Friday, December 16th 2011

Radeon HD 7970 Tessellation Performance Figures Surface

Among the bits and pieces (read: slides) of AMD's press presentation that we're getting, a slide that's definitely missing is performance against competitive or previous generation graphics cards across a range of applications/games. Instead, there's a slide detailing tessellation performance improvements of the Radeon HD 7970 over the previous-generation HD 6970. On average, AMD is looking at about 1.5x (50%) improvements in the tests that it run. One has to also take in to account that the HD 7970 is a faster GPU overall, compared to HD 6970, and of course, that these are AMD's figures.


Source: ComputerBase.de
Add your own comment

53 Comments on Radeon HD 7970 Tessellation Performance Figures Surface

#1
pjl321
Little Worried

I am not sure how well nVidia did in Tessellations but normally only the stuff that AMD want to get leaked actually gets leaked, from this i would read that the Tess improvement is 7970's biggest improvement.

I hope i am wrong.
Posted on Reply
#2
dieterd
by: btarunr
One has to also take in to account that the HD 7970 is a faster GPU overall, compared to HD 6970
- and it better be like more than 50% faster because price is +50% :mad:

Summary - this is bad news: AMD PR slides tell that this is only 50% faster in tessellation nothing else PR has to say to us? considering that this is one of AMD famous PR slideis - it is way less then what they say.
Posted on Reply
#3
gorillagarrett
by: dieterd
- and it better be like more than 50% faster because price is +50% :mad:

Summary - this is bad news: AMD PR slides tell that this is only 50% faster in tessellation nothing else PR has to say to us? considering that this is one of AMD famous PR slideis - it is way less then what they say.
Mate, looks like its 2.5x faster in tessellation on average and 60%-80% faster in those games.
Posted on Reply
#4
Completely Bonkers
Agree with pjl321. So AMD has added some additional tessellation units to try and catch up with nV tessellation performance. However, as we know, increased tessellation usually results in lower fps due to all the extra shading work that is needed for all those extra vertices.

So the "selective leak" focusses on raw tessellation performance only for the WOW factor, because fps performance is lackluster.

Well, if they get similar real world games performance as the 6xxx series but at half the power consumption on the new fab, then it's still a major improvement.
Posted on Reply
#5
air_ii
by: btarunr
On average, AMD is looking at about 1.5x (50%) improvements in the tests that it run.
How did you come up with an average of 50% based on that chart, if all results are above 50% ;)?
Posted on Reply
#6
nt300
This is great news. Tessellation Performance looks good for HD 7970 over the HD 6970. The closer to launch the more we learn about HD 7970s true performance. Now give us some gaming benchmarks because I can easily see 50% or more faster than existing HD 6970.
Posted on Reply
#7
Zubasa
by: air_ii
How did you come up with an average of 50% based on that chart, if all results are above 50% ;)?
Because AMD's fail marketing tell's you its 50%+ so its most likely its less than 50% :laugh:
Posted on Reply
#9
entropy13
by: pjl321
I am not sure how well nVidia did in Tessellations but normally only the stuff that AMD want to get leaked actually gets leaked, from this i would read that the Tess improvement is 7970's biggest improvement.

I hope i am wrong.
If it's all about tessellation alone:



That's current gen cards.
Posted on Reply
#11
entropy13
by: gorillagarrett
Honestly mate, i wouldn't trust any benchmarking utility out there.The 6970 comes pretty close to the gtx580 in the Unigine heaven benchmark, which is a pretty heavy tessellation benchmark.We're talking about millions of polygon per frame here:

http://tpucdn.com/reviews/Palit/GTX_560_Ti_Twin_Light_Turbo/images/heaven_1920_1200.gif
Honestly mate, Unigine is a benchmark too. But you make a sweeping statement about not trusting any benchmarking utility out there, mate.

So what is it really mate? And besides mate, if you further check on Tech Report's revisions on their methodology, FPS isn't exactly the ultimate factor, mate.
Posted on Reply
#12
gorillagarrett
by: entropy13
Honestly mate, Unigine is a benchmark too. But you make a sweeping statement about not trusting any benchmarking utility out there, mate.

So what is it really mate? And besides mate, if you further check on Tech Report's revisions on their methodology, FPS isn't exactly the ultimate factor, mate.
I have not heard or seen this Tessmark utility used anywhere.The Unigine heaven benchmark has been long being supported by both vendors in their drivers, so it's a more standard and supported benchmark than this Tessmark is.

Also, I'd rather wait for the reviews next week than argue these performance numbers...
Posted on Reply
#13
buggalugs
Looks pretty good to me. The tesselation performance jump from the 5870-6970 was much less than that.
Posted on Reply
#14
Damn_Smooth
I don't care about these marketing slides at all. I want these cards to be good simply so I don't have to wait a year for Nvidia. Unless that's changed. If Nvidia can release anywhere near the same time frame, I will go with whichever has the best price/performance. I can't wait for W1zz's review.
Posted on Reply
#15
Bundy
It's great news that news cards are coming. I would not believe any of the current marketing though.
Posted on Reply
#18
Benetanegia
by: air_ii
OBR has been known to spread fud for years...
Yeah that's what people said about Bulldozer too and everything turned out to be true. Not saying it's going to be true with HD7970, but specs do not suggest more than a 30% increase anyway. Then the new architecture can improve performance per SP a bit, but the new arrangement* of cores might also actually degrade performance a bit, it's always posible.

* In Cayman 16 shaders (4d VLIW) were directly inside a single SIMD unit and the SIMD unit was directly connected to the uncore.
Now it's shaders (1d) inside a SIMD, inside a CU, connected to uncore. One more step that could or not degrade performance a few percents. The closest example of such an arrangement is Fermi, and it did suffer a substantial performance degradation from the new modular aproach, compared to the old arrangement. The benefits are very predictable performance increase and that scaling will never suffer as you increase the number of SMs (Nvidia) or CUs (AMD).
Posted on Reply
#19
air_ii
by: Benetanegia
Yeah that's what people said about Bulldozer too and everything turned out to be true. Not saying it's going to be true with HD7970, but specs do not suggest more than a 30% increase anyway. Then the new architecture can improve performance per SP a bit, but the new arrangement* of cores might also actually degrade performance a bit, it's always posible.
That's true (about BD), but if you constantly say something is going to be crap from a certain IHV, you're boud to be right sometimes ;). I'm not saying he's right or wrong, only that he's not a reliable source.
Posted on Reply
#20
Benetanegia
by: air_ii
That's true (about BD), but if you constantly say something is going to be crap from a certain IHV, you're boud to be right sometimes ;). I'm not saying he's right or wrong, only that he's not a reliable source.
I only know him for his very early BD reports. At the time they seemd horrible and people said he was just a troll so I didn't believe them. Then the architecture was completely revealed with all details, I made my calculations, they seemed to support the idea that BD was not as good as it was first being reported. According to my calcs (10-20% faster than 2600k) it wouldn't be as bad as he was saying, but the huge difference between the hype and his numbers started to shrink. According to the specs and how resources were arranged and shared, the efficiency of each resource and IPC needed to increase by a lot in order to increase performance over previous gen. It didn't so in the end he was fully right.

The pattern is similar this time, so I'm afraid that it could actually be true. This time is the reverse tho, I've known the architecture for a long time, but without knowing the specs I could not make any assuption regarding final performance, then the specs came in, then his performance numbers, but the effect is the same.

With 2048 SPs you really need them SPs to be a lot faster/efficient than they were if you want the card to be significantly faster. It is fairly posible to have much faster SPs, but like with BD we again depend on a net "IPC" increase in order to get a significant performance gain. Since I don't want to believe in fairies again, I'm counting on SPs being fairly equal hence with 30% more SPs and 30% more TMUs and similar clocks, 30% performance increase is kinda the maximum I would expect. According to OBR numbers, net "IPC" or perf-per-shader might actually be slightly down and wihle it was not something I would have expected, it's not really something completely imposible.
Posted on Reply
#21
dir_d
We also dont know if the CUs are locked to clock speed or they went with a more NVIDIA approach and sped them up. So there is a slither of hope, very slim though.
Posted on Reply
#22
Wrigleyvillain
PTFO or GTFO
by: air_ii
That's true (about BD), but if you constantly say something is going to be crap from a certain IHV, you're boud to be right sometimes ;). I'm not saying he's right or wrong, only that he's not a reliable source.
Even a broken clock is right twice a day.
Posted on Reply
#23
Steevo
Who cares about tessellation performance when the original tests on many things were done with tessellation being performed on surfaces that were not in the frame. I care more about the triangle/geometry performance than I do the ability to break them down with tessellation, you can't tessellate what you haven't rendered yet.
Posted on Reply
#24
HalfAHertz
Meh... enough benchmarks. Show us some tessellated boobies!
Posted on Reply
#25
Benetanegia
by: dir_d
We also dont know if the CUs are locked to clock speed or they went with a more NVIDIA approach and sped them up.
They are still locked if we are to believe the rumors.

Again I don't expect shaders themselves to be very different from what they were.
So there is a slither of hope, very slim though.
Yeah, but I wouldn't count on that. I'd rather be surprised than dissapointed. Again.

And the complete lack of any performance slides other than tesselation performance is not good really.
Posted on Reply
Add your own comment