.

Witcher 2 at ultra settings (ubersampling on) is a rendering farm application. It renders every frame a few times to achieve much better image quality than any regular AA method. Crysis 2 isn't nearly as gpu-heating even without v-sync. W2 power usage is somewhere between peak and maximum.

That's good to know, but do you have some actual figures you could reference? We are after all just hobbyists discussing a subject.

I don't have any laboratory measurements to show, but lot of people are saying they are getting same high gpu temps and fan speeds with Furmark stress test and with W2 ubersampling enabled. W2 isn't cpu limited or memory bandwith limited, just raw GPU processing with ubersampling, and power draw is similiar to maximum.

W2 is also partly a stealth genre game, so you really don't want any undersized or loud case or video card cooling when playing the game.

Throwing my 2 cents in...in general I've found the actual power use while gaming is far below the max. However, take a look at my signature. When I was playing SWTOR, the average power use came within 20W of my stress test results. I think this was a case of my PC barely having the horsepower to provide a decent fps experience even with some quality features dialed down.

Oh yeah, I noticed that one day, Steve, but forgot all about it. Now we've got something concrete to go on! Guess there are some crazy games out there; honestly, I can't see the benefit of torturing a graphics card like that when engines used in Battlefield 3 or Spec Ops: The Line look pretty amazing already. I would include Crysis 2 in that, but it feels so poorly optimised.

So, I guess if you're playing Witcher 2 or SW:ToR, you'd better get a cool-running card.

Just started Guild Wars 2...guess I better grab the Kill-a-Watt.

Sapphire HD7850 Dual-X @ HT4U added.

Sapphire @ Hardware-mag.de and XFX @ Hexus added.

What about a six months old review:

http://www.overclockersclub.com/reviews ... _dd/16.htm

They call those XFX as "DD" (CDFC) but by the photos they would look like the "BE" (CDBC).

I'm only compiling reviews that show SPL testing.

HIS iPower IceQ Turbo @ Guru3D added.

Powercolor passive added. Review @ Hexus.