AMD: R9xx Speculation

Don't worry, be happy. This is simply because of the well organised mess served by AMD. But nothing really special- a lot of rumours fighting which one is the corect one. But as far as I know we expect Barts ala 6700 series to be launched later this month. Perhaps there was a rumour that 6800 series will be launched, but there is also one that something is delayed. Total mess.....

 

Wasn't there some version of that 6700 slide with more at the bottom?

 

hang on, youve taken two seperate issues and turned them in to one point. to address the first paragraph: dont come on to the 'AMD: R9xx Speculation' thread on beyond3d.com and moan that checking bench marks for performance on graphics cards is a waste of time. thats just nonsensical, (not saying you did by the way) I dont think thats snobery, I just think your complaining to the wrong audience. Your point would be lost here. If you dont like spending minutes/hours on understanding tech and resulting purchases then this place prob aint the right place.


To address second paragraph, I whole heartedly agree that abuse of name rebadging to give the perception the product is next gen when it isnt is bordering on fraudulant. I think thats is total foul play. And the general public dont deserve to be conned like that.

 

http://vr-zone.com/articles/amd-radeon-hd-6000-series-in-trouble-/9999.html

It might be Cayman is 6800 series GPU, and after all non-sense discussion here about Barts being 6800 series.....

 

I'm actually not sure that slide is about 6800 at all. Here's a question for you:
If it is indeed for 6800, why isn't it in the title? It has multiple reference to 6800 series but neither of them actually refers to an actual part.
From what I understand, it might as well be a 6700 series paper, with architectural/performance references to 6800 series because they are the same and 6800 won't be available for test until later date.
You CAN actually write about Cayman (Cypress) and then just say Barts (Juniper), the product in discussion here, is just 2/3 (1/2) of what's just described.
In fact almost ALL Juniper/GF104 articles are referring to Cypress/GF100 to explain the "benefits" other than real performance.

So, it adds up to:
a)Unless you can show us what's been cut off from that page, I don't think it's convincing, not enough any way.
b) there is no reference about Barts or Cayman anywhere on that page. Not even NI.
c) And the whole Barts before Cayman is pure rumor anyway.

Until you can convince me and many others that ALL three of those concerns are nonsense. I will not directly associate 6800 with Barts, and I will deem Cayman more appropriate for 6800 series. -- That means I think Cayman *should* be named 6800 series by convention, not necessarily that I think it *will* definitely happen.

Edit: Look, all I'm saying is that there is no conclusive evidence that Barts will be called 6800 series or otherwise. Please stop assuming that everybody must agree with you on that matter.
Many people would deem marketing a 6800 part under $300 (which is exactly what Barts is gunning for) to be a mistake, that doesn't mean AMD won't try doing exactly that.

 

Whats the mistake in a 6800 part selling under $300 (actualy i think everyone would wish a 6850 to be sold for $200 )
So you want to say that so many people are completly insane.

 

All these cards, are X870 parts, meant to declare a product's status in its respective series, in which case, the X870 is used for the higher end single gpu product. In that respect, they are all products that are meant for a specific audience.

You do have a point doing the 3870-5770 association, but my main argument here was the exponential increase of the prices. Apart from the obvious die size differences of the rv670 with Cypress, the rest of the card's costs, do not justify the resulting price of the 5870 compared to 3870.

I mean they basically have the same bit bus width, hence similar PCB complexity, the 3870's 55nm and GDDR4 were as cutting edge as the 5870's 40nm and GDDR5, so the only main differences, manufacturing wise, would be a few more grams of copper/aluminum for the heatsink, a more robust VRM and the actual main chip itself. Do these things justify a 150 euros price increase from the 3870 to the 5870?

 

Yes they do, because at this end of the spectrum the market is paying for relative performance, not a straight BOM with no margin. The name indicates a relative positioning within AMD's own stack, not a pricing model.

 

Thank you good sir.

That's is perfectly understandable. So all in all, AMD is making more money from the 5870 than it did with the 3870.

I only wish they could settle for less (same goes for NV), but I guess it is me who will have to settle.

 

racca: I'm not trying to convince anybody. I have no reason for it. I just asked a few persons, why are they believe in the rumours, which I have a reason (or two) to consider unreliable.

One of the reason is quite simple. All the three slides which were presenting Barts as HD6700, were fakes. It means, that their authors had no acces to real info, or their target was to confuse people (= spread purposely false info). So even the information, that Barts is HD6700 cannot be considered as a fact. Next set of (semi-censored) slides were leaked by OBR. Those looked quite persuasively. Please notice, that the previous slides were conceived as ATi's press presentations. But ATi's press presentations are created typically 8-12 days before launch. They can't be created earlier, because clocks aren't finalized at that time (usually). So all the presentations, which were "leaked" month before launch and which are styled like the press materials, are very likely fakes. On the other hand, the one, which was leaked by OBR, was styled like typical internal document. That's the first reason why I consider it to be real. The second reason is the list of features. OBR leaked (partially censored) list of feature-set - the uncensored parts were identical to the list, which was leaked by another source (expreview?) about a week ago. That's the second reason why I believe it. OBR also said, that according to the presentation there's not any Barts-based board, which would be called HD6700. And finally, OBR removed the article - reputedly because he won't to ruin relations with AMD.

I'm definitely not an OBR's fan. But I must admit, that at this moment I consider him as the most appealing public source of info.

 

Everyone still believing that Barts is 6750/6770, please leave the topic!

 

Its me! But where would I go?

Of course among all these fakes and extremely tight-lipped AMD we still have no idea which parts are which, hell, even generations NI/SI themselves were spread as misinformation just a month ago. Still IMO Barts are 6700, and Caymans 6800, and I will stay this opinion until actual data appears.

 

Sh*t.. what happened to the time people believed what I said?.. or at least took it as somewhat credible.


okay.. there's more things that separate the 6850 and 6870 than just clocks etc.

Heck, you have to get the whole N.I. = Evergreen .. out of your head..

 

This is thread is a proof of how well AMD misinformed all the people. Nobody believes nobody. Good job AMD.

 

Dude... a NUMBER does not make a PRODUCT.

 

No need to misinform, it's just the nature of The Internet. If there is no real information, The Internet will make stuff up.

 

He's just sad he has to leave...

 

6770 & 6850?
Now that will be evil... and dumb

 

Sad man making typos is sad.