***AMD Radeon HD7970 GHZ Edition - Official Reviews Thread***

[blackened23]

Love when amd fanboys bust out the super high res. Homeless what percentage of people run 3 screens? Its like 1%. 95% of gamers run a 1920x1080 screen so why should we care about super high res benchmarks?

No offense Don, you're OK in my book :thumbsup: but if we all only cared about 1080p what the hell are we doing here? You say 120hz perhaps but who gives a crap? You could get a cheese 560ti 448 for 1080p and it would be fine. You might have to lower a setting here or there but you would barely notice it in games, I can 100% say that if I only cared about 1080p i'd probably use an xbox 360 because its cheaper. I have to differ with you here, this isn't gamespot this is AT.

Who CARES what 99% of people are doing? Steam survey says that 10% of users are using HD3000 and HD4000 graphics, should we switch to that??!

Freaking of course high resolution matters!!!!!!!!!!!! I'm not saying I prefer either brand, but high resolution performance DOES matter and thats why we have 1000$ worth of GPUs. 1080p is for little girls (figuratively) or cheapeskates with 23 inch trash TN panels from a Wal-Mart sale . Lets not forget this, we are enthusiasts. Saying "this %'age uses high res" is so stupid especially in light of the ENTHUSIAST audience here - REMEMBER we own 1000$s in GPUs bro. Falling back on something like that is kinda weak, it is what it is. I don't think there's a clear winner either way and arguing back and forth is just fruitless - just get whatever brand you like because both cards are damn CLOSE in performance....something we should be happy about (close competition)

 

[Homeles]

No offense Don, you're OK in my book :thumbsup: but if we all only cared about 1080p what the hell are we doing here? You say 120hz perhaps but who gives a crap? You could get a cheese 560ti 448 for 1080p and it would be fine. You might have to lower a setting here or there but you would barely notice it in games, I can 100% say that if I only cared about 1080p i'd probably use an xbox 360 because its cheaper. I have to differ with you here, this isn't gamespot this is AT.

Freaking of course high resolution matters!!!!!!!!!!!! I'm not saying I prefer either brand, but high resolution performance DOES matter and thats why we have 1000$ worth of GPUs. 1080p is for little girls (figuratively) or cheapeskates with 23 inch trash TN panels from a Wal-Mart sale . Lets not forget this, we are enthusiasts. Saying "this %'age uses high res" is so stupid especially in light of the ENTHUSIAST audience here - REMEMBER we own 1000$s in GPUs bro. Falling back on something like that is kinda weak, it is what it is. I don't think there's a clear winner either way and arguing back and forth is just fruitless - just get whatever brand you like because both cards are damn CLOSE in performance....something we should be happy about (close competition)

Not everybody wants a 60 foot monitor taking up all of their desk space.

 

[blackened23]

When you're talking about 1200$ worth of graphics in your computer thats not the kind of setup you hook to a wal mart TN panel.

High resolution performance matters, and arguing otherwise for a 500$ product is downright stupid, period. If you want to be in the 99% just use your HD4000 graphics and play at 1366x720. We are a different market and for enthusiasts that want the best - high resolution and high performance matters. Like I said if you're content with a 19" thats great but if you buy a 500$ GPU for that I feel bad for you.

 

[Silverforce11]

DX11 is a API not a implementation set of rules for techniques. How something is implemented has nothing to do with the API.

You should look at more reviews...

Ah, that must be the reason why the performance lost of Kepler is a result of the Adv. Lighting option and not of MSAA...
BTW a 6970 is (much) faster than a GTX580. Do you really think it's a compute problem?

You wont convince many ppl that dx11 direct compute is biased like PhysX. IF you can prove somehow that what AMD did is fishy even remotely close to the invisible ocean and flat surface tessellation on Crysis 2, then you may have a point. Otherwise, its the lack of compute performance on these Keplers thats causing the problem, specifically because the EGO game engine is moving to using direct compute for more features. It's a sign of things to come. First it was tessellation and HDAO, now directcompute.

As to Civ 5, its only directcompute function is in leader animations, not during gameplay. If you want to interpret the results of the other benches where the 7970 is still leading.. include this:

 

[blackened23]

:thumbsup:

You wont convince many ppl that dx11 direct compute is biased like PhysX. IF you can prove somehow that what AMD did is fishy even remotely close to the invisible ocean and flat surface tessellation on Crysis 2, then you may have a point. Otherwise, its the lack of compute performance on these Keplers thats causing the problem, specifically because the EGO game engine is moving to using direct compute for more features. It's a sign of things to come. First it was tessellation and HDAO, now directcompute.

He claims that AMD is cheating in Dirt showdown, which is interesting. The new beta drivers fixes showdown performance, it was due to a driver bug (per a dev post at the nvidia forums). They fixed Shogun 2 performance as well. Note that both games perform signifigantly faster with 304 compared to 301.42.

I guess nvidia is CLEARLY never at fault and is incapable of having driver bugs. Must be stolen AMD cheats correcting the situation.

 

[SirPauly]

1080p is for little girls (figuratively) or cheapeskates with 23 inch trash TN panels from a Wal-Mart sale . Lets not forget this, we are enthusiasts.

Nah, I have a 27 inch, 1080p, 120hz monitor and enjoy 3d stereo gaming and this may be defined as an enthusiast set-up, too -- offers a lot of immersion. Personally think 2d gaming is for little girls, hehe!

 

[-Slacker-]

BF3 gets linked to comparo's because millions of people play the game. If you dont play BF3 you have issues :whiste:

In other words, you're so much of a nv fanboy that you're willing to insult anyone who doesn't play the one of the run-of-the-mill, 13 in a dozen military shooters that you keep banging on about? Wow.

 

[blackened23]

Nah, I have a 27 inch, 1080p, 120hz monitor and enjoy 3d stereo gaming and consider this may be defined as an enthusiast set-up, too -- offers a lot of immersion. Personally think 2d gaming is for little girls,, hehe!

Fair enough. 3D vision is alright, but really not my thing......widescreen gaming though -- i've been warming up to it, its pretty awesome :thumbsup:

Different strokes for different folks I guess. My only point is that this kind of stuff matters for 500$ cards, I cringe when someone says "only 1080p matters". Thats a bad argument for a high end card. Of course widescreen gaming / high res gaming matters when you're talking about a creme de la creme top end video card.

What does someone who enjoys widescreen gaming buy? a 560ti? Hell no, they buy the best. Thats what these cards are MADE for, so *of course* 3 screen gaming matters. Arguing otherwise is -- again -- a horrible argument. Please do note i'm not saying either side is better, but whether you buy a GE7970 or a GTX 680, hell yeah widescreen gaming and high res performance matters! :thumbsup:

 

[skipsneeky2]

Nah, I have a 27 inch, 1080p, 120hz monitor and enjoy 3d stereo gaming and consider this may be defined as an enthusiast set-up, too -- offers a lot of immersion. Personally think 2d gaming is for little girls,, hehe!

That is so wrong,lol.

Full time gaming on my 37'' 1080p 60hz t.v and it is more then plenty for me.

 

[Silverforce11]

A few months ago a single gtx580 was great for 1080p... what's changed? There's hardly any new game out since then that stresses it without turning on a gimmick feature.

 

[Skurge]

High res does matter. The new transfromer prime runs at 1920x1200, The new ipad runs at 2048x1536, the macbook pro runs at 2880x1800. You guys are talking as if 1920x1080 is fine. It's not. How rediculous is it that $500 can have higher resolution displays than enthusiasts like us? How can you be happy with 1920x1080 displays? I can understand you want a 120hz display, but you are saying you are happy with your 1920x1080, yet you are spending over $1000 on GPU's. Tell me how that makes ANY sense.

I may be cheaping out and getting those Korean 2560x1440 IPS monitors. If you have the money for 2x 690s then get 3 of those and OC the suckers to 100hz+.

OT: We need to send a message to Samsung, LG, Dell, etc that we won't accept their sub 150ppi displays. Power to the People

I see no point looking at which $500 card is faster at 1080p cause that is just pointless. Some people say once it's over 60fps it doesn't matter cause our monitors are only 60hz, well I don't care about 1080p either.

/rant

 

[SirPauly]

That is so wrong,lol.

Full time gaming on my 37'' 1080p 60hz t.v and it is more then plenty for me.

 

[SlowSpyder]

1080P is a very popular resolution, and I think it is a valid point to include benches at that res when comparing cards. But, with that said, when talking about $500 - $1200 cards, I doubt most of those are used to power 1080P setups.

 

[sontin]

You wont convince many ppl that dx11 direct compute is biased like PhysX. IF you can prove somehow that what AMD did is fishy even remotely close to the invisible ocean and flat surface tessellation on Crysis 2, then you may have a point.

Look at the screens of Dirt:Showdown. It's a minimal effect and cost more than 50% of the performance on Kepler. That's much more than the whole DX11 patch of Crysis 2. The same thing happend in Sniper 2. DirectCompute costs nearly 40% on Kepler and looks much more worse than Crysis 2.

Otherwise, its the lack of compute performance on these Keplers thats causing the problem, specifically because the EGO game engine is moving to using direct compute for more features. It's a sign of things to come. First it was tessellation and HDAO, now directcompute.

You sound like a broken record. There is no "lack of compute performance". It's AMD implementation which caused the bad performance. Even in Metro2033 - in which Kepler shows his weakness - the difference is only
half as huge.

As to Civ 5, its only directcompute function is in leader animations, not during gameplay. If you want to interpret the results of the other benches where the 7970 is still leading.. include this:

People playing the game and not the leader benchmark.

 

[Elfear]

Fair enough. 3D vision is alright, but really not my thing......widescreen gaming though -- i've been warming up to it, its pretty awesome :thumbsup:

Different strokes for different folks I guess. My only point is that this kind of stuff matters for 500$ cards, I cringe when someone says "only 1080p matters". Thats a bad argument for a high end card. Of course widescreen gaming / high res gaming matters when you're talking about a creme de la creme top end video card.

What does someone who enjoys widescreen gaming buy? a 560ti? Hell no, they buy the best. Thats what these cards are MADE for, so *of course* 3 screen gaming matters. Arguing otherwise is -- again -- a horrible argument. Please do note i'm not saying either side is better, but whether you buy a GE7970 or a GTX 680, hell yeah widescreen gaming and high res performance matters! :thumbsup:

+1

Always makes me laugh a bit when people cling to the 1080p benchmarks like the last lifeboat in the middle of the ocean when discussing $500-1200 graphics solutions. A 6870 or 560Ti would be great for 95% of gamers at 1080p.

 

[guskline]

Fair enough. 3D vision is alright, but really not my thing......widescreen gaming though -- i've been warming up to it, its pretty awesome :thumbsup:

Different strokes for different folks I guess. My only point is that this kind of stuff matters for 500$ cards, I cringe when someone says "only 1080p matters". Thats a bad argument for a high end card. Of course widescreen gaming / high res gaming matters when you're talking about a creme de la creme top end video card.

What does someone who enjoys widescreen gaming buy? a 560ti? Hell no, they buy the best. Thats what these cards are MADE for, so *of course* 3 screen gaming matters. Arguing otherwise is -- again -- a horrible argument. Please do note i'm not saying either side is better, but whether you buy a GE7970 or a GTX 680, hell yeah widescreen gaming and high res performance matters! :thumbsup:

blackened23, what monitor setup are you running on the rig in your sig?

 

[HurleyBird]

Look at the screens of Dirt:Showdown. It's a minimal effect and cost more than 50% of the performance on Kepler. That's much more than the whole DX11 patch of Crysis 2. The same thing happend in Sniper 2. DirectCompute costs nearly 40% on Kepler and looks much more worse than Crysis 2.

The performance hit on this kind of lighting does seem excessive for the IQ increase, but it's probably a necessary advancement in computer graphics. SSAO was the same way, now every game uses it. Obviously neither are able to jump out at you the same way as something like tessellation can.

 

[SirPauly]

+1

Always makes me laugh a bit when people cling to the 1080p benchmarks like the last lifeboat in the middle of the ocean when discussing $500-1200 graphics solutions. A 6870 or 560Ti would be great for 95% of gamers at 1080p.

I don't think it is clinging or even funny but personally find 1080p, 1600p and multi-monitor display resolutions very important over-all. To dismiss 1080p or 1600p, or multi-monitor display resolutions is kinda odd to me.

Maybe many tech sites do as well -- because they offer 1080p, 1600p and multi-monitor display resolutions.

 

[Dark Shroud]

I'm poor & with almost no hardware budget. Yet I still have a 23" high res 2048x1152 monitor for gaming. I'd love to get a 120hz panel but right now I'll settle for what I have and invest in SSDs or video cards instead.

Those Korean monitors sure are tempting though.

 

[hawtdawg]

You guys are all [awesome people that I feel privileged to converse with]

Hey, remember the rule about no personal attacks? I guess not.:|
-ViRGE

 

[Elfear]

I don't think it is clinging or even funny but personally find 1080p, 1600p and multi-monitor display resolutions very important over-all. To dismiss 1080p or 1600p, or multi-monitor display resolutions is kinda odd to me.

Maybe many tech sites do as well -- because they offer 1080p, 1600p and multi-monitor display resolutions.

Don't get me wrong, I'm not advocating that review sites quit running benchmarks at 1080p. It's still one of the most popular monitor resolutions. The point I was trying to make is that I don't get people who cling ONLY to 1080p and dismiss any other resolution because the user base is small. Well yea, so is the user base for high-end graphics cards.

 

[lopri]

I may be cheaping out and getting those Korean 2560x1440 IPS monitors. If you have the money for 2x 690s then get 3 of those and OC the suckers to 100hz+.

Where can I get the information about this monitor overclocking? Out of curiosity. I remember back in the days some people overclocked their PS/2 mice.

 

[psolord]

Hey guys, instead of arguing about Steam surveys and the importance of whatever resolution over another, why don't you start a poll to see what resolution people in Anandtech forums are using? Just to make it real you know.

 

[HurleyBird]

The new beta drivers fixes showdown performance, it was due to a driver bug (per a dev post at the nvidia forums).

So how do the Nvidia cards perform now? It will be interesting to see how much of the performance delta was due to the new compute features and how much was due to the bug. It will also be interesting to see if GTX 580 is faster than HD 6970 now, especially since the former is more compute oriented but the later has more theoretical compute performance.

 

[Olikan]

Hey guys, instead of arguing about Steam surveys and the importance of whatever resolution over another, why don't you start a poll to see what resolution people in Anandtech forums are using? Just to make it real you know.

don't really matter.... we will know how many ppl don't use the card at full potential...

"if 6 billion people belives in a stupid idea, it doesn't change the fact that it's a stupid idea"