AMD: R9xx Speculation

Anyhow this is similar to the situation of GF104 , nobody knew the correct number of shaders , nobody knew about super-scalar arrangement , and all leaks were wrong , except for those that came 1 weak or so before launch.
well, hints for changed arrangement (and exact!) were available a month before launch :)
 
Radeon HD 6850: 960 ALUs @ 775 MHz (~230mm² / 1488 GFLOPS / ~$200) vs HD 5770 (170mm² / 1360 GFLOPS / $125) vs GTX 460 768 MB (368mm² / $140) :cry:

Radeon HD 6870: 1120 ALUs @ 900 MHz (~230mm² / 2016 GFLOPS / ~$275) vs HD 5850 (334mm² / 2088 GFLOPS / $230) vs GTX 460 1 GB (368mm² / $190) :cry:

Note: Prices are taken from Newegg.
If those 3dmark numbers are any indication, the HD6850 should rather be compared to GTX460 1GB, which would hence cost pretty much the same (not that the GTX 460 768MB is much slower, with current prices this has much better price/performance ratio than GTX 460 1GB, since it's only about a 10% perf difference but a 30% price difference - maybe 768MB memory just doesn't sell... ). Since as the sole performance metric you included gflops you could as well include that for GTX460 - which would be only 900 gflops... Also listing the HD5770 there is a bit pointless imho, if anything HD5830 would make more sense.
 
Since as the sole performance metric you included gflops you could as well include that for GTX460 - which would be only 900 gflops...
While we can't use GFLOPS to compare a GeForce vs Radeon, we can use it to compare 2 different Radeon.
When doing so a Radeon HD6850 (1488 GFLOPS) ~ HD 5770 (1360 GFLOPS), and a HD6870 (2016 GFLOPS) ~ HD 5850 (2088 GFLOPS).

Also listing the HD5770 there is a bit pointless imho, if anything HD5830 would make more sense.
Maybe, but the Radeon HD5830 offers 1792 GFLOPS vs 1360 for the HD5770, and the HD6850 = 1488 GFLOPS.
That's why I chooses to compare the HD6850 with the HD5770.


It is likely that these numbers do not represent the true performance of the new architecture (I really hope so), but for the moment these are the only numbers that we have at hands.
 
Also listing the HD5770 there is a bit pointless imho, if anything HD5830 would make more sense.

5830 was death after gtx460 came out(maybe even before :LOL:). Anyone buying a 5830 over a gtx460 needs to be realy misinformed.
The weak and expensive 5830 is actualy the main reason for the barts release first.
 
While we can't use GFLOPS to compare a GeForce vs Radeon, we can use it to compare 2 different Radeon.
When doing so a Radeon HD6850 (1488 GFLOPS) ~ HD 5770 (1360 GFLOPS), and a HD6870 (2016 GFLOPS) ~ HD 5850 (2088 GFLOPS).
Even for similar architectures, I don't think comparing just the gflops numbers makes a whole lot of sense. You could basically just as well compare rops, and figure HD68xx is faster than HD58xx (and way faster than HD57xx, of course). Even using a single "real world" benchmark (the 3dmark score) makes way more sense than to just pick gflops.

Maybe, but the Radeon HD5830 offers 1792 GFLOPS vs 1360 for the HD5770, and the HD6850 = 1488 GFLOPS.
That's why I chooses to compare the HD6850 with the HD5770.
You're just giving way too much significance to that metric. Also, we know HD5830 is underperforming for the gflops it has compared to HD5850.

It is likely that these numbers do not represent the true performance of the new architecture (I really hope so), but for the moment these are the only numbers that we have at hands.
But making up performance assumptions about those numbers just doesn't make sense, only numbers we have or not.

5830 was death after gtx460 came out(maybe even before :LOL:). Anyone buying a 5830 over a gtx460 needs to be realy misinformed.
Certainly, but really the HD5830 is more comparable both in price (and very, very likely performance) to HD6850 than HD5770 - if the HD6850 isn't faster than HD5830 I'm going to be disappointed.
The weak and expensive 5830 is actualy the main reason for the barts release first.
Not sure that's really the case, but it's certainly at a price (and performance) segment where AMD is somewhat lacking compared to nvidia.
 
5830 was death after gtx460 came out(maybe even before :LOL:). Anyone buying a 5830 over a gtx460 needs to be realy misinformed.
The weak and expensive 5830 is actualy the main reason for the barts release first.

5830 was as expensive as a gtx 460 768 and at higher resolutions played better due to the bigger frame buffer. You could have gotten 5830s on rebate as low as $130 and two of them for $260 was a good deal
 
5830 was as expensive as a gtx 460 768 and at higher resolutions played better due to the bigger frame buffer. You could have gotten 5830s on rebate as low as $130 and two of them for $260 was a good deal

Here in europe its around 200 € and the cheapest gtx460 768 around 160 €.:rolleyes:
 
5830 was death after gtx460 came out(maybe even before :LOL:). Anyone buying a 5830 over a gtx460 needs to be realy misinformed.
The weak and expensive 5830 is actualy the main reason for the barts release first.
Hind sight... But 5830 would probably have been much better received/perceived and sold better had it been released at $199. I see a lot of "what's the best card for under $200", or "what's the best bang for the buck" type of messages in gaming forums. The 460 took that segment easily, even though the better version was above $200 initially.
 
Radeon HD 6850: 960 ALUs @ 775 MHz (~230mm² / 1488 GFLOPS / ~$200) vs HD 5770 (170mm² / 1360 GFLOPS / $125) vs GTX 460 768 MB (368mm² / $140) :cry:
More like GTX460 1GB performance and price.
Radeon HD 6870: 1120 ALUs @ 900 MHz (~230mm² / 2016 GFLOPS / ~$275) vs HD 5850 (334mm² / 2088 GFLOPS / $230) vs GTX 460 1 GB (368mm² / $190) :cry:
GTX470 performance and price I think.
 
I guess the 5830 price is largely dependent on the lack of suffciently defective cypress dies. Why sell cheap 5830s when the production cost is virtually the same as more expensive 5850. Then it's better to sell them by raising the 5830 and lower the 5850 price to adjust demand. You should be present in both price ranges ofcourse, but don't have to be equally attractive.
 
If u click on the product u will see that the average price is much higher than cheapest on the main page. But yes it also depends on the taxes in EU country.
This truth applied to both products ;)
By the way I checked the price for "reliable" retailers and they are pretty much the same. (actually the HD5830 is some euros cheaper but less than ten euros is irrelevant).
 
What the heck are those " :cry::cry: " supposed to mean?

1) Pre-order prices are ALWAYS too high. As soon as the initial demand wave is over, prices will definitely normalize at a lower level, something like 150-160$ for 6850 and 200-220$ for 6870.

Always? :D

Both the 5870 and 5850 had preorder prices that were exactly the same price as AMD's MSRP. Well that just blew your always theory right out of the water. In fact, there was also virtually no price gouging for the first month or so of 58xx sales. Prices didn't in fact start to go up until AMD revised their MSRP to reflect that supply was so far under demand that it would be fiscally irresponsible of them as a business to not increase the price. At least in the US, and from what I remember also in the EU.

As to the :(, yes, that's absolutely a slap in the face that it's called 68xx. One nice thing for AMD moving to names rather than number schemes for their cards is that it's easier for them to obscure the fact that Barts would have been named Rv940 and Cayman would have been Rv970 had they used the traditional naming schemes.

Either way from everything revealed thus far, 68xx is a big FU to your everyday consumer. And I can only imagine every card going down the line will be similar in offering little to no performance improvement over the cards that they replace. Yay for PR. :p

I'm just waiting to see final performance numbers for 68xx in reviews. If it's similar to what's been revealed thus far, I won't be buying any AMD products for 1 or 2 generations.

Regards,
SB
 
Back
Top