AMD Radeon HD 6970 already benchmarked? Enough to beat GTX480 in Tesselation?

[darkswordsman17]

Resolution and tesselation levels are the same, according to this AA settings also match:
http://www.hexus.net/content/item.php?item=27053&page=10

And the CPU is even more beefy in that review:
http://www.hexus.net/content/item.php?item=27053&page=8

I think it's safe to say that just plugging those numbers into a chart dosn't give any clue to actual performance.

But I already see posts on forums claming that the charts shows 6970 beating the GTX480...I'd would be carefull making such a claim right now....if the chart is the "source".

Wait, did Hexus just take the results from the other person and plug it into their own chart? At least they did mention that the person claimed to have benched a 480 and 5870 on the same system and the results they got, but...why plug the one card into your own chart?

In short, the chart is stupid, but the numbers may or may not be. If they aren't, then the 6970 will be impressive indeed. If not, well then we'll see what they did manage to accomplish. Even just matching the 480 would be quite impressive if they can keep the rest of the performance good, along with the relatively low power consumption.

 

[Spyhawk]

You completely contradicted yourself.


Why does AMD need to beef up its tesselator if no games on the market (are) able to use it? Wouldn't they be wasting silicon as well?

What Im trying to get at is that these gpus arnt thought up and made in months. Years of planning go into them. I think its around 3 years. So AMD and NV pretty much decided around 3 years ago how much tesselation performance was gonna be needed for games today.

Now tell me this. Is there a DX11 game atm that runs poorly on AMD hardware due to insufficient tesselation capabilites ? The answer is no so you see in the end, AMD wasnt wrong in its approach on tesselation.

 

[Lonbjerg]

Wait, did Hexus just take the results from the other person and plug it into their own chart? At least they did mention that the person claimed to have benched a 480 and 5870 on the same system and the results they got, but...why plug the one card into your own chart?

In short, the chart is stupid, but the numbers may or may not be. If they aren't, then the 6970 will be impressive indeed. If not, well then we'll see what they did manage to accomplish. Even just matching the 480 would be quite impressive if they can keep the rest of the performance good, along with the relatively low power consumption.

Page hits I presume as the tessellation performance on AMD vs NVIDIA is a hot topic right now, ever since AMD's whine.

 

[Arkadrel]

What Im trying to get at is that these gpus arnt thought up and made in months. Years of planning go into them. I think its around 3 years. So AMD and NV pretty much decided around 3 years ago how much tesselation performance was gonna be needed for games today.

Now tell me this. Is there a DX11 game atm that runs poorly on AMD hardware due to insufficient tesselation capabilites ? The answer is no so you see in the end, AMD wasnt wrong in its approach on tesselation.

epic post +1 interwebs to you.

 

[Lonbjerg]

What Im trying to get at is that these gpus arnt thought up and made in months. Years of planning go into them. I think its around 3 years. So AMD and NV pretty much decided around 3 years ago how much tesselation performance was gonna be needed for games today.

Now tell me this. Is there a DX11 game atm that runs poorly on AMD hardware due to insufficient tesselation capabilites ? The answer is no so you see in the end, AMD wasnt wrong in its approach on tesselation.

Lets revisit that thought in 6 months when more DX11 games have come out.

And no matter how it's spun, having AMD's implementation being so much slower at higher tessellation levels (where tesselation makes the most sense) is not a catalyst for progress, infact it's the oppiste.
It will force devleopers to either use lower levels of tesselations i ngames (because AMD hardware runs poorly with it)...or introduce low/high tesselation modes in their games in order to achieve the desired levels of performance.

The DX11 standard goes up to 64x tessellation, making hardware that can only run "mildy" good at 1-11x tesseallation is really setting the bar way to low.

It now more looks like a checkbox feature rather than a good implementaion.
And AMD knows this...otherwise they wouldn't spend their PR resources at whining over "too much tesselation"...when infact it's part of the DX11 standard.

 

[brybir]

Lets revisit that thought in 6 months when more DX11 games have come out.

And no matter how it's spun, having AMD's implementation being so much slower at higher tessellation levels (where tesselation makes the most sense) is not a catalyst for progress, infact it's the oppiste.
It will force devleopers to either use lower levels of tesselations i ngames (because AMD hardware runs poorly with it)...or introduce low/high tesselation modes in their games in order to achieve the desired levels of performance.

The DX11 standard goes up to 64x tessellation, making hardware that can only run "mildy" good at 1-11x tesseallation is really setting the bar way to low.

It now more looks like a checkbox feature rather than a good implementaion.
And AMD knows this...otherwise they wouldn't spend their PR resources at whining over "too much tesselation"...when infact it's part of the DX11 standard.

I think the point was made a few posts ago, but what you and everyone else seems to be missing out of this is the time period we are talking when it comes to making decisions of "how much and of what" in developing video cards.

Read Richard Huddy's statements on Anandtech, the design for Cyprus was finalized in 2008, at least a year before it was ever released. So, AMD, and really all chip companies pushing technology boundaries, are taking educated guesses on what is going to be coming in the future. So, they thought "we could use 6 units of tessellation because we think its the best when all things considered for our company", sometimes they will be right and sometimes they will be wrong. It happens, and it happens to everyone. It even happens to everyone in this forum in their own lives. This question becomes even more complex in a chicken/egg scenario with developers and hardware vendors all collectively trying to "guess" what the future holds.

Nice thing about it is that no matter how much complaining we all do here, the process actually does more the entire industry forward over time, sometimes it just takes longer than others.

 

[Lonbjerg]

I think the point was made a few posts ago, but what you and everyone else seems to be missing out of this is the time period we are talking when it comes to making decisions of "how much and of what" in developing video cards.

Read Richard Huddy's statements on Anandtech, the design for Cyprus was finalized in 2008, at least a year before it was ever released. So, AMD, and really all chip companies pushing technology boundaries, are taking educated guesses on what is going to be coming in the future. So, they thought "we could use 6 units of tessellation because we think its the best when all things considered for our company", sometimes they will be right and sometimes they will be wrong. It happens, and it happens to everyone. It even happens to everyone in this forum in their own lives. This question becomes even more complex in a chicken/egg scenario with developers and hardware vendors all collectively trying to "guess" what the future holds.

Nice thing about it is that no matter how much complaining we all do here, the process actually does more the entire industry forward over time, sometimes it just takes longer than others.

Actually AMD has no excuse to botch up their tessellation engine:
http://en.wikipedia.org/wiki/TruForm
They have had hardware tesselllation since the Radeon 8500.
They should have wiped the floor with NVIDIA's performance in tessellation, but they made it to be the other way around.


I suspect this is another area where AMD's small die philosophy once again hit performance/features in a bad way.

 

[shangshang]

What Im trying to get at is that these gpus arnt thought up and made in months. Years of planning go into them. I think its around 3 years. So AMD and NV pretty much decided around 3 years ago how much tesselation performance was gonna be needed for games today.

Now tell me this. Is there a DX11 game atm that runs poorly on AMD hardware due to insufficient tesselation capabilites ? The answer is no so you see in the end, AMD wasnt wrong in its approach on tesselation.

Once developers start using tessellation en masse, then it'll bite AMD in the collective ass.

 

[Scali]

To me you come accross as a hard core AMD/ATI hater....Just seems like most of your posting is anti AMD/ATI

Must be you then.
For your information, I have owned and supported many Radeons, mostly when they offered superior DirectX implementations and/or features, such as the PS1.4 on the Radeon 8500, or the SM2.0 on the Radeon 9700.
I actually bought a Radeon 5770 recently, because AMD was the only one with DX11 hardare at the time.
How is that anti AMD/ATi?

I just support nVidia's great tessellator implementation now, because, just like the above-mentioned Radeons, they currently have a technological advantage.

How is that biased, green shades or anything?

Maybe it was in the way you read it. I didn't see anything offensive in it.

You were giving me the third degree over nothing.

 

[Scali]

Your join date shows that you have been around for years, but I don't recall (maybe I'm wrong here) you being upset with Nvidia for not having any tessellation capability for years while AMD did.

Given the astounding success that was AMD's tessellator, is that any surprise?
I actually did have a Radeon 8500, and did play around with TruForm a bit, but I was underwhelmed by it. I never did much with the tessellation of the later Radeons I owned. Instead I tried to implement my own tessellation via GPGPU. It wasn't until DX11 that tessellation became fully programmable. Now all we needed was decent performance.

Also, as I understand it tessellation was supposed to be in DX10, but because Nvidia did not have that built into their hardware they were able to get Microsoft to push it back to DX11.

Ah good, some rumours to make nVidia look like the bad guy again. Haven't seen that before...

You stand out as particularly vocal about Nvidia's superior tessellation abilities, but I don't recall any posts by you lambasting Nvidia for not supporting it at all or delaying adoption.

No, see above.
But I have lambasted nVidia for various other things, such as having a very poor SM2.0 implementation in their GeForce FX.... for not having DX10.1 and DX11 hardware on time, for cheating in 3DMark, and probably lots of other things that I can't think of right now.

 

[Grooveriding]

What Im trying to get at is that these gpus arnt thought up and made in months. Years of planning go into them. I think its around 3 years. So AMD and NV pretty much decided around 3 years ago how much tesselation performance was gonna be needed for games today.

Now tell me this. Is there a DX11 game atm that runs poorly on AMD hardware due to insufficient tesselation capabilites ? The answer is no so you see in the end, AMD wasnt wrong in its approach on tesselation.

Exactly. And it is widely accepted that the 6970 is going to be faster than a GTX 480 and this card is coming out in less than a month. Given AMD's slide showing their progressing tessellation hardware plan

They are implementing a scalable tessellator in the 6970. The tessellation performance of the 6870 is already sufficient for every game on the market and outperforms the GTX 460 in many games and where it doesn't it's a draw between the two.

With the improved tessellation ability of the 6970 along with it being a faster card than the GTX 480, there is no reason to believe it will not continue where the 6870 left off and then some with being not just the fastest single-gpu card on the market but more than ample to play any DX11 game on the market and play it faster than the GTX 480 will.

 

[blastingcap]

Thanks, makes me feel just that much better about skipping HD68xx which is basically just a refresh of Evergreen. Cayman looks more interesting as it's a new-ish architecture, but unfortunately it will likely be priced over my budget.

Exactly. And it is widely accepted that the 6970 is going to be faster than a GTX 480 and this card is coming out in less than a month. Given AMD's slide showing their progressing tesselalation hardware plan

They are implementing a scalable tessellator in the 6970. The tessellation performance of the 6870 is already sufficient for every game on the market and outperforms the GTX 460 in many games and where it doesn't it's a draw between the two.

With the improved tessellation ability of the 6970 along with it being a faster card than the GTX 480, there is no reason to believe it will not continue where the 6870 left off and then some with being not just the fastest single-gpu card on the market but more than ample to play any DX11 game on the market and play it faster than the GTX 480 will.

 

[Lonbjerg]

I wonder if scalability just means they added one more tesselation engine...the numbers would indicate it.

 

[Skurge]

I wonder if scalability just means they added one more tesselation engine...the numbers would indicate it.

They added more units the 6800 series and they said that, they are saying its scalable now. So its not the same thing. As much as you would want it to be.

 

[AtenRa]

Now tell me this. Is there a DX11 game atm that runs poorly on AMD hardware due to insufficient tesselation capabilites ? The answer is no so you see in the end, AMD wasnt wrong in its approach on tesselation.

With only one year with DX-11 hardware in the market and you jump to conclusion that AMDs approach was the right choice? On the same note, we can conclude by judging the results of CIV-5 and HAWX-2 tessellation performance that AMDs approach was not right and NVs was.

 

[Aristotelian]

Page hits I presume as the tessellation performance on AMD vs NVIDIA is a hot topic right now, ever since AMD's whine.

Where was the whining, exactly? This is the second time I've asked this sort of question and the second time I'm implying that people on this forum need to read articles more carefully.

It has also been a hot topic ever since Nvidia's PR machine->forum presence took up the gauntlet - and that's not a member call out, just a statement of fact.

 

[Skurge]

Its just funny that after the release of barts, the hot topic on these forums is it tessellation power. Not the fact that it brought down prices of other cards, not that it has 3d, not that it can support 6 monitors. Non of the pros, only its single con. I'd say nVidia PR did a great job.

 

[Idontcare]

Its just funny that after the release of barts, the hot topic on these forums is it tessellation power. Not the fact that it brought down prices of other cards, not that it has 3d, not that it can support 6 monitors. Non of the pros, only its single con. I'd say nVidia PR did a great job.

The hot topic is whatever is controversial. AMD's email to editors regarding HAWX 2 provided all the fuel that the fire ever needed.

The other stuff you listed is more of the same stuff talked about over and over again, that no one really felt the need to continue talking about it over and over again is of no surprise.

3D was talked to death before because of the controversy that it was NV only. AMD adding support makes it a non-controversy topic.

Eyefinity was talked to death before because of the controversy over the bezels. Increasing eyefinity monitor support from 3 to 6 might impact all of 10 members here, maybe fewer. Nothing new and controversial here.

Prices dropping was hyped and talked about for at least a month leading up to the release. How much longer you want people to keep reposting their same posts over and over?

Nope, today's controversy is brought to you squarely and firmly by AMD releasing their email to the reviewers combined with the fact that their tessellator engine is indeed lower performing than Nvidias.

And I have a feeling the forum will be talking about it until Cayman comes, maybe even after.

 

[Scali]

The hot topic is whatever is controversial. AMD's email to editors regarding HAWX 2 provided all the fuel that the fire ever needed.

Yup, so it's AMD's PR, and not nVidia's PR... and I would say AMD's PR is NOT doing a good job.

 

[Aristotelian]

Nope, today's controversy is brought to you squarely and firmly by AMD releasing their email to the reviewers combined with the fact that their tessellator engine is indeed lower performing than Nvidias.

Right. And AMD had no reason to send that email out that was directly caused by Nvidia. It is simply that "AMD's tessellation units aren't as strong as Nvidia's" in isolation that caused it. That's exactly what everyone believes in this scenario. Where's the 'eye rolling' emoticon?

Today's controversy is brought to you, in part, by 'the way it's meant to be portrayed' (TWIMTBP) and the correlated forum presence.

Boss: "Gaming benchmarks on the 6800 are fine? Crossfire is fine? Uh oh! Run the tessellation machine!"

Underling: "But boss, where are the tessellation games...the gaming benchmarks are fine in both camps?!"

Boss: "Forget the games kid! (stamps out his cigarette) Run Unigine! Run some synthetic tessellation only benchmarks! That we beat 'em in something will stop the cards from flying off the shelves!"

Underling: "Yes Sir!"

Underling runs to different forums: "Hey guys! Seen this latest benchmark that has little impact on how games in general are played right now? Give it a long look! (Damn I better hurry up, the 6800 has just been released, gotta stop the flow of the 9:1 DX11 marketshare holder!)" Victory Green Team! Yay!!!!

Underling: "Oh wait, the 6850 actually is pretty good bang for the buck....hrm, better not let my boss see that I run AMD"

Blah blah blah blah, I see through it. I'm not the only one.

 

[Scali]

Except nVidia's PR has not commented on anything. It's all been AMD's PR, trying to explain why certain DX11 benchmarks shouldn't be run on their DX11 hardware, with lots of colourful slides, interviews with Huddy taking digs at nVidia etc.

 

[Genx87]

Tesselation is a major component of DX11. So is GPGPU. Why is it AMDs DX11 chips arent good performers compared to the competition in 2 of 3 major selling points of DX11? What purpose does it serve to put DX11 bullet point on the chip if it isnt great at performing that role? Feels like a DX9\10 chip with a DX11 checkbox.

 

[n0x1ous]

Its just funny that after the release of barts, the hot topic on these forums is it tessellation power. Not the fact that it brought down prices of other cards, not that it has 3d, not that it can support 6 monitors. Non of the pros, only its single con. I'd say nVidia PR did a great job.

Barts is great for the most part, but the 3D kinda blows I think. People already think of it as a niche feature with Nvidia's system and that has the whole package ready to go out of the box. With AMD's you are on your own to make it all work. If Nvidia's 3D is a niche I don't know what the word would be for AMD's....

 

[Scali]

If Nvidia's 3D is a niche I don't know what the word would be for AMD's....

I dunno, checkbox feature seems appropriate, yet again?

 

[badb0y]

Tesselation is a major component of DX11. So is GPGPU. Why is it AMDs DX11 chips arent good performers compared to the competition in 2 of 3 major selling points of DX11? What purpose does it serve to put DX11 bullet point on the chip if it isnt great at performing that role? Feels like a DX9\10 chip with a DX11 checkbox.

Wrong, Almost every game that uses DX11 runs fine on the Radeons, by the time a game comes out using a lot of tessellation power these cards will have run their course.

Benchmarks only serve to what a card is cabable of, not how it will perform in games and don't even try to bring up that HAWX benchmark since even the 5770 averaged 40 fps LOL it's so demanding man.