OTOH, shouldn't the supply constraints be coming down soon?
AMD: R9xx Speculation
- Thread starter Lukfi
- Start date
OTOH, shouldn't the supply constraints be coming down soon?
Silent_Buddha
Legend
Well that speculative chart on the previous page was estimating ~450 USD for 6870. I'm going to be a very sad panda if that's the case. 
Although if it comes by default with 2 GB of memory (extremely doubtful) I wouldn't be too teary eyed.
Lowered supply constraints just means prices should fall over time (even without much comp) rather than the product getting EOL'd at a higher price than it launched for a variety of reasons. First among them being that a higher price is going to lower demand, and greater supply of wafers means a greater supply. Thus it'll reach equilibrium sooner rather than later (or never) and eventually supply will outstrip demand, at which point to maximize sales they'll have to drop price to stimulate demand. In other words, it'll still launch at the higher price, but over time it's much more likely to drop in price than what happened with a supply constrained 5870.
Of course, if the Die is ~15% larger they'll need 15% or more wafers just to maintain the same supply of chips and cards.
Regards,
SB
Although if it comes by default with 2 GB of memory (extremely doubtful) I wouldn't be too teary eyed.Lowered supply constraints just means prices should fall over time (even without much comp) rather than the product getting EOL'd at a higher price than it launched for a variety of reasons. First among them being that a higher price is going to lower demand, and greater supply of wafers means a greater supply. Thus it'll reach equilibrium sooner rather than later (or never) and eventually supply will outstrip demand, at which point to maximize sales they'll have to drop price to stimulate demand. In other words, it'll still launch at the higher price, but over time it's much more likely to drop in price than what happened with a supply constrained 5870.
Of course, if the Die is ~15% larger they'll need 15% or more wafers just to maintain the same supply of chips and cards.
Regards,
SB
Yes. The benchmark tool is from 2007 and says "Vista" in the title bar when running on Windows 7.
Fake they may be, but that reasoning doesn't make much sense. If you're going to mock up an Unigine benchmark, why would you bother manipulating an image when it's much easier to just edit the results html-file (of which that's a screenshot).
DavidGraham
Veteran
Just like people faked the hell out of Fermi scores before launch ..
Honestly if these are fakes I will be disappointed , we haven't had much of an architectural fun for along time now ..
Honestly if these are fakes I will be disappointed , we haven't had much of an architectural fun for along time now ..
http://www.chiphell.com/thread-119587-1-1.html
nApoleon confirmed Vantage P and Unigine scores in those screenshots.
nApoleon confirmed Vantage P and Unigine scores in those screenshots.
Last edited by a moderator:
Excuse my ignorance but, who is nApoleon? He may just as well be bull**ing.
He leaked specifications of Rv770 in Mar.2008.He also leaked some pics of evergreen before launch.
So the numbers for HD5800 and GTX400 are true?
Maybe they are/were expecting a "GTX 495".
That depends on how much they want to sell them for. They could use it as a vehicle to sell lower leakage defect parts, so in comparison to the 6850 they would have otherwise gone into, they would get higher margins.
DavidGraham
Veteran
If my memory serves me right , he did confirm some Fermi rumors too , and they turned out to be false ..
After seeing the inconsistent text style and alignment of the Unigine benchmark screenshot, I would say that this myth is busted.
DarthShader
Regular
You say the benchmarks are likely to be fake, but I think they show results that are well within expectations. If the 1920 shader in groups of 4 rumours are true, then you have a 20% increase in shaders and up to 20% increase in efficiency. That's up to 44% faster than Evergreen. There could be more, since memory bandwith seems to go up by 33%. If you add increased teseletion power and consider early drivers, a 30% boost at the moment sounds very reasonable to me, maybe around 40% at launch and about 50% when drivers mature. FWIW, IMO, YMMV etc.
You say the benchmarks are likely to be fake, but I think they show results that are well within expectations. If the 1920 shader in groups of 4 rumours are true, then you have a 20% increase in shaders and up to 20% increase in efficiency. That's up to 44% faster than Evergreen. There could be more, since memory bandwith seems to go up by 33%. If you add increased teseletion power and consider early drivers, a 30% boost at the moment sounds very reasonable to me, maybe around 40% at launch and about 50% when drivers mature. FWIW, IMO, YMMV etc.
1920 SPs/ 4 wide VLIW = 480 TPs. 480 TPs/16 TPs per SIMD = 30 SIMDs. 30 SIMDs *4 TMUs per SIMD = 120 TMUs. How big would be this GPU? Bigger than R600?
Similar threads
