| Monday, July 14 2008 |
As we inch closer to the R700 release, which previews prove to have dethroned the GeForce GTX 280 convincingly, taking the crown for the fastest graphics card there is, NordicHardware reports AMD could be giving 'final-touches' to a fresh-wave of GPUs for early 2009. The timing is a critical factor here as that's the time-range the proposed NVIDIA G300 carrying 384 SP's and 512bit GDDR5 memory bus is marked for. The R800 cards could carry 2,000 stream processors, with its unit processors the RV870 holding 1,000 stream processors each. They could enter the realm of either 40nm or 45nm fabrication process.
Source: Nordichardware
Source: Nordichardware
User comments
Sounds good. Wasn't there some rumours that AMD(ATI) was skipping 45nm and going straight for 40nm? Either way IMO just increasing the amoung of stream processors to 1000 isn't enough to compete with G300 (GT200 doesn't exist it's G200, look @ the chips - why do ppl call it GT200?) without further optimization. Any info about if it's based on R600 or some new design? thanks :)
R870 already? gotta be kidding me...
200 more stream processors after a die-shrink could hint at the die-size remaining the same. :)
Maybe just 200+ streams but a die shrink to 45 or 40nm is definitely a clock speed increase to the 1Ghz barrier maybe 1,2Ghz or even 1,5Mach:D
1000 isnt the right number .. nordichardware are just guessing
Does 1280 look realistic?
no
I wouldn't put anything past AMD at this point but the overkill factor is definitely starting rear up in huge numbers.
i think the 40nm cards will be like wat happened with die shrink of HD2xxx series and producing Hd3xxx series . I dont think there will be much performance improvement, power requirements may be reduced.
WHat is interesting to see is the GPU's moving to a smaller fabrication process than CPU's .
WHat is interesting to see is the GPU's moving to a smaller fabrication process than CPU's .
I am quite surprised that GPUs are moving this fast. I wonder if we will see much improvement on heat production with smaller GPU. (vs 55nm, probably not much?)
I think this architecture is really suffiecient, and addding more rops and giving more shaders and clockiing it past 1ghz should give it a great boost probably another 80-90% considering shader increase and clock increase. but I think ati will go for more shaders than clock increase as that will be better off for future games, I love the effeciency of the current design and only refining it would make sense. And nvidia has some serious work to do, they can't have a giant chips with all them rops and tmu's that are bearly effecient for the numbers they represent.
keep the 800 and clock then independently like Nvidia does and that will be pure Pwnage! :rockout:
by: WarEagleAUIMO, I think this might be a viable route for them . . . hell, the R700s were hinted at having seperate clocks long before we saw the actual specs . . .
keep the 800 and clock then independently like Nvidia does and that will be pure Pwnage! :rockout:
either way, though, it's looking like ATI won't be holding back on the green-camp beat down next series.
by: WarEagleAUI wish they would do that, but knowing AMD, they'll try to squeeze every last SP they can in there. The R800 will be on a level we can just dream about now & the RV870 will be the next single GPU king. They won't just shrink it, they'll stuff it. I don't mind it but just how many SPs can they fit on one die :rolleyes:
keep the 800 and clock then independently like Nvidia does and that will be pure Pwnage! :rockout:
Get 40/45nm out quick. The urgency is on cooler and lower power.
ATI IS Bringin sexy back!
Was there much of a performance increase gleaned from the die shrink,going from hd2xxx to hd3xxx?
Whatever the specs of the r800,its good to see whats coming,even if its not written in silicon yet
Whatever the specs of the r800,its good to see whats coming,even if its not written in silicon yet
2000 sp+ per core and 40nm tech in 2009 for r870
separate GPU CLK/ shader CLK
separate GPU CLK/ shader CLK
Just gonna have to skip these middling RV770/R700 products. ;)
by: tigger69There was an immense improvement to power consumption while performance remained the same. Something like a 100W drop at load from R600 while the architecture remained mostly the same. Powerplay brought idle down even further. It allowed them to create 3870X2. ATI also saves lots of money on the cost of each GPU, complexity of the board and its cooling apparatus.
Was there much of a performance increase gleaned from the die shrink,going from hd2xxx to hd3xxx?
Whatever the specs of the r800,its good to see whats coming,even if its not written in silicon yet
Experience with 55nm RV670 undoubtedly helped with development of RV770.
Only 200 stream processor more in one kernel. But how many kernel in one card or chip. Maybe 3x or 4x :respect: :rockout:
i hope they bring some CPUs to go along with those monster GPUs that should bring Nvidia+Intel to their knees....
We'll see. I remember reading somewhere that ATI was going to head towards at Tick/Tock strategy similar to that of Intel. If that's the case, it will be more like HD2xxx to HD3xxx, some performance increase, smaller thermal envelope and less power draw.
by: Rurouni StrifeI've heard similar as well . . .
We'll see. I remember reading somewhere that ATI was going to head towards at Tick/Tock strategy similar to that of Intel. If that's the case, it will be more like HD2xxx to HD3xxx, some performance increase, smaller thermal envelope and less power draw.
but we can't really call Intel's strategy tick/tock anymore . . . they've gone over to full-blown hop scotch.
Yeah, AMD isn't likely to make a horrible mistake like that *cough...sandbagging...cough*, when they showed us so much power in this series. We've gotten greedy, very greedy. If the performance of the next series is similar to this one, many of us will just skip it & AMD doesn't want that now do they.