AMD: R9xx Speculation

[Entropy]

http://unigine.com/clients/
i see 6 games (1 released) and twice more non-game applications build with their engine.
How much applications were built using 3dmark's engine?
And their tests are exactly that - well-looking tests, no way to predict which way future development will go.

I never said Futuremarks tests were better - in fact, at the time of their introductions, they have been notoriously bad at mapping to then current application performance. In time, some of them proved to be more useful than others, but by that time newer, shinier versions have been introduced and are used by most reviewers and e-penis measurers.

I never saw a good analysis of just how successful the Futuremark tests have been at being "forward looking". It would be a prety difficult study to do well. The saving grace of game benchmarking is that any tests or game that use a reasonably balanced set of resources, will predict the performance of any other reasonably programmed game fairly well because GPU resources tend to scale together.

The reason the unigine benchmarks fail in spite of this, is that they explicitly emphasize one particular feature (tesselation) way out of proportion to its actual use.

 

[no-X]

I think many versions of 3D Mark were balanced to give similar score on directly-competing hardware irrespective of its architeture or real qualities. This was the case of 3DM 2006, FP16 filtering, X1900 vs. 7900GT and 2900 vs. 8800 later.

 

[Mize]

Okay, nvidia's cards on now on the table. They have "called" AMD.

Whatcha got AMD?

 

[Entropy]

Addendum.
The above comments were regarding MadOnion/FutureMark game tests. To their credit, along with the game tests, they have also provided more targeted feature tests, which can be be both interesting and useful. For a long time, I got better application performance predictions out of the pure fill-rate tests than the game tests.

 

[DavidGraham]

On the other hand, we all know that AMD is planning to unveil its high-end Cayman design before the end of the year—and there’s not a ton of time between then and now. Regardless of how Cayman performs, it’s a fairly safe bet that prices will shift to reflect the relationship between that that card and this one. There's a twist, though. Back when the Radeon HD 6870/6850 launched, we were given a date for the Cayman debut. As we inch closer to it and nobody anywhere knows anything about it (board partners haven't seen cards, system builders are still in the dark), it begins to look like there may be delays. From what I've heard, AMD won't even be briefing its partners until after that original embargo date passes.

http://www.tomshardware.com/reviews/geforce-gtx-580-gf110-geforce-gtx-480,2781-16.html

 

[Squilliam]

@Mize: Cayman is broken, hot and unfixable. Well maybe not, but it would be really funny if Charlie did an article like this one for Cayman. Maybe he can talk about how they don't have the experience to produce large chips or something of the sort.

 

[ECH]

Okay, nvidia's cards on now on the table. They have "called" AMD.

Whatcha got AMD?

This reminds me of the 2900XT "supposed" release. I hope this isn't the case again.

 

[Mize]

@Mize: Cayman is broken, hot and unfixable. Well maybe not, but it would be really funny if Charlie did an article like this one for Cayman. Maybe he can talk about how they don't have the experience to produce large chips or something of the sort.

That would be funskies. Although we know from Chiphell that people do have Caymans so Tom's is at least wrong on that point. AMD has been amazingly good at keeping leaks down of late...much to my disappointment.

 

[Topman]

@Mize: Cayman is broken, hot and unfixable. Well maybe not, but it would be really funny if Charlie did an article like this one for Cayman. Maybe he can talk about how they don't have the experience to produce large chips or something of the sort.

Why not a 'B' plan ???

Look the Techreport review. A dual Barts can compete with GTX580 (in performance, price, power consumption, etc).
It's not hard. AMD has several dual-GPU card designs.

 

[DavidGraham]

Regarding the debate around Civ 5 earlier , it looks like GTX 480/580 do get faster as resolution increases :

Finally, those GTX 480 and 580 numbers aren’t wrong – it really does get faster at higher resolutions for reasons we can only assume are due to the triangles getting larger and easier to rasterize and/or cull.

http://www.anandtech.com/show/4008/nvidias-geforce-gtx-580/10

So it is true , (as postulated by Mr.Alex) , and the reason for it sounds logical too.

 

[rpg.314]

http://www.tomshardware.com/reviews/geforce-gtx-580-gf110-geforce-gtx-480,2781-16.html

If there is any truth to that, then I'd say AMD made an excellent call in deciding to go for a semi-refresh in Barts.

 

[ferro]

If there is any truth to that, then I'd say AMD made an excellent call in deciding to go for a semi-refresh in Barts.

5850 and 5870 were rendered obsolete by Barts at a much lower price point. AMD needs to follow up quickly with new and improved performance parts.

 

[Nihilist]

5850 and 5870 were rendered obsolete by Barts at a much lower price point. AMD needs to follow up quickly with new and improved performance parts.

From where I'm from, the 5850 is selling for the same price as the 6850 and is both slightly faster in the majority of titles and is more power efficient as well (if ever so slightly). Being unimpressed by the tessellation speed and AF improvements convinced me yesterday that a 5850 is at the moment a better purchase.

 

[ferro]

From where I'm from, the 5850 is selling for the same price as the 6850...

Did 5850 prices drop when Barts was launched? That's what happens with obsolete products. It doesn't mean that a 5850 is a bad choice at 6850 prices of course. Au contraire.

Until Cayman launches AMD is missing performance products in the €250 - €350 price range though.

 

[ECH]

We had an opportunity to interview Stanley Ossias, Director, Mobile Discreet Graphics Product Management at AMD right about the time that the GTX 580 NDA ended. Basically the message was that AMD expected this improved version of Fermi and is completely set to counter it now and with the further release of HD 6000 series. They feel that they have the right product and the right price point and that competition is good. We will have more later on and in a separate published interview. Stay tuned as this graphics war gets more and more exciting and ABT is reporting from right on the battlefront’s front lines.

source (at the bottom/conclusion)

I can't make heads or tails of the comment made. Or is that the gist of the comment? In any case I'm left wondering:
-When will AMD counter?
-Are they confident they have a better product?
-Why isn't the message more reassuring?
-Why didn't they announce a date when they will showcase their cards?

Perhaps there is more to the story then what was shown but it does leave more questions than answers for me.

 

[Picao84]

source (at the bottom/conclusion)

I can't make heads or tails of the comment made. Or is that the gist of the comment? In any case I'm left wondering:
-When will AMD counter?
-Are they confident they have a better product?
-Why isn't the message more reassuring?
-Why didn't they announce a date when they will showcase their cards?

Perhaps there is more to the story then what was shown but it does leave more questions than answers for me.

Seriously, since that pointless article from Charlie that sounded like a reassuring that HD6970 is coming, that Im really suspicious of whats going on.. Was that article really needed, if they are quite confident???

 

[Harison]

source (at the bottom/conclusion)

I can't make heads or tails of the comment made. Or is that the gist of the comment? In any case I'm left wondering:
-When will AMD counter?
-Are they confident they have a better product?
-Why isn't the message more reassuring?
-Why didn't they announce a date when they will showcase their cards?

Perhaps there is more to the story then what was shown but it does leave more questions than answers for me.

It was standard PR answer, without any details as usual. In line with AMDs deafening silence about not launched products.

-When will AMD counter? - This year IMO, dont see any reason to believe rumors about postponed launch.

-Are they confident they have a better product? - From the response it looked more like Cayman would be slightly slower, slightly cheaper, hence not "the faster card", but "the right product and the right price point".

-Why isn't the message more reassuring? - I doubt that guy was authorized to spill the beans, nor its AMD policy lately to give the details before launch.

-Why didn't they announce a date when they will showcase their cards? - Its coming, my guess AMD is testing GTX580 to the bits to see what last second tweaks they have to do, along with 24/7 working drivers team. Was Nov. 22 official or rumored date? If official, thats when we'll know more, if not everything.

 

[Gipsel]

Regarding the debate around Civ 5 earlier , it looks like GTX 480/580 do get faster as resolution increases :

http://www.anandtech.com/show/4008/nvidias-geforce-gtx-580/10

So it is true , (as postulated by Mr.Alex) , and the reason for it sounds logical too.

It may sound logical to you but it simply isn't as I already explained back then.

 

[Harison]

Seriously, since that pointless article from Charlie that sounded like a reassuring that HD6970 is coming, that Im really suspicious of whats going on.. Was that article really needed, if they are quite confident???

Why do you think AMD was behind that article? Charlie quoted his sources, thats it. Its good that AMD learned how to keep secrets incredibly well, but in their place I would "leak" few benchmarks to spoil GTX580 launch :smile:

 

[DavidGraham]

It may sound logical to you but it simply isn't as I already explained back then.

Ok , so what's your take on this ? it is obviously not a test bug , a driver bug maybe ?