• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

HD 7970 vs HD 6950 Crossfire- some benchmarks and impressions

S

spacing guild

[H]ard|Gawd
Joined
Dec 9, 2010
Messages
1,356
Sapphire HD 6950 2GB Crossfire (locked) OC 865 /1345 -VS- Sapphire HD 7970 3 GB OC 1075 / 1550

1080p 1920 x 1080 2500k @ 4.6 Ghz

This is pretty much what I expected and probably most of you expected. It's just about dead even between the two. To choose an HD 7970 over 6950 crossfire, you're paying a $50 to $100 premium for slightly better min. framerates....and most importantly...BETTER framerates with a little over half the power draw and a full gig more of vRAM for higher resolutions. Smooth as butter vs the occasional stutter with a Crossfire set-up. I saw temps hit 80c...but the fan was bearable in my opinion.

I might have better overclocking potential but 1075 / 1550 is where I feel safe 24/7 ATM.


3dMark 2011 P9160 P9040
3dMark Vantage P31,445 P31315
Heaven 2.5
shaders high, tess normal, Anistrophy 4, AA off
1875 1825

Warhead frost benchmark:

avg. 74.43 min: 60.76 max: 87.48 no AA 64-bit enthusiast
avg. 66.68 min:52.99 max: 77.46 4xAA 64-bit enthusiast

No AA 64-bit enthusiast 67.23 avg. 52.7 min. max. 79.18
4x AA 64-bit enthusiast 57.32 avg. min. 43.47 max. 69.33


Crysis Island benchmark:

no AA 76.93 AVG.
4xAA 66.665 avg.


no AA 71.74 avg.
4XAA 56.945 avg.


Metro 2033 Frontline benchmark:

4xAA 16XAF Very High DX10 advanced physx on
49.40 fps avg. max. 250.38 min. 8.74
AAA 16 AF Very High DX10 Ad. PhysX on
60.01 fps. avg 275 max. 10.87 min.


4x AA/16X ad. physx on avg. 46.77 max. 141. min. 10.70
AAA /16X ad. physx on avg. 57.70 max. 186.88 min. 11.82

Now for a little BF3 testing...:D
 
Last edited:
Very nice results, thanks for posting. I'll definitely be showing this to a few friends who were debating.
 
I have a 6950 CFX setup so I was hoping someone would post some numbers like this.

Too bad I just can't afford it right now
 
60hz display?


Any chance the resolution may be to low? Bottleneck?
 
darkstarcow said:
Your choice of blue text made my eyes bleed.
Click to expand...

I see it fine. I'll change the color...

SonDa5 said:
60hz display?


Any chance the resolution may be to low? Bottleneck?
Click to expand...

Yes I have a 60 hz monitor. I'm not sure what you're referring to. My benches are on par with everything i've seen. My core doesn't seem to want to go passed 1125 without artifacting. If I was @ 1260 core like [H] and other people i've seen, the numbers would admittedly be better.



Uploaded with ImageShack.us



Uploaded with ImageShack.us
 
You should use Orange.........

Awesome results though, I'm upgrading from a single 6970 to (at the moment) a single 7970 so it gives some perspective for me.

Thanks for posting.
 
spacing guild said:
Now for a little BF3 testing...:D
Click to expand...

Can't wait :D

I've been contemplating on whether to get another HD 6950 or go with an HD 7970.

My top games are BF3 & BC2, would you please mention any issues that you encounter with both card setups with these games? Thanks

*subscribed*
 
Last edited:
[D]razio;1038250876 said:
Can't wait :D

I've been contemplating on whether to get another HD 6950 or go with an HD 7970.

My top games are BF3 & BC2, would you please mention any issues that you encounter with both card setups with these games? Thanks

*subscribed*
Click to expand...


Never played BC2.

All I can say about BF3 is...I play a lot of Metro because it's totally noob-spam. I used to fly and cap flags in BF2 but I'm too lazy these days so I get my kicks from machine- gunning bottlenecks with extended clips. (Battlefield 2142 was great that way also)
This is a good measure to judge BF3 with because of all the action and destruction of environments and things exploding all around you all the time.

To be quite honest... I never had a problem with HD 6950 Crossfire in BF3. Always got steady, smooth great framerates but I knew it would be that way. I noticed that one 6950 is decent @ 1080p on ultra (it will dip to the 20's FPS outdoors under heavy fire) so I knew two 6950s would put me right where I wanted to be.

I upgraded to HD 7970 for more vRam and smoother framerates in other games. I notice in the Metro 2033 benchmark with everything maxed, the 6950s will stutter a bit with all the light and smoke effects. Obviously, with a single cards...those dips to 30 FPS and below are much better. Don't get me wrong... my 7970 runs BF3 excellently..I never dip below 50 fps on Ultra and I'm thoroughly happy about it. But I was wanting a single GPU solution while keeping the same or better FPS in most games.

And I like to support to ATi, especially when they put out superior products. Please change the name back, AMD.

Sufficient to say, this card rocks.
 
I'd really like to see some benches for 5040x1050 with 7970 vs. 6950 xfire.

Adding a 6950 would cost me $220.
Selling my 6950 and purchasing a 7970 would cost $600-220= $380.

So, there's quite a jump in price, and I think a second 6950 would be the route to go.
 
Nice work. Not to be a complainer, but it's hard to quickly compare the two setups with the formatting of each benchmark different for each card. ;)
 
Before I upgraded to the 7970 I had 5870s in crossfire, which, apart from smaller vram pool and issues that created with AA, perform nearly identical to the 6950s. While I bought 2 7970s, my PSU wasn't up for the task so I'm stuck with crossfire disabled until the new one arrives, so I'm in a pretty similar situation as you.

I've only played BF3 so far, but from what I've seen w/ 1 7970, the measured framerate is definitely lower than the 5870 crossfire setup, but it plays noticeably better. And by plays, I definitely mean feels, totally qualitative. Chalk it up to "micro stuttering" or whatever affects crossfire so much, but one thing I learned with the 5870s is that the FPS readout never really matched the intuitive feel of the game, and that 90 FPS on crossfire only felt about as smooth as 60 FPS without.

For a more qualitative analysis, I can say that both solutions, 5870x2 and 7970x1, handled AO poorly and that had to be disabled, and I cranked shadows down to low (because IMO low looked better than anything else besides ultra, I'd rather have low res shadows than fewer shadow sources...). I also had 0xMSAA for both. However for the rest with the 5870x2 I used Medium, and with the 7970 I use Ultra, and IMO, the single 7970 in ultra is more playable than the 5870x2 even when set to Ultra vs Medium. All of this is in 2560x1600.

New PSU arrives tomorrow, I'm interested to see if it makes a big difference when I light up the 2nd 7970.
 
J Macker said:
I'd really like to see some benches for 5040x1050 with 7970 vs. 6950 xfire.

Adding a 6950 would cost me $220.
Selling my 6950 and purchasing a 7970 would cost $600-220= $380.

So, there's quite a jump in price, and I think a second 6950 would be the route to go.
Click to expand...

I'll tell you right now. I messed with eyefinity and HD 6950 Xfire. I did not have a good experience. In Crysis 2, even if I set everything to lowest, I was still getting low framerates.
I picked up a third for tri-fire and I got negative scaling. I saw a jump in synthetics and Heaven 2.5 and small bump in BF3 if I messed with the settings long enough. I just couldn't get it to work so I gave it up and went back to single monitor. I'm not sure if the third card being at 4x was the issue...
 
Keep us update on your second 7970 for bf. I am curious if two 7970 stutters making it less enjoyable vs single 7970. What psu are you on now? I'm going to cfx on my ax850 if I can.
 
addictedto60fps said:
It should make a huge difference, as the crossfire Battlefield benchmarks I've seen are pretty good. Not quite perfect x2 scaling, but pretty damn close at 2560x1600.

Edit - good example here: http://www.guru3d.com/article/radeon-hd-7970-crossfire-review/19
Click to expand...

Well my point above was that measured FPS isn't as useful a benchmark with crossfire because or the "microstuttering" or whatever you want to call it. Check this out:

http://techreport.com/articles.x/21516/1

It's an interesting enough situation that I'd really love to see [H] dig into it to. [H] really blazed the trail in videocard reviews by focusing on "playable settings" instead of raw FPS, I think the next logical step (at least in Crossfire reviews) is to address this issue.

WiL11o6 said:
Keep us update on your second 7970 for bf. I am curious if two 7970 stutters making it less enjoyable vs single 7970. What psu are you on now? I'm going to cfx on my ax850 if I can.
Click to expand...


Indeed I will. In the short sessions I got in with crossfire enabled before my PSU shutdown the entire system, I noticed that AO didn't introduce nearly as much mouse lag as it did with a single 7970 and even worse 5870x. More tomorrow, I've been registered but lurking for years, but now I need to hit the 25 post mark so I can post my 5870's (and now an 850w PSU) in FS so I'll keep the updates coming.

I went with the Enermax MaxRevo 1350W, from a Thermaltake Toughpower 850W. The price grabbed my attention, the [H] review and Editor's Choice Gold Award made it a no-brainer. Turns out I was having issues with power even with the 5870's, I just didn't know it. I thought my RAM was bad and couldnt hold the spec'ed 1600MHz, but in retrospect I think it was a power issue with the overclocks. I did some research this week and found a bunch of TDP graphs for RAM, GPU, and CPU overclocking, and they take a real rapid spike once you pass a certain low-overclocked threshhold. 850W just isn't enough to power a i7 920 with a massive overclock, 5 HDDs, 2 DVD-RWs, and 2 cutting edge GPUs. We'll see. I'm expecting to push my CPU and RAM both a few hundred MHz higher with the new PSU, as well as the GPUs.
 
Last edited:
J Macker said:
I'd really like to see some benches for 5040x1050 with 7970 vs. 6950 xfire.

Adding a 6950 would cost me $220.
Selling my 6950 and purchasing a 7970 would cost $600-220= $380.

So, there's quite a jump in price, and I think a second 6950 would be the route to go.
Click to expand...

with eyefinity vrams more important and the extra gig on the 7970 is far more future proof then the 2 GB on the 6950's. especially as more and more games come out using more and more vram. even BF3 can max out 2 gigs when cranking up the graphics. not to mention you can always throw in a second 7970 later on down the road.. with the 6950's you're stuck theres nothing you can add to it to give it more vram.

spacin9 said:
I'll tell you right now. I messed with eyefinity and HD 6950 Xfire. I did not have a good experience. In Crysis 2, even if I set everything to lowest, I was still getting low framerates.
I picked up a third for tri-fire and I got negative scaling. I saw a jump in synthetics and Heaven 2.5 and small bump in BF3 if I messed with the settings long enough. I just couldn't get it to work so I gave it up and went back to single monitor. I'm not sure if the third card being at 4x was the issue...
Click to expand...

the 4x slot was probably an issue and why you didn't see such a huge change in BF3, as far as crysis goes, they are now on their 3rd game and they still suck when it comes to crossfire and SLI support so it doesn't surprise me.
 
sirmonkey1985 said:
the 4x slot was probably an issue and why you didn't see such a huge change in BF3, as far as crysis goes, they are now on their 3rd game and they still suck when it comes to crossfire and SLI support so it doesn't surprise me.
Click to expand...

I'm considering moving my 2nd card to the 8x slot on my MB for airflow reasons, do you know of any good PCIE comparison benchmarks? I know that as of a year or two ago the performance difference was largely statistical, but I don't have any recent numbers.
 
satrafton said:
I'm considering moving my 2nd card to the 8x slot on my MB for airflow reasons, do you know of any good PCIE comparison benchmarks? I know that as of a year or two ago the performance difference was largely statistical, but I don't have any recent numbers.
Click to expand...

I found this here, actually:

http://www.hardocp.com/article/2010/08/23/gtx_480_sli_pcie_bandwidth_perf_x16x16_vs_x8x8/

This is exactly what I'm looking for, but if anyone has anything newer it would be worth while.
 
a 7970 looks good for a single 30" panel and maxed game settings
my 2 6950 cards are staying though for single card configuration use
a 4870x2 and a GTX9800 will be available for a lucky purchaser,I don't hoard the tech
 
Awesome, thanks for the review :) I've been considering selling my pair of 6950s to help fund a 7970 purchase and this is helping me justify that a bit more!
 
spacing guild said:
Metro 2033 Frontline benchmark:

4xAA 16XAF Very High DX10 advanced physx on
49.40 fps avg. max. 250.38 min. 8.74
AAA 16 AF Very High DX10 Ad. PhysX on
60.01 fps. avg 275 max. 10.87 min.


4x AA/16X ad. physx on avg. 46.77 max. 141. min. 10.70
AAA /16X ad. physx on avg. 57.70 max. 186.88 min. 11.82
Click to expand...

AMD Cards can run PhysX now? I thought that was proprietary tech for Nvidia hardware?

I'm considering getting a 97x0 as an upgrade from my 2x GTX 285s, but I don't want to go much over $500 this time for a card. Any good deals on these cards floating around? I don't think I want to drop that much on a 580 that's almost EOL.
 
Last edited:
satrafton said:
Before I upgraded to the 7970 I had 5870s in crossfire, which, apart from smaller vram pool and issues that created with AA, perform nearly identical to the 6950s. While I bought 2 7970s, my PSU wasn't up for the task so I'm stuck with crossfire disabled until the new one arrives, so I'm in a pretty similar situation as you.

I've only played BF3 so far, but from what I've seen w/ 1 7970, the measured framerate is definitely lower than the 5870 crossfire setup, but it plays noticeably better. And by plays, I definitely mean feels, totally qualitative. Chalk it up to "micro stuttering" or whatever affects crossfire so much, but one thing I learned with the 5870s is that the FPS readout never really matched the intuitive feel of the game, and that 90 FPS on crossfire only felt about as smooth as 60 FPS without.

For a more qualitative analysis, I can say that both solutions, 5870x2 and 7970x1, handled AO poorly and that had to be disabled, and I cranked shadows down to low (because IMO low looked better than anything else besides ultra, I'd rather have low res shadows than fewer shadow sources...). I also had 0xMSAA for both. However for the rest with the 5870x2 I used Medium, and with the 7970 I use Ultra, and IMO, the single 7970 in ultra is more playable than the 5870x2 even when set to Ultra vs Medium. All of this is in 2560x1600.

New PSU arrives tomorrow, I'm interested to see if it makes a big difference when I light up the 2nd 7970.
Click to expand...

I had a very similar 5870 xfire to 7970 experience.
http://hardforum.com/showthread.php?p=1038437070#post1038437070
 
I have a 6950 crossfire setup as well. I've been wondering about the comparison of the 7970 against the 6970's or 6950's. It's pretty much what I was thinking like you said and I see no real reason to upgrade and waste the money. I feel okay with the 6950's. The only thing that upset me at first was the heat output in summer but my case has decent airflow so the temps are not too bad. I don't really notice too much stutter with crossfire setup so I can't really say it bugs me.
 
You must log in or register to reply here.
Back
Top