Would not be enough against a good clocked G92...
And R680 seem definitely to be Dual-RV670/RV680.
So, do we know anything about RV670 yet?
- Thread starter nicolasb
- Start date
Would not be enough against a good clocked G92...
And R680 seem definitely to be Dual-RV670/RV680.
I sincerely doubt that this actually was the case but I would definitely "lol" if it was and AMD fell for it
. Either way, how much flexibility did AMD really have in this situation anyway. They have a very cool running chip in RV670 but how much can they really play with core speeds after seeing what Nvidia is putting forward?jimmyjames123
Regular
I sincerely doubt that this actually was the case but I would definitely "lol" if it was and AMD fell for it
Where is Dave Baumman when you need him?
NV has been very successful at misdirection lately. On top of that, everyone thought they were moving in the direction of smaller and smaller die size, but then they pull out the G92 which has a surprisingly large die size.
It seems that NV's 8 series architecture is so flexible that it's very tough for the competition to get a good read on what variants NV can come out with. For instance, you have various memory bus width's already in play, various numbers of stream processors running at a varying rate of speed, in addition to the standard variation in memory speeds.
Should be very interesting to see how the RV670 compares to the G92 in games. We know that the 2900XT compared well to the GTX in 3dmark06, but not so well on games (especially with AA/AF enabled). One thing that shocks me is that we have yet to really see any leaked benchmarks of any kind for the RV670.
Last edited by a moderator:
Dunno why anyone would assume that this is anything close to being vaugely non-fictional given the template of that slide.
NV isn't the only one successful at misdirection... Sudden NV roadmap changes at the end of september concerning the final config and price of G92GT kinda suggest that they expected something different (less powerful) from RV670 and most of all they expected it in Q1 08.
I'll quote this:
It's drafted in price order, not performance order
Yeah. It also explains the mess that is now called the Geforce 8 lineup from Nvidia.
You can thank the performance level to AMD (RV670) for being competitive. It went from 64SP->96->112SPs.Still, exceptionally well performing.
775 MHz instead of 825 Mhz is quite a drop
Maybe RV670 doesnt need to be 825 to be competitive?
Or conversely they dropped it to hit lower price levels. Either way, too early to call.*edit* The Diamond page is offline now, didnt they break the NDA last time too?
Jawed
Legend
It would seem the priority is a single-slot cooler, which I guess puts a lower ceiling on core clock. So the single-slot thing allows them to market 4-way CrossFire with their top card.Still, exceptionally well performing.
775 MHz instead of 825 Mhz is quite a drop
Will anyone actually build such a system? Anyone who isn't a reviewer, that is? The only people I can think of will be those who like to play 3DMark.
Jawed
Still, exceptionally well performing.
775 MHz instead of 825 Mhz is quite a drop
Possibly, only if you believe that the leak info was ture
Also, when they drop the max speed, that means more yeild of the chip could pass to the line at that price point too, which I tend to agree with what serenity posted.
Edit: Oop my spelling
Bouncing Zabaglione Bros.
Legend
Possibly, only if you believe that the leak info was ture
Also, when they drop the max speed, that means more yeild of the chip could pass to the line at that price point too, which I tend to agree with what serenity posted.
But if you clock lower, your performance goes down, and you have to charge less.
jimmyjames123
Regular
Most definitely. No doubt, NV must have been pretty surprised by RV670's reportedly good performance/price ratio, in addition to it's reportedly small die size, especially after seeing the 2900XT.
Fortunate for NV that their architecture seems quite flexible in terms of tweaking by using different combinations of shader processors, shader domain clock speed, memory bus width, core clock frequency, memory clock frequency, etc in order to boost performance on expectations of good performance from ATI/AMD.
It probably could use better proofreading.
It only states DX 10.0 support.
That and the AA feature is mentioned twice.
It only states DX 10.0 support.
That and the AA feature is mentioned twice.
Still over 2 million sold computers the last quarter is nothing to look down upon
Even if Apple sells 8 million Macs (of all descriptions) this year, it will still amount well under 5% of the 220M + computers sold annually each year--a number which is growing again. Yes, Apple computer sales should not be looked down on, but neither should they be exaggerated.
<hardcore gamer mode>
You're right. Apple computer sales should not be looked down on, but rather spat upon.
</hardcore gamer mode>
You're right. Apple computer sales should not be looked down on, but rather spat upon.
</hardcore gamer mode>
The_Wolf_Who_Cried_Boy
Newcomer
With the focus being crossfire/multigpu rendering and ATI's history of limiting vendor freedom with clockrates is it at all likely a dual slot factory overclocked edition will be released for us single card prefering luddites?
I thought the HD3870 is a dual Slot card?
Could it be the yields or we were fooled along with the 800+ Mhz on core rumour?
