AMD Radeon HD 6970 and 6950 Video Card Review @ [H]

I'm waiting to see the [H]'s comparison of these boards in CF/SLI

From what I've read elsewhere the 69XX boards scale better in CF than Fermi, creating this rather weird phenomenon:

HD6950 < GTX570 < HD6970 < GTX580 < GTX570SLI < GTX580SLI < HD6950CF < HD6970CF, at least in some titles.


but I don't trust these other sites as much as I do the [H] so I'd like the [H] take on this :)
 
Zarathustra[H];1036570768 said:
I'm waiting to see the [H]'s comparison of these boards in CF/SLI

From what I've read elsewhere the 69XX boards scale better in CF than Fermi, creating this rather weird phenomenon:

HD6950 < GTX570 < HD6970 < GTX580 < GTX570SLI < GTX580SLI < HD6950CF < HD6970CF, at least in some titles.


but I don't trust these other sites as much as I do the [H] so I'd like the [H] take on this :)
Click to expand...

I didn't see 6950CF beat 580SLI unless you're looking at the F1 2010 results.
Nvidia would probably release a driver update for that game, AMD needs a number of driver updates for various games using the 6970.
The 6970 has the ability to perform more consistanly than it has.
 
Sottilde said:
Not entirely. Power usage these days is much improved from back in the day but a tiny chip like the 5450 uses less power, period. There is much less logic in the chip and far less streaming processors, which means that there is far less leakage. Not to mention, most low-end cards run so cool that they don't even need heatsinks with fans. The 5450 is one of my favorite cards, if you can't tell... it's tiny, cheap, and does the job very, very well.

Also a smaller chip will have a narrow memory bus, etc... which has far less leakage.

See this chart - notice that the 5000 series has much improved idle power consumption on the high end, but it still can't beat a bare-bones chip:

21605.png
Click to expand...


WTF? My 4870 was chewing through MUCH MORE JUICE?
 
TwistedAegis said:
Do we have to rehash this same bullshit with EVERY goddamn video card review?! Can someone just pull all the usernames that post these arguments and forward them to their own forum? Or 4chan?

Great review guys. As a proud owner of a 30" Dell, I still value my money. I used a large severance package (thanks, financial crisis!) to purchase it; I don't usually have $1,200 lying around to just buy whatever I would like.

That being said, if you have two cards that perform at nearly the same level, even if money is not the issue, why pay 40% more? That's valuable information. What's more is that given the sporadic performance of the 6970 I think it is safe to say that it is driver related; once that gets worked out we should see better performance. And talk about scalability; for 150% of a GTX 580, you can have CFX 6970s, with a full 1 GB more memory.
Click to expand...

+1 Well said
 
Sottilde said:
Not entirely. Power usage these days is much improved from back in the day but a tiny chip like the 5450 uses less power, period. There is much less logic in the chip and far less streaming processors, which means that there is far less leakage. Not to mention, most low-end cards run so cool that they don't even need heatsinks with fans. The 5450 is one of my favorite cards, if you can't tell... it's tiny, cheap, and does the job very, very well.

Also a smaller chip will have a narrow memory bus, etc... which has far less leakage.

See this chart - notice that the 5000 series has much improved idle power consumption on the high end, but it still can't beat a bare-bones chip:

21605.png
Click to expand...


WTF? My 4870 was chewing through THAT MUCH JUICE?

I gather a 6950,0r 5970 would STILL use less than my 4870 did?
 
Brent_Justice said:
I knew this is exactly what people were going to say. That's why I made the first paragraph you see in the conclusion specifically address this topic.

I don't think the intention was to ever beat out the GTX 580. Certainly not at that price.
...
Click to expand...
Are you sure Cayman wasn't originally targeted higher? I think my disappointment lies in that so many of the Cayman rumors indicated 1920SPs, which suggests that the HD6970 has SIMD units disabled. If this is true, then a "full" Cayman would probably be faster than the GTX580. So why would AMD disable units when it had the chance to take the "performance crown"? Maybe yield issues. Maybe AMD wants to avoid a large gap between HD6870 and HD69XX. Maybe Cayman was only 24 SIMDs from the get-go.

Either way, HD6970 is not the card I was hoping for. I wanted a $400 - 450 single-gpu high-performance card. That price range is pretty much empty (aside from CF setups, which I want to avoid) and I'd rather not settle for HD6970 or spend extra for HD6990.
 
Alaska Wolf said:
WTF? My 4870 was chewing through THAT MUCH JUICE?

I gather a 6950,0r 5970 would STILL use less than my 4870 did?
Click to expand...

This should probably be a different thread, we're getting off-topic here. But yes. AMD/ATI introduced very significant idle power improvements in the 5000-series.
 
Thanks, el jefe y otros...good job as usual.

(Insert Ray Liotta voiceover here: "Low-res is for schnooks.")
 
Is there any consensus on which brand of 6950 would be better? I see the Gigabyte and Asus are already sold out on the Egg. But the XFX , Sappire, and His are still there.
My dead 4870 was a Sapphire.
 
You can fix/tweak/customize idle clock speeds and voltages in about 5 minutes with any of the cards.
Oh right, yeah, you probably can't do that in CCC yet. My mistake.
 
XFX is always going to back up to its unlimited lifetime warranty. To some thats incredible peace of mind, to others, who cares.

My 4890 is XFX, and with this release, I might just be going back to that well again. Again, thats my personal experience, no issues, but you do pay that premium (albeit slight as it may be) to have that extra warranty.
 
Alaska Wolf said:
Is there any consensus on which brand of 6950 would be better? I see the Gigabyte and Asus are already sold out on the Egg. But the XFX , Sappire, and His are still there.
My dead 4870 was a Sapphire.
Click to expand...

Sapphire, ASUS, XFX and Gigabyte are what I prefer. If I had to put them in order of preference it would be:

ASUS > XFX > Gigabyte > Sapphire

Though I will not pay the XFX (tax) premium if I can find the other 3 brands for cheaper or same price.
 
Thanks for the replies about brands! I've been warming up my car so I can go to the store and run some errands. I think I might pull the trigger on one of these when I return in a few hours.

Thanks again!

Man it's nasty outside today!

Brrrr.
 
Alaska Wolf said:
Is there any consensus on which brand of 6950 would be better? I see the Gigabyte and Asus are already sold out on the Egg. But the XFX , Sappire, and His are still there.
My dead 4870 was a Sapphire.
Click to expand...
Just to clarify, the ASUS hasn't even started selling yet!
Still waiting on NewEgg to stock them in the first place.
 
WorldExclusive said:
I didn't see 6950CF beat 580SLI unless you're looking at the F1 2010 results.
Nvidia would probably release a driver update for that game, AMD needs a number of driver updates for various games using the 6970.
The 6970 has the ability to perform more consistanly than it has.
Click to expand...

I was referring to these results. First chart on page.
 
I have had many ASUS cards without any problems. Sapphire is good as well, but not top-tier like ASUS.

Never used XFX. The lifetime warranty isn't as big of a deal to me, they are more money because of it, and some of their revised PCB designs (those boards not based on reference design) have used inferior components from what I have read.
 
Zarathustra[H];1036571031 said:
I was referring to these results. First chart on page.
Click to expand...

OK, that's fine. AMD has extraordinary texture fill rate. Great for games like Metro and Crysis.
The pontential is there just still kinda weary about many of the games being so driver dependent.
When will AMD fix it? Who knows.
 
Alaska Wolf said:
Is there any consensus on which brand of 6950 would be better? I see the Gigabyte and Asus are already sold out on the Egg. But the XFX , Sappire, and His are still there.
My dead 4870 was a Sapphire.
Click to expand...

I personally ordered a Sapphire because it, unlike the other brands, comes with a mini-DP to DP adapter. And all the boards this round are reference boards, so there shouldn't be any difference between them.

2-year warranty is enough for me. I'll be upgrading by then.
 
So from the reviews im reading.. 6970 is better for high end resolutions such as 2560x and crossfire doesnt suck a D now?
 
Kenjiwing said:
So from the reviews im reading.. 6970 is better for high end resolutions such as 2560x and crossfire doesnt suck a D now?
Click to expand...

CF is up and down depending on game. I play Metro, BFBC2, F1 2010 a lot now so it's doing fine. In other games not so much.
 
Climber said:
You're going to be happy with either of the cards to be honest at that resolution. Its like asking if you want a GT500 or a Z01. Both are awesome and will draw double takes whereever you drive them




Check out the DCS: A-10 Warthog. While not an F22, it is a Warthog!

Fallout 3 and Mass Effect are two great RPG's you really shouldn't miss.

Assassins Creed 2 is a very beautiful and detailed game.

Batman Arkham Asylum is another you shouldn't miss.

Quite a few games that you will enjoy I think.

Also check out Minecraft! ;)
Click to expand...

Ditto on Minecraft, insanely addicting.
 
Vexerz said:
CF is just ridiculous this round. CF > SLI actually for Cayman vs Fermi.
Click to expand...

I would love to see some serious eyefinity + CF reviews this round. I really hope they fixed the stuttering issues. FRAPS used to report anywhere from 50-80 FPS in eyefinity in BF:BC2 with my 5970, but it never felt like it was more than 30. It actually felt smoother with Catalyst AI disabled (crossfire off).
 
Kenjiwing said:
...and crossfire doesnt suck a D now?
Click to expand...

I would be careful to draw that conclusion.

CF does benchmark higher than SLI in some titles, leading to a role reversal in standalone vs SLI benchmarks. Single GTX580 is king, but in CF/SLI 6970 takes lead.

That being said, this does not speak to any of the many other reliability, stuttering and other issues people have experienced with CF in the past.

I think we are still in a wait and see hold pattern as far as CF goes.
 
Sottilde said:
I would love to see some serious eyefinity + CF reviews this round. I really hope they fixed the stuttering issues. FRAPS used to report anywhere from 50-80 FPS in eyefinity in BF:BC2 with my 5970, but it never felt like it was more than 30. It actually felt smoother with Catalyst AI disabled (crossfire off).
Click to expand...

Its been know that the 58xx series (5970 based on) were inconsistent when it came to CF. All the other 5 series were fine. Can't go wrong with 2 x 6950s for multi display.


As for CF in general.
Crossfire benches for Cayman are only a few clicks away. Yet they continue to deny CF is better this round. Notice all the criticisms in this thread also apply to SLI as well but the naysayers don't mention that. I guess we still have some doubters so just wait for [H] review.
 
srangara said:
Roger.
Click to expand...

I read somewhere that the HD6970 and GTX580, based on the setbacks in die shrinks are likely to reign at the top of single GPU boards through 2011. The only faster boards in this time period are likely to be multi-GPU boards.

I'm not sold on this yet, but it would be nice if this were true. That kind of longevity for a high end video card would be pretty cool...
 
I think that new Catalyst drivers will be coming out on the horizen which should improve the performance of these new AMD cards. Anyone else feel like this is a real possibility?
 
Brent_Justice said:
We use default driver settings on both AMD and NVIDIA control panels. It is the only fair way to compare. We are looking for the out-of-box experience.
Click to expand...

You guys are aware about the degradations AMD has put into their drivers recently at the default quality filtering setting? It was confirmed by 2 german sites and now guru 3d as well. They net up to 10% extra performance at that filtering setting. IMO the only fair way to bench would be HQ vs HQ since neither has degradations there and after all they are both high end cards. Oh well, I knew something didn't seem right with the results. I'll have to extrapolate and consider the extra up to 10% performance from AMD and deduct it from their results which bring me back to what I found before this review. I need to do this because I would be cranking to HQ especially using one of these beasts.

GTX 570 and 6970 trade blows, 6950 is slower than GTX 570 by a little and the best value overall at $299.99, 6970 is not worth the premium over a GTX 570 unless you are going surround also. 580 is even further away from 6970 than this review showed.

Here is a quote form Hexus

We like to benchmark with 'out-of-the-box' performance with driver-default settings. But there's been plenty of hoo-hah over AMD reducing driver-default image quality for post Catalyst 10.9 drivers. The situation is documented here. Bearing this in mind, we've manually adjusted the image-quality settings for the Radeon HD 6950 - disabled Catalyst A.I and slid filtering quality to High Quality - and shown it in all the graphs. It's identified by the IQ suffix. This, we hope, provides an apples-to-apples comparison to NVIDIA's default image quality.

However, please bear in mind that we will continue to use driver defaults in all future reviews, unless there's a marked difference in image quality. Further, the out-of-the-box HD 6950's numbers are the ones you should concentrate on.
Click to expand...
 
Kaldskryke said:
Are you sure Cayman wasn't originally targeted higher? I think my disappointment lies in that so many of the Cayman rumors indicated 1920SPs, which suggests that the HD6970 has SIMD units disabled. If this is true, then a "full" Cayman would probably be faster than the GTX580. So why would AMD disable units when it had the chance to take the "performance crown"? Maybe yield issues. Maybe AMD wants to avoid a large gap between HD6870 and HD69XX. Maybe Cayman was only 24 SIMDs from the get-go.

Either way, HD6970 is not the card I was hoping for. I wanted a $400 - 450 single-gpu high-performance card. That price range is pretty much empty (aside from CF setups, which I want to avoid) and I'd rather not settle for HD6970 or spend extra for HD6990.
Click to expand...

I wouldn't trust rumors.

I agree there is room there for a higher end card. I would like to see the dual-GPU card fill this gap.

You cannot deny the awesome performance at $369 however.
 
UtopiA said:
Just to clarify, the ASUS hasn't even started selling yet!
Still waiting on NewEgg to stock them in the first place.
Click to expand...

You are correct it appears! I like the idea of the lifetime warranty, but wonder if Asus might have better build quality. I've use their boards for years now. XFX I've heard mixed things about. The extra $10 for the XFX is NBD.

NOW I'm vacillating between the 6950 and the 6970. Wonder if the 6970 is noticeably louder on low usage? I'm also more or less OK spending another $60 if it would translate into a better gaming experience on my 52". I think it IS time for me to get into this end of computing!

It's going to be a while before I can spring for a new monitor or two, or three...

ASUS EAH6950/2DI2S/2GD5 Radeon HD 6950 2GB 256-bit GDDR5 PCI Express 2.1 x16 HDCP Ready CrossFireX Support Video Card with Eyefinity
Be the first to review this product...

OUT OF STOCK. ETA: 12/21/2010.


ASUS EAH6970/2DI2S/2GD5 Radeon HD 6970 2GB 256-bit GDDR5 PCI Express 2.1 x16 HDCP Ready CrossFireX Support Video Card with Eyefinity
Be the first to review this product...

OUT OF STOCK. ETA: 12/21/2010
 
Lord_Exodia said:
You guys are aware about the degradations AMD has put into their drivers recently at the default quality filtering setting? It was confirmed by 2 german sites and now guru 3d as well. They net up to 10% extra performance at that filtering setting. IMO the only fair way to bench would be HQ vs HQ since neither has degradations there and after all they are both high end cards. Oh well, I knew something didn't seem right with the results. I'll have to extrapolate and consider the extra up to 10% performance from AMD and deduct it from their results which bring me back to what I found before this review. I need to do this because I would be cranking to HQ especially using one of these beasts.

GTX 570 and 6970 trade blows, 6950 is slower than GTX 570 by a little and the best value overall at $299.99, 6970 is not worth the premium over a GTX 570 unless you are going surround also. 580 is even further away from 6970 than this review showed.

Here is a quote form Hexus
Click to expand...

Thank you for your feedback, we are going to continue to test fairly by testing at default driver settings from both NV and AMD and report the gameplay experience to you.
 
Lord_Exodia said:
You guys are aware about the degradations AMD has put into their drivers recently at the default quality filtering setting? It was confirmed by 2 german sites and now guru 3d as well. They net up to 10% extra performance at that filtering setting. IMO the only fair way to bench would be HQ vs HQ since neither has degradations there and after all they are both high end cards. Oh well, I knew something didn't seem right with the results. I'll have to extrapolate and consider the extra up to 10% performance from AMD and deduct it from their results which bring me back to what I found before this review. I need to do this because I would be cranking to HQ especially using one of these beasts.

GTX 570 and 6970 trade blows, 6950 is slower than GTX 570 by a little and the best value overall at $299.99, 6970 is not worth the premium over a GTX 570 unless you are going surround also. 580 is even further away from 6970 than this review showed.

Here is a quote form Hexus
Click to expand...

Fact you can't see the difference with the naked eye makes it a moot point, if i you have to zoom an image 100% to see the diffrence then this is just sour grapes from twits in green. :mad:
 
Alaska Wolf said:
You are correct it appears! I like the idea of the lifetime warranty, but wonder if Asus might have better build quality. I've use their boards for years now. XFX I've heard mixed things about. The extra $10 for the XFX is NBD.
Click to expand...

Aren't these all reference boards at this point? Unless I am mistaken, they are all pre-built and the OEMs just slap different stickers on them so I don't think there is any build quality difference between any of them.
 
I never thought I'd wish I could afford a $3000 monitor for games. Of course I never thought at 60 years of age I'd have to go back to school to learn how to use my telephone either. I can text, surf, and take videos OK but I screw up answering and placing calls. I'm informed I'm a champion Butt Dialer though. WTF?

Those 2560 monitors are TOO spendy for me. What would be a GOOD 2 or three monitor substitute for my HDTV? I gather the 6950 or 6970 OR the GTX 570 can run both my HDTV AND a two or three monitor setup?
Click to expand...

i see your running on a 55" tv. To give you some input alaska, i have a 40" sharp aquos 1080p led tv.

I can max just about any game out at this resolution with my 6870. Tvs do not make the best montiors, it takes a very good tv to make a semi decent monitor. I have a top of the line tv, but even with it i have some small text reading problems. It looks great in games, and i love to be able to sit on the sofa 7-8 feet away playing with a xbox 360 controller. Problem with such a big screen and FPS games, using mouse and keyboard, you have to sit close to it, While great for aiming across the map with sniper rifles and stuff, it does strain your eyes a good bit. I have a hannis G 28" monitor as well which runs 1900x1200, it has a much much sharper picture, and no text problems. I actually prefer it to my 40" tv. I just hate watching tv on a 28" monitor. Since my pc/tv/blueray is all in one, i stick with the 40".

the 6970 /6950 are great cards, and are really meant for people playing over 1080 res/ eye infinity. @ kyle great review man, ever since i shut down inside-hardware.net I have always liked your reviews the best to keep me up to date with the pc gaming industry, you do things the right way.
 
I wouldn't say I'm dissapointed, but expectations were really high on cayman. People were expecting at least 20% higher performance than the 580. Never the less great performance per dollar.

The thing that tips the scale IMO is memory size. Specialy with the 570 vs 6950. From what I've seen @ 1920x1200 the 570 can meat or beat the 6970, but at 2560 even the 6950 beats it.

BTW the line has become a lot blurier than ever. Even though the games you tested corroborate your findings, there are other games that give different results and even tips the scale to one side or the other.
 
Stoly said:
I wouldn't say I'm dissapointed, but expectations were really high on cayman. VERY FEW People were expecting at least 20% higher performance than the 580. Never the less great performance per dollar.

.
Click to expand...

Hell no to the original so I fixed it for you. I've been checking the [H] cayman daily since it began (closed one) and another forum.
 
You must log in or register to reply here.
Back
Top