Tell us your purchase, HD 6900s.

[adairusmc]

Mine showed up today!

Dorkenstein - What resolution do you game at and what CPU are you running?

 

[Adul]

i am going from an 8800GTX to 6950. Though I am having to wait until christmas to open it.

 

[Tegeril]

I'm also going from an 8800GTX (that is getting RMA'd at the same time for overheating issues, yay lifetime warranty from EVGA) to a 6950 but I'm having trouble deciding which brand. I'm leaning towards XFX for the lifetime warranty but the ASUS cooler is also appealing. Any suggestions?

 

[MegaWorks]

Mine showed up today!

Dorkenstein - What resolution do you game at and what CPU are you running?

Lucky I was waiting for mine today but they didn't show up, maybe tomorrow.

 

[Dark Shroud]

I'm also going from an 8800GTX (that is getting RMA'd at the same time for overheating issues, yay lifetime warranty from EVGA) to a 6950 but I'm having trouble deciding which brand. I'm leaning towards XFX for the lifetime warranty but the ASUS cooler is also appealing. Any suggestions?

Unless that lifetime warranty is important to you the Asus one is probably the better card. I almost wish I had waited to buy my 6970 in favor of the Asus model.

 

[Rezident]

Spent the weekend reading the reviews. Initially disappointed with the 6xxx performance versus rumours but that’s my fault for believing rumours. Looking again at the benchmarks, it’s a very nice boost over my current card (4890) at 19 x 12. I’m doing a full Crysis/warhead playthrough over Christmas and my 4890 just can’t do that properly, looking forward to it now.

580 GTX performance is great but it’s just way too expensive in Ireland (circa. €500!). 6970 2Gb for €329 looks pretty sweet to me, ordered yesterday. Happy Christmas to me!

 

[HumblePie]

Looking at my main rig, I don't really have a reason to upgrade what I have just yet. Only reason I'm even considering it though is to get rid of the 4870x2 card in my "htpc" that uses a bit too much power for that. Of course it does double as a second gaming "lanbox" machine as well. Still, I can get a couple of 6970's for my main, put a 5870 in my htpc, and then just sell off one of the 5870's and the 4870x2 to recoup some of the costs. It's tempting...

 

[jacktesterson]

Asus has best cooling lately. There 6850 cooler is best in that class by far

I sold the 5970 and bought two 6950's.

It's not a huge upgrade but i sold the 5970 for 435.00 so didn't cost that much.

 

[Daedalus685]

How are the temps for everyone?

I've noticed that mine are ridiculously low at load... So low that I'm inclined to believe the sensor is not working... lol

Most reviews peg the load temperatures at around 90C. I can not for the life of me get above 76C, and the fan never ramps beyond 35%. I just ran 12 passes of crysis frost map at max settings (8AA 1920*1200), temperature reached 76 by pass 5 and never climbed above it. I sit at about 56C @26% fan speed at idle. My ambients are also not all that great in my office (average room temperature) though my air flow is exceptional.. Perhaps reviews were using furmark for all the temp testing.

I wanted to point out for those coming from something in the 4800 series like myself that the 8AA performance of the 6900 is unreal. At the cost of perhaps 2fps to average and minimums i can go from 4x to 8x in crysis. The same jump would have cost me playability on the 4890 crossfire (minimums would get tanked and average would drop a good bit as well).


On another note, I am acceptably stable at 935/1435. Folks should note that when using a second display the cards will NOT downclock the memory at idle, nor will the core drop as low.

I have found the following:

With one display and overdrive off the clocks will drop to 250/150 core/mem at idle
with a second display regardless of overdrive you are looking at 500/whatever the mem is set to
With overdrive and one display you will have 350/about half of the set memory speed

Edit:

There is currently a bug in the drivers (10.12a which is actually an older release version than 10.12 but supports the 6900 where 10.12 does not) that causes the secondary display to go blank while running a full screen game/benchmark. This happens only 90% of the time and is strange. Not the end of the world but makes monitoring GPUz a bit of a pain. Full screen windowed mode (such as the style blizzard games and civ 5 use) allow the second display to work fine. Seems as long as the mouse is permitted to track to the second display it will show up alright. Mouse cursors show up in a far lighter colour in WoW than they used to as well, which isn't so much of a problem as it is different.

 

[Grooveriding]

How are the temps for everyone?

I've noticed that mine are ridiculously low at load... So low that I'm inclined to believe the sensor is not working... lol

Most reviews peg the load temperatures at around 90C. I can not for the life of me get above 76C, and the fan never ramps beyond 35%. I just ran 12 passes of crysis frost map at max settings (8AA 1920*1200), temperature reached 76 by pass 5 and never climbed above it. I sit at about 56C @26% fan speed at idle.


On another note, I am acceptably stable at 935/1435. Folks should note that when using a second display the cards will NOT downclock the memory at idle, nor will the core drop as low.

I have found the following:

With one display and overdrive off the clocks will drop to 250/150 core/mem at idle
with a second display regardless of overdrive you are looking at 500/whatever the mem is set to
With overdrive and one display you will have 350/about half of the set memory speed

What case are you using?

 

[Daedalus685]

What case are you using?

I have a cooler master HAF with some custom airflow additions.. It is certainly not lacking in the cooling department but I've never seen my own set up be that different from a review before unless the review was crossfire/sli while I was not.

 

[Grooveriding]

I have a cooler master HAF with some custom airflow additions.. It is certainly not lacking in the cooling department but I've never seen my own set up be that different from a review before unless the review was crossfire/sli while I was not.

I can believe it. I went from a HAF to an 800D with a 480 SLI setup and my temps went up 15C in my gpus. I had to play with airflow and add a bunch of fan to knock 10C off the cards.

The HAF is a monster for air cooling
:thumbsup:

 

[Dorkenstein]

Mine showed up today!

Dorkenstein - What resolution do you game at and what CPU are you running?

Q9550, gaming at 1920x1200. Sorry for the late reply.

 

[evolucion8]

I ordered the Gigabyte version of the HD 6970, hope that it arrives before Christmas. I sold my HD 4870 Crossfire setup, the incoming card should bring me between 15% and 75% of improvements in performance depending of the scenario.

 

[Qbah]

I ordered the Gigabyte version of the HD 6970, hope that it arrives before Christmas. I sold my HD 4870 Crossfire setup, the incoming card should bring me between 15% and 75% of improvements in performance depending of the scenario.

Good luck with that:

HD4870x2 vs HD5870 (= pretty much equal cards)
http://www.anandtech.com/bench/Product/174?vs=162
http://www.techpowerup.com/reviews/ATI/Radeon_HD_5870/30.html

HD5870 vs HD6970
http://www.techpowerup.com/reviews/HIS/Radeon_HD_6970/29.html

You'll get an average of 15% more. At 1920x1200 it's ~13.5%. At 2560x1600 it's only ~16% more.

I'd advise you revise your expectations a bit

 

[Daedalus685]

Good luck with that:

HD4870x2 vs HD5870 (= pretty much equal cards)
http://www.anandtech.com/bench/Product/174?vs=162
http://www.techpowerup.com/reviews/ATI/Radeon_HD_5870/30.html

HD5870 vs HD6970
http://www.techpowerup.com/reviews/HIS/Radeon_HD_6970/29.html

You'll get an average of 15% more. At 1920x1200 it's ~13.5%. At 2560x1600 it's only ~16% more.

I'd advise you revise your expectations a bit

He likely means in situations where the second GPU is not used, or where dx11 is. In the games I play my 6970 improved over my 4890 crossfire anywhere from 0% to 120% when you look at crysis at 4AA and wow as the outliers there.

 

[Qbah]

He likely means in situations where the second GPU is not used, or where dx11 is. In the games I play my 6970 improved over my 4890 crossfire anywhere from 0% to 120% when you look at crysis at 4AA and wow as the outliers there.

Okay, in games where scaling sucks that's true - but which modern game is that? Even Crysis has good scaling now.

And no DX11? The only game that actually gives better performance with the same visual quality is the new WoW expansion (and that's like a 30% boost in GPU limited situations) - there's a review about it somewhere. All the other DX11 games offer visual advantage when enabling DX11, so technically it's infinitely better - you can't really use things like tessellation on the old one

Is the HD6970 better than a HD4870x2? No doubt about it. It's faster and offers great IQ improvements. But saying it's up to 75% faster is just plain wrong.

 

[Daedalus685]

Okay, in games where scaling sucks that's true - but which modern game is that? Even Crysis has good scaling now.

And no DX11? The only game that actually gives better performance with the same visual quality is the new WoW expansion (and that's like a 30% boost in GPU limited situations) - there's a review about it somewhere. All the other DX11 games offer visual advantage when enabling DX11, so technically it's infinitely better - you can't really use things like tessellation on the old one

Is the HD6970 better than a HD4870x2? No doubt about it. It's faster and offers great IQ improvements. But saying it's up to 75% faster is just plain wrong.

But it is up to that much faster in various older games that one might play. In wow my performance more than doubled going from 4890 crossfire to the 6970 simply because crossfire/sli will never work and has never worked in any dx9 game run in windowed mode, coupled with the dx11.

Up to does not mean average, it means in certain crazy situations that may or may not be pertinent you could see that.

 

[Qbah]

But it is up to that much faster in various older games that one might play. In wow my performance more than doubled going from 4890 crossfire to the 6970 simply because crossfire/sli will never work and has never worked in any dx9 game run in windowed mode, coupled with the dx11.

Up to does not mean average, it means in certain crazy situations that may or may not be pertinent you could see that.

Okay. I just don't like it when people mention extreme situations when comparing two products - personal preference, I guess The HD6970 is definitely a better card - faster, quieter, zaps less power and has no CF-related issues. I'm sure he'll be happy with his upgrade.

 

[Daedalus685]

Okay. I just don't like it when people mention extreme situations when comparing two products - personal preference, I guess The HD6970 is definitely a better card - faster, quieter, zaps less power and has no CF-related issues. I'm sure he'll be happy with his upgrade.

aye, for sure.

Just that my reasons for upgrading my similar set up were because of a silly out-lier . Or at least in large part.. that and dx11 + better 8AA.

 

[Spike]

How are the temps for everyone?

I've noticed that mine are ridiculously low at load... So low that I'm inclined to believe the sensor is not working... lol

Most reviews peg the load temperatures at around 90C. I can not for the life of me get above 76C, and the fan never ramps beyond 35%. I just ran 12 passes of crysis frost map at max settings (8AA 1920*1200), temperature reached 76 by pass 5 and never climbed above it. I sit at about 56C @26% fan speed at idle. My ambients are also not all that great in my office (average room temperature) though my air flow is exceptional.. Perhaps reviews were using furmark for all the temp testing.

I wanted to point out for those coming from something in the 4800 series like myself that the 8AA performance of the 6900 is unreal. At the cost of perhaps 2fps to average and minimums i can go from 4x to 8x in crysis. The same jump would have cost me playability on the 4890 crossfire (minimums would get tanked and average would drop a good bit as well).


On another note, I am acceptably stable at 935/1435. Folks should note that when using a second display the cards will NOT downclock the memory at idle, nor will the core drop as low.

I have found the following:

With one display and overdrive off the clocks will drop to 250/150 core/mem at idle
with a second display regardless of overdrive you are looking at 500/whatever the mem is set to
With overdrive and one display you will have 350/about half of the set memory speed

Edit:

There is currently a bug in the drivers (10.12a which is actually an older release version than 10.12 but supports the 6900 where 10.12 does not) that causes the secondary display to go blank while running a full screen game/benchmark. This happens only 90% of the time and is strange. Not the end of the world but makes monitoring GPUz a bit of a pain. Full screen windowed mode (such as the style blizzard games and civ 5 use) allow the second display to work fine. Seems as long as the mouse is permitted to track to the second display it will show up alright. Mouse cursors show up in a far lighter colour in WoW than they used to as well, which isn't so much of a problem as it is different.

Thank you very much. I was wondering why my 5870 on my lanbox with one monitor will downclock to almost nothing (so far down that the fans turn off most of the time) but my 6970 will only go down to 500 on my main pc which has 2 monitors.

I also noticed the second screen going blank issue but only twice, and only with Just Cause 2. Otherwise the card has performed wonderfully with the only problem being hitting 89-91c in BF:BC2 though I blame that on low air flow more than anything else.

 

[evolucion8]

Good luck with that:

HD4870x2 vs HD5870 (= pretty much equal cards)
http://www.anandtech.com/bench/Product/174?vs=162
http://www.techpowerup.com/reviews/ATI/Radeon_HD_5870/30.html

HD5870 vs HD6970
http://www.techpowerup.com/reviews/HIS/Radeon_HD_6970/29.html

You'll get an average of 15% more. At 1920x1200 it's ~13.5%. At 2560x1600 it's only ~16% more.

I'd advise you revise your expectations a bit

With recent driver improvements, the HD 5870 started to distance themselves from the HD 4870X2. specially with the minimum FPS. Also look at the HardOCP review of the HD 6970. While the performance difference isn't great, the playability settings was drasticly different allowing the HD 6970 to run the game at higher settings being only 5% slower than the GTX 580 in average.

Or look at LegionHardware review, Metro 2033 showed a performance difference of over 72% between the HD 5870 and the HD 6970, hence, my performance estimation numbers. (Specially when the DX11 path showed better performance than the DX10 path) Plus I don't have to deal with Microstuttering, high power consumption and scaling issues, it still an upgrade.

http://www.legionhardware.com/articles_pages/gigabyte_radeon_hd_6970_2gb,9.html

"The Radeon HD 6970 was impressive when testing with Metro 2033 at 2560x1600, as it was found to be 11% faster than both the GeForce GTX 570 and HD 5970. Meanwhile the Radeon HD 6970 was just 9% slower than the GeForce GTX 580 and a whopping 72% faster than the HD 5870. "

 

[flopper]

6950x2
watercooled.
idle 43c/34c
happy

 

[betasub]

^ Well, that's one neat way of solving the air-cooling problem for adjacent 69x0 cards
Me, I'll wait for a baord partner to introduce a non-stock cooler/PCB before getting my second 6950.

 

[Burpy]

- XFX 6950 replacing eVGA 9800GT

- XFX 6950 Installed