Please Post Links to Reputable 6900 Reviews When They Go Up

Damn dude I'm in shock at this point, fudo was right... about everything!!! OMG, 1536 SP, 6950 in a class of its own, less performance than GTX 580, he was spot on!, not only that the 570 is a real winner! What a comeback by nvidia this round. BTW, to answer your question I think nvidia will push the price down a bit to add more pressure, imagine now the 570 trades blows with 6970 and is cheaper, imagine they lower the price by $20.00 or so, that would kill the reason to buy a 6950 as 570 beats it for only $29.99 more, and that would kill the reason for a 6970 as the 570 trades blows for $39.99 less. They can lower the price of 580 too and I believe they will. Bring on the GTX 560 series!!

lol you spoke 9 mins too soon about the 570.. 6950 better in games and 50$ cheaper. yeah great comeback nvidia :D
 
I don't know what to think.
The HardOCP review is showing quite different results than other websites...
 
The 570 looks like it is even or better at 1920, while the 6970 pulls ahead at 2560, so the resolution you are using matters.
 
I don't know what to think.
The HardOCP review is showing quite different results than other websites...

That's exactly the point I realized after comparing A2A and Hard's style. In the end, what looks like a 20% deficit in A2A turns into one AA level when you look at H's. H's reviews are the most indicative of real world performance, far more than any A2A comparison when it comes to seeing which GPU will give you the best gameplay experience.

TLDR: [H]ardOCP = Win.:)
 
That's exactly the point I realized after comparing A2A and Hard's style. In the end, what looks like a 20% deficit in A2A turns into one AA level when you look at H's. H's reviews are the most indicative of real world performance, far more than any A2A comparison when it comes to seeing which GPU will give you the best gameplay experience.

TLDR: [H]ardOCP = Win.:)
[H] only tests 2560x1600 from what I see, in which case they make the same conclusion as other sites...
Every other bench clearly says, the 6900 series wins at 2560x1600. It looks the 570 is winning slightly (Mostly with AA enabled) at my resolution.

I don't see how testing five games at one resolution is really useful, generally.
At least, you can't make any sweeping statements about the cards.
 
[H] only tests 2560x1600 from what I see, in which case they make the same conclusion as other sites...
Every other bench clearly says, the 6900 series wins at 2560x1600. It looks the 570 is winning slightly (Mostly with AA enabled) at my resolution.

I don't see how testing five games at one resolution is really useful, generally.
At least, you can't make any sweeping statements about the cards.

Yeah you're right. The 6900 series gains performance relative to the GTX 500 series as resolution goes up.TPU Clearly demonstrates what you're saying.

If anything, it shows that a performance deficit at the same resolution with the same settings doesn't usually make a big difference in the grand scheme of things.
 
The one other thing I am considering when looking at this is that the 570 is running on a bit more mature driver than the 6970, so there may be some decent gains over the next few weeks. Hard to say tho as AMD hasent had the best track record in regards to drivers. Just have to see.

Personaly I'm gonna listen to the [H] on this one replace my 285 with the 6970 instead of the 570 I have been drooling on for the past week. <crossing fingers>
 
The one other thing I am considering when looking at this is that the 570 is running on a bit more mature driver than the 6970, so there may be some decent gains over the next few weeks. Hard to say tho as AMD hasent had the best track record in regards to drivers. Just have to see.

Personaly I'm gonna listen to the [H] on this one replace my 285 with the 6970 instead of the 570 I have been drooling on for the past week. <crossing fingers>
I'm in the same boat...
Take a gamble on the future 6970 drivers? I don't know... And the $20 price premium. In a tough spot right now.
 
Personaly I'm not counting on drivers as much. I'm running at a resolution where both cards seem pretty even over all. So even if there are not major driver improvments I still "win". Although if nVidia drops the 580 into the sub 425 range I'll proboly have to run out and kick myself in the crotch a few times.
 
Was going to pick up a 6970 2GB to replace my 5870 1GB for Eyefinity (4320x900), but after seeing a few reviews I may hold off. The 6970 seems to have inconsistent performance increases over the 5870 ranging anywhere from a measly 5% (Bad Company 2) to a whopping 20-30% (Metro 2033) at higher resolutions.

Still can't decide...
 
Darn just ordered a 570.

Should I cancel since it's only $20 more and I'm going triple screen?
 
lol you spoke 9 mins too soon about the 570.. 6950 better in games and 50$ cheaper. yeah great comeback nvidia :D

Now your right if you are only taking this review with only one resolution into consideration but everywhere else seems to show a different story, I'll dig further myself to see what that's attributed too, perhaps everyone else is using 10.11. Either way still looks good for nvidia with the GTX 580. Prices are sure to come down soon. I may get dual 6970s or 580s depending. Gotta see that surround and multi card scaling now. Edit: Also 6970 is very close to GTX 570 and 6950 even in the [H] review so 6970 has me confused and so do the results I'll be back soon after reading more elsewhere.
 
Last edited:
The HD 6950 is impressive - you get better performance and better tessellation than a 5870, but with slightly less power consumption, all for $300.

The HD 6870, however, is an enigma. It costs $70 more for %10 better performance, and for that tiny performance increase you get 50w more power consumption! This is almost as much as the GTX 570 (I checked 3 sites including the [H], and those power consumption numbers are consistent)!

And it's not at all clear why the power increase is so high, because the card is using the same 2GB of ram that the 6950 has, and I can't imagine clocking the thing at 880 MHz requires a major voltage increase (versus say the 850 MHz clock of the 5870, which sips power).
 
blah, was hoping for better overall performance but since I got 1 27" incher and 2 coming on Friday the 6970 wins, besides ATI's driver are probably not at their best right now. Give it a few updates and you might see slightly better AA performance. Cancelled the 570 and just order a 6970. Either way I can't lose at this performance bracket.
 
So, what is the general consensus of which card(s) to go with a multi-monitor setup ? (res: 3x 1920x1080)

Assuming a budget of 500-600~.
 
I think two 6950's is the sweetspot for that budget.
Just based on what the reviews are saying.

~25% faster than the 580 and 5970 on average...
 
Yeah, scaling looks insanely good with the 6950's. In every game tested it was nearly 100%.
 
I think two 6950's is the sweetspot for that budget.
Just based on what the reviews are saying.

~25% faster than the 580 and 5970 on average...

Two 6950 would be the way to go.

I'll keep that in mind for my january build!

I just want sandy bridge to hit already :D

q66/gtx260 -> 2600K/6950 xfire look to be in the works...!
 
Tech Report has their review up now also. The 6950 CrossFire option looks incredible - kills the 580 at about the same price.
 
Darn just ordered a 570.

Should I cancel since it's only $20 more and I'm going triple screen?

A little hard to say but since you are doing triple screen, it appears that might be wise.

OTOH, I'm doing single 1920x1200 and I can effectively get a 570 for $331 shipped (since it comes with a $20 thing I was gonna get anyway) plus I really want to get into doing some CUDA coding and I also run a lot of Premiere Pro CS5 (which at the moment uses CUDA only for effects accel) so I'm almost certain I will stick with the 570 order (and i could swear dark skin tones looked different and weirder on the one AMD card I tried with blu-ray, not sure why either maker should mess with the default video output colors but it seems like there were some differences) the 6970 really doesn't seem to give anything too much on average, wins some games, loses some, wins some effects computations and loses others at my resolution and even if it ends up doing a bit better in comparison to the 570 than I think and matches a bit closer to HardOCP's 2560x results it would still be tough to give up CUDA (and it is $40 more, for me) I'd need a rather definitive win from it which I don't think it will deliver in my situation.

EDIT: although techpowerup also had the 6970 looking a good deal better with some of the DX11 games than the 570, oddly they even had AMD killing nvidia for dirt 2 EDIT it's late messed that up they had DX 11 turned off for Dirt2
 
Last edited:
A little hard to say but since you are doing triple screen, it appears that might be wise.

OTOH, I'm doing single 1920x1200 and I can effectively get a 570 for $331 shipped (since it comes with a $20 thing I was gonna get anyway) plus I really want to get into doing some CUDA coding and I also run a lot of Premiere Pro CS5 (which at the moment uses CUDA only for effects accel) so I'm almost certain I will stick with the 570 order (and i could swear dark skin tones looked different and weirder on the one AMD card I tried with blu-ray, not sure why either maker should mess with the default video output colors but it seems like there were some differences) the 6970 really doesn't seem to give anything too much on average, wins some games, loses some, wins some effects computations and loses others at my resolution and even if it ends up doing a bit better in comparison to the 570 than I think and matches a bit closer to HardOCP's 2560x results it would still be tough to give up CUDA (and it is $40 more, for me) I'd need a rather definitive win from it which I don't think it will deliver in my situation.

Either way it's all good for gaming; I only do light photoshopping and video editing. Only a temp video card until next year when I build an 8 core Sandy Bridge in 12 months.
 
Man, that's a lot of pages. How did they have time to test/write all that.

23 games; i wouldn't call them old; i test 7 DX11 games. How? i got no sleep for 4 days/nights
:(

Did anyone see this as the reason for Cayman's delay?
http://www.chiphell.com/thread-148220-1-1.html
http://www.xtremesystems.org/forums/showthread.php?t=263668

2334064bfmt2tgjc2b4dtg.jpg

The notches on the ones i got are tiny; nothing like the pic. It does not compromise the casing at all.
 
Last edited:
Back
Top