• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

NDA for Cayman is lifted 22nd?!

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
blah blah

While it appears AMD will still hold the performance crown it gets old with stupid leaks here and there. I know they want to confuse nvidia and all but what nvidia already has in queue is all they have. They can't suddenly pull out a 28nm gpu or anything. AMD should just go ahead and have a preview day or something to dispell all of the trash based news.

Either way, if any of the so-called 'leaks' is any indicator then AMD's new GPU's will be a beast (In a good way, and well maybe in a not so good way as well)

Anyways I hope Anandtech doesn't give in to nvidia's pr trash and stick a stupid oced video card in the AMD GPU launch again. :thumbsdown:
 
blastingcap said:
P.S. Do I detect a bit of pride where you wrote: "Google GT200 anti-aliasing investigation - it's the first hit"? 😉 Nice article.
Click to expand...

Well, not pride in the sense of arrogance. The reason he told you how to find the article is because he himself cannot link it per the TOS. He is following the rules. As far as I am concerned, BFG is the authority around here when it comes to graphics cards...
 
Grooveriding said:
Whereas from rumors and some of these leaked specs the 6970 has some major changes, new additions and is a bigger die compared to the 5870. I think there is more room for a bigger performance leap than what nvidia could do with what is basically a full shader count GTX 480 in the 580.
Click to expand...

Of course there is more room for AMD to make a performance leap. Their die size with HD5870 was much smaller than GF100 and their performance/mm^2 was superior already, while their tessellation/geometry engines were not very advanced. Couple these 2 advantages and sub-part tessellation setup in the 1st DX11 generation with a new revised VLIW4 architecture design with 50% more shaders (480 / 320), and a giant leap in tessellation/geometry performance through parallel tessellation engines, and you can get ~35-50% performance increase.

Don't forget that GTX580 is ~ 35%+ faster than the HD5870. So HD6970 has to be a much greater leap from HD5870 than GTX580 was from the 480.

I am very impressed in their ability to improve ROP efficiency by 2-4x, which means they didn't need to go beyond 32 ROPs. The focus on performance/mm^2 just permeates this presentation. Very impressive if it translates into direct real-world benefits (unlike up to 2x the tessellation of HD6870 over HD5870...)

It would be immensely impressive if HD6970 can defeat the GTX580 since AMD's original "small-die" strategy only positioned dual-GPU designs against 1 monolithic NV core design. Can't wait for the benches. ^_^
 
Last edited:
RaistlinZ said:
I hope Anand doesn't throw in an overclocked 580 into his 6970 review, I swear to God.
Click to expand...

The reason behind including GTX460 FTW was its MSRP of $229, which was $10 below HD6870's MSRP, making it a direct price competitor (and as you have seen with HD6870 prices going up, this only made the point stronger).

OTOH, factory pre-overclocked GTX580s cost more than $499. Therefore, it makes no sense at all to compare pre-overclocked 580s, unless of course HD6970's MSRP is also above $499 (but hopefully not!).
 
Last edited:
I am very impressed in their ability to improve ROP efficiency by 2-4x, which means they didn't need to go beyond 32 ROPs. The focus on performance/mm^2 just permeates this presentation. Very impressive if it translates into direct real-world benefits (unlike up to 2x the tessellation of HD6870 over HD5870...)
Click to expand...

^ This... how the hell they managed to get 2x-4x performance out of their ROPs? that seems like a huge leap.

The fact that they made over 10% smaller Shaders, that work more efficent is pretty nice too... but it doesnt sound as impressive as that ROP leap they took.

parallel tessellation engines... meh... that seemed the obvious solution to get closer to nvidias tessellation strength.
 
Arkadrel said:
parallel tessellation engines... meh... that seemed the obvious solution to get closer to nvidias tessellation strength.
Click to expand...

They didn't implement it in the HD68xx series. So this type of setup must take up a lot of extra transistors, thus increasing die space on the same 40nm process. Since Barts was a $239 part, that was probably out of the question if they intended to keep the die size small/yields high for a mainstream part.
 
Last edited:
ArchAngel777 said:
Well, not pride in the sense of arrogance. The reason he told you how to find the article is because he himself cannot link it per the TOS. He is following the rules. As far as I am concerned, BFG is the authority around here when it comes to graphics cards...
Click to expand...

I was not implying arrogance at all.
 
RaistlinZ said:
I hope Anand doesn't throw in an overclocked 580 into his 6970 review, I swear to God.
Click to expand...

Actually I hope that when the gtx560 comes out, Anand compares it to not only reference models, but factory overclocked hd6850 and hd6870's models within the same price range as well. How particular cards performs at a specific price range is ultimately what is most important.
 
RussianSensation said:
Of course there is more room for AMD to make a performance leap. Their die size with HD5870 was much smaller than GF100 and their performance/mm^2 was superior already, while their tessellation/geometry engines were not very advanced. Couple these 2 advantages and sub-part tessellation setup in the 1st DX11 generation with a new revised VLIW4 architecture design with 50% more shaders (480 / 320), and a giant leap in tessellation/geometry performance through parallel tessellation engines, and you can get ~35-50% performance increase.

Don't forget that GTX580 is ~ 35%+ faster than the HD5870. So HD6970 has to be a much greater leap from HD5870 than GTX580 was from the 480.

I am very impressed in their ability to improve ROP efficiency by 2-4x, which means they didn't need to go beyond 32 ROPs. The focus on performance/mm^2 just permeates this presentation. Very impressive if it translates into direct real-world benefits (unlike up to 2x the tessellation of HD6870 over HD5870...)

It would be immensely impressive if HD6970 can defeat the GTX580 since AMD's original "small-die" strategy only positioned dual-GPU designs against 1 monolithic NV core design. Can't wait for the benches. ^_^
Click to expand...


I'm guessing it will not be a clean sweep in the 6970's favor, but I think in the realm of actual game performance, not synthetic benches, the 6970 will have an edge on the 580.

The domain of 2560x1600 30" monitors, eyefinity/surround 3 monitor setups and texture intensive games like Crysis will be the 6970's stomping grounds imo.

Crysis 2 will likely be the next big gpu-bench to be run, depending on how much so called 'extreme' tessellation :whiste: ends up in that game due to TWIMTPB involvement it still may go to the 580.

I believe the 580 will still be faster in those excessive tessellation synthetic benchmarks, but the 6970 will be faster in gaming in general and particularly in high-resolution scenarios.

The Crossfire scaling improvements seen in the 6870/6850 also need to be carried over to the 6970/6950 and of course, the 6990, to really make this a good release.
 
tviceman said:
Actually I hope that when the gtx560 comes out, Anand compares it to not only reference models, but factory overclocked hd6850 and hd6870's models within the same price range as well. How particular cards performs at a specific price range is ultimately what is most important.
Click to expand...

the gtx 560 will be faster then the 6870 without a factory overclock.
 
RaistlinZ said:
I hope Anand doesn't throw in an overclocked 580 into his 6970 review, I swear to God.
Click to expand...

Its not the end of the world, holy crap. Although, I do like how much more organized sites that stick to stock only cards are. I think every launched product should have a follow up article discussing the impacts of overclocking different parts of that specific architecture.

I get frustrated on here the people who say: just overclock X card and now its the same thing as XX card. No, that doesn't work in the real world. If you are going to overclock one card you can do the other one. It seems as if your post is implying that you cannot limit yourself to only comparing stock cards.

Because I'm such a nice guy, if Anand does include an overclocked card, I will edit it out of the graph, just for you, so you don't have to freak out.
 
happy medium said:
the gtx 560 will be faster then the 6870 without a factory overclock.
Click to expand...

I believe so too (although speculation, it seems like the most likely outcome), amd will have to countery by selling their 68xx cards cheaper.

@Ben90

I hope we see stock vs stock... oc cards should have their own reviews vs stock to show their improvements over their own stock card. They should not be in stock card launch's.
 
BFG10K said:
I wrote a piece that describes it in detail, but I can't link to my own articles.

Google GT200 anti-aliasing investigation - it's the first hit.
Click to expand...

fortunately, I CAN link to it:

http://alienbabeltech.com/main/?p=10638

nice article btw.

edit: drat, somebody already beat me to it!

ArchAngel777 said:
Well, not pride in the sense of arrogance. The reason he told you how to find the article is because he himself cannot link it per the TOS. He is following the rules. As far as I am concerned, BFG is the authority around here when it comes to graphics cards...
Click to expand...

agreed. ABT was smart to grab him before AT started hiring more people.
 
Last edited:
Grooveriding said:
There is a lot of talk on Chiphell about very sparse concrete data on performance and it's rumored AMD is dead-set on nvidia not getting any specifics about 6970/6950 performance.

Chiphell also has some rumors that nvidia wants to release an overclocked 580 at 6970's release. As I said before, 6970 is likely going to overall be faster than a 580.

It's worth noting when debating 580's performance, the 580 is much like what the GTX260 216core was to the GTX260, with the addition of improved thermals. Clock for clock the 580 is 6% faster than a GTX 480, which is not surprising as it is a 480 with the full 512 shaders available which gives it that 6% lead.

Whereas from rumors and some of these leaked specs the 6970 has some major changes, new additions and is a bigger die compared to the 5870. I think there is more room for a bigger performance leap than what nvidia could do with what is basically a full shader count GTX 480 in the 580.

AMD also has to contend with NV managing to get out a dual-gpu card, they could pull that move around 6970's launch and prior to 6990's launch. NV is doing a big push to try and get themselves back the halo crown and it looks like AMD is trying to counter their efforts by keeping solid performance data on their new parts out of nvidia's hands.
Click to expand...

would be interesting if they're just holding out instead of struggling. certainly possible in light of the whole "460 ftw" fiasco that AT helped perpetuate when 68x0 launched.
 
Ben90 said:
Its not the end of the world, holy crap. Although, I do like how much more organized sites that stick to stock only cards are. I think every launched product should have a follow up article discussing the impacts of overclocking different parts of that specific architecture.

I get frustrated on here the people who say: just overclock X card and now its the same thing as XX card. No, that doesn't work in the real world. If you are going to overclock one card you can do the other one. It seems as if your post is implying that you cannot limit yourself to only comparing stock cards.

Because I'm such a nice guy, if Anand does include an overclocked card, I will edit it out of the graph, just for you, so you don't have to freak out.
Click to expand...

It's definitely a hot topic, "include or not include OC cards", and in CPU reviews it is done all the time but not so much met with the same degree of "community" backlash as it were...

At any rate I'm one of those folks that could not care less...I'm fully capable of reading graphs, plucking out the data that is relevant to myself, and making conclusions based on data accumulated across multiple reviews, so if a OC'ed 580 is present in a 69xx review my panties are most certainly not going to be getting wadded up if you know what I mean.

That said, when I am in a rush to skim through a review and I want to just nab that superficial "so basically this is what it means..." message from a CPU review I do absolutely hate it when the graphs are all chock-full of 5 or 6 different OC configs that are not communicating to me what the headlining product is doing.

I think it was tom's that use to do this the worst when it came to CPU reviews, trying to find the relevant CPU SKU for the review in the backdrop of their 50-CPU tall bar graphs was mind-numbingly dull.

So I can appreciate the sentiment behind why some people want to see no OC cards, on the other hand I have been there where it just didn't matter to me. This difference is subtle, and usually involves consideration of how empty my margarita bucket is at the time that I am reading the reviews 🙂 😛
 
RussianSensation said:
The reason behind including GTX460 FTW was its MSRP of $229, which was $10 below HD6870's MSRP, making it a direct price competitor (and as you have seen with HD6870 prices going up, this only made the point stronger).

OTOH, factory pre-overclocked GTX580s cost more than $499. Therefore, it makes no sense at all to compare pre-overclocked 580s, unless of course HD6970's MSRP is also above $499 (but hopefully not!).
Click to expand...

I'm gonna go out on a limb here and postulate that if 6970 > gtx 580 you will see it launch in the $499+ range. AMD has done a good job of keeping us guessing here. I would have bet a LOT of money a week ago that 6970 would be <gtx 580 and launch at $399. now? I think that anything between +/- 15% of gtx 580 is in play, and price of $399-$599. I still think that preponderance of the evidence points more to <580 and <$499 price point, but it's at least not a slam dunk imho right now.

HOWEVER, if they miss the mid-dec launch date then evidence goes back overwhelmingly to failure.
 
RaistlinZ said:
I hope Anand doesn't throw in an overclocked 580 into his 6970 review, I swear to God.
Click to expand...


I'll echo this.


Looking forward to benchies and evaluation. Though anandtech.com lost a good amount of the value I place in them after the mishandled 6850/6870 review.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top