AMD: Southern Islands (7*** series) Speculation/ Rumour Thread

They're not limited to reference designs, and most of them go beyond.

Most of the marketing is centered around the specific features and properties of a partner's card: the bundle (games, cables and adapters, software), the cooling system, the custom board design, the added memory, the dual-bios, the custom video connectors, the high-quality capacitors, and exotic stuff like dual-GPU boards using mainstream GPUs, low-profile cards, passively or water cooled models, etc. Most of these things don't apply to CPUs.

Besides, discrete desktop graphics cards are often bought directly by consumers, especially high-end cards, while CPUs are mostly bought by OEMs, on whom the kind of marketing provided by partners would be lost.

No it can't, or certainly not to the extent that partners do it. AMD doesn't have unlimited resources. In fact they've recently fired 10% of their workforce.

 

One of AMD's branding issue regarding GPU is that some big partners are most of the time allowed to use the marketing/communication money to offer price cuts. It kills smaller players and does nothing to improve brand perception. They need to change that but it won't be easy

 

I can be wrong, but i have the feeling after the initial first months who follow a release, there's surely more AIB specific cards who are sold, instead of the reference one. ( remember the blue pcb ones and version2 ( who was offtly not as good of the reference one ) .. With all the different models who are released and a certain competition between AIB. ( with better cooler, clock speed, or OC specific features, sometimes just cheaper )

I have allways buy reference boards, Like that im sure they have the best power, and ram parts. Even if they dont come with the best coolers ( sound wise ). Personally this make 10years all my hardware have been watercooled, so i couldnt care less about the fan sound.

The only non reference i will buy are cards like the MSI Lightning and why not the last Asus DirectCUII products...

 

This is not brand building.

 

A $549 HD 7970, coming from a $370 HD 6970 is brand building?

Of course there are some serious problems with the costs of manufacturing in the latest process technologies, but that could be communicated to the customers, who wonder about these rising launch prices in AMDs product families.

 

I think they already tried to price Cayman to match GTX 400-series, but 500-series forced their hand. They want to make people think that these are worth the high price. AMD's "crown jewels" like was mentioned in their analysts day. I'm pretty certain that the manufacturing side does not explain the increase. They now have very strong market availability across their entire 7000-series range. Curious to see how things unfold. I haven't been a fan of the pricing, but I think that 7850 and 7750 are good value and positioned attractively.

 

7850 starts to shine when OC'd, that's how I always look at these, because I would OC them. I'm not a huge fan of the 7770, it looked quite bad compared to the outgoing models unlike the 7850. However I haven't really looked at retail 7850s and what type of boards and possible OC room/restrictions they come with. The launch reviews said that the retail boards will be simpler/poorer than the 7870 board, which was used for the review units. I'd think there are some decent custom versions for that card also?

 

Uh no. Building a brand is about creating extra value within the name. You do want a perception of value, but that's not the same as just being cheaper. Quality, performance, reliability, prestige are things that build brand value. How does selling similar products at a significant discount achieve any of that?

 

If the brand you want to create is 'Rolecks' (intentionally spelled that way) maybe.

 

Sustainable pricing is part of good brand. Look at Apple.

HD 7000 series was a far step from former AMD/ATi product generations, which were characterized by very good price per performance ratio.

But with stalling costs per transistor were a probably look in a bad future for higher end GPUs until new solutions are coming up.

 

Imagine you bought a Ferrari for $200.000 a few months ago. Now there's a new Porsche on the market - and Ferrari decides to suddenly lower their prices by 25%. Very bad decision as

(a) it's basically like admitting that the Ferrari is inferior to the new Porsche and can't compete at the "old" price.

(b) people who bought the Ferrari at the original price will feel screwed over - a lot of unhappy customers right there.

(c) observers might deduce that it's not worth buying a new Ferrari at all - as fragile price stability basically destroys the feeling of "value" attached to a purchase.

(d) Peope interested in buying a new Ferrari in the future won't buy one right after market introduction anymore - but instead wait for a price cut. Results in a general slowdown of sales.

 

They won't create a strong brand over night, but I really don't buy the whole mercedes vs volkswagen crap in the graphics world. It's more like Ford vs GM. Or Honda Vs Hyundai or something. They aren't worlds apart in what they can demand for a product, but you're not going to become a Mercedes by acting like the Yugo.

 

Guess they need to do some brand building then, if thats your general opinion of them... And we've come full circle.

 

I think 1000 would be the highest they would want to put their cards through without putting any extra effort into binning. Any more and they have to raise stock clocks just to safely get all their cards stable. OEM models are always clocked conservatively.

http://forums.anandtech.com/showthread.php?t=2228094

The worst overclock that was achieved was 1015 mhz. Alot of the cards hit 1100+ but when was the last time a videocard maker left 0 overclocking room in their cards. I think 1050mhz would piss alot of manufactures off selling OC cards, because most of their OC models would have a hard time selling with a 50 dollar premium and no difference in clock speed.

And this is all assuming Nvidia doesn't do the same thing and raise the clock speed of their models. The gtx 680 has very good overclocking ability as well.

The best thing AMD can do without stepping on too many feet is lower the price moderately($479 for a 7970, $399 7950, $329 or $299 7870) and get the hd8xxx out later this year and be more aggressive with the clocking. Selling at the same price or higher is only going to make them lose marketshare. The gtx 560 ti was the worse performer than the 6950 and it killed the 6950 for sales because it was slightly cheaper(than the 1gb model anyways).

 

a) what if it is (and most likely)? Now prices have to remain high just because some silly person with too much money and weak character needs some artificial price constant to keep him/her feel/thinking he/she did a good purchase?

b) Foolish/greedy downsides. Anything i buy expensive has to remain so, because im special and never make dumb purchases.

c)Good, reality kicks in, consumers didn't fell for marketing. And i bet they would/will even feel better.

d) Slow downs will always occur. Can even occur from prices being way too high for what the product is or its quality. Or the market simply being oversupplied ( in the case of medium/high end graphic cards, has it ever?).

 

I think AMD could talk to their partners, have them liquidate stock of the current OC models and then make a reference model with 1050~ (remember that the process will get better with time), and the new OC models start over with the 1100. In theory, anyway

I think you are right. I mean, it would be a double-edged sword for AMD to raise the performance of their HD7970 because it would noticeably lessen the HD8000's performance increase, and with the current OC models AMD could do it right. To me, AMD should set the price of reference models at US$470 and the OC models up to US$550.

In actuality, I think the price is right on the 7970. The 7970 RTSO of the 680 in GPGPU, and in this market it counts. The difference in performance isn't so high.