NVIDIA GF100 & Friends speculation

[Sontin]

Why don't we talk about this:

it's still BS because the units are THERE, why disable them?

http://www.xtremesystems.org/forums/showpost.php?p=4198908&postcount=905

So, nVidia will release GF100 with <10 SM/Cluster? That's funny.

 

[Razor1]

Why don't we talk about this:

http://www.xtremesystems.org/forums/showpost.php?p=4198908&postcount=905

So, nVidia will release GF100 with <10 SM/Cluster? That's funny.

hmm AMD wouldn't have know the performance of Fermi till sometime this week, and even then drivers are still going to help Fermi conciderably, Huddy's statement was well before that, and no nV's high end Gf doesn't have disabled units, and performance......

8X MSAA performance has been fixed btw on Fermi.

 

[neliz]

A barrel of laughs, really.

 

[digitalwanderer]

Why did he quote you from another forum Neliz, are you being unfaithful?

 

[neliz]

Why did he quote you from another forum Neliz, are you being unfaithful?

Even a prostitute needs pausing.

 

[boxleitnerb]

And power consumption? Are we talking closer to 5870 or 5970?

 

[DegustatoR]

Why don't we talk about this:

Yeah. I think we may even talk about other science fiction stories while we're at it.

 

[neliz]

Yeah. I think we may even talk about other science fiction stories while we're at it.

I think even zed-x "nda expires any day now" said I was right about <80 TMU's

GF100 has more ROPs then G200 and less TMUs then G200, but that is not a problem, has new units - raster a geometry units

 

[OlegSH]

What's a new raster and geometry blocks? Maybe it's hardware triangle cull unit to reduce load of triangle setup? And raster block, did nvidia already have something like tile coalescing?

 

[Rys]

"I was right" ? You missed the A and T chars there, man. Really, they should just post themselves, save you the bother

 

[FrameBuffer]

hmm AMD wouldn't have know the performance of Fermi till sometime this week, and even then drivers are still going to help Fermi conciderably, Huddy's statement was well before that, and no nV's high end Gf doesn't have disabled units, and performance......

8X MSAA performance has been fixed btw on Fermi.

Are you suggesting that ATI and the HD5000 series won't benefit from driver updates ?

As far as 8X MSAA, it's about damn time. ATI's only been dominATIng nV in such for how many generations now ?? Now only if ATI will fire back and fix their piss poor AF performance. (and 120Hz issues)

 

[Ailuros]

Are you suggesting that ATI and the HD5000 series won't benefit from driver updates ?

Both IHVs have very capable driver teams.

As far as 8X MSAA, it's about damn time. ATI's only been dominATIng nV in such for how many generations now ?? Now only if ATI will fire back and fix their piss poor AF performance. (and 120Hz issues)

Well that's the cute thing about heated competition; since it's inevitable to avoid all possible shortcomings (I'm not aware of an IHV named GOD for absolute perfection) they're forced to work hard enough to fix them asap.

The real question with each architecture is if an IHV has made more wrong than right design decisions and that's still a book with the majority of pages left blank for GF100.

 

[trinibwoy]

I am absolutely shocked and dismayed that there hasn't been a single leak as yet and it's been days since the "deep dive". What good is the internet any more? It's like we're back in the dark ages I appreciate neliz's info but it raises more questions than answers.

 

[Silus]

Are you suggesting that ATI and the HD5000 series won't benefit from driver updates ?

As far as 8X MSAA, it's about damn time. ATI's only been dominATIng nV in such for how many generations now ?? Now only if ATI will fire back and fix their piss poor AF performance. (and 120Hz issues)

Not many. They didn't dominate anything during R600 and RV670.

 

[Silus]

I am absolutely shocked and dismayed that there hasn't been a single leak as yet and it's been days since the "deep dive". What good is the internet any more? It's like we're back in the dark ages I appreciate neliz's info but it raises more questions than answers.

I have to agree. Many have been talking about next monday as "the day", but I expected some leaks before that and this far, we got almost nothing...

 

[Razor1]

Are you suggesting that ATI and the HD5000 series won't benefit from driver updates ?

Did I say that drivers for ATi's cards won't? Why do people just assume things?

Unigine benchmark with tesselation, Fermi is much faster then Cypress as it is right now around upper 50%.

 

[DemoCoder]

It may be there there's more improvement in drivers to be made on Fermi because the architecture has changed a lot and they're probably in the stage of just getting stuff working without crashing, whereas the Cypress stuff is less radical, and more mature. The tessellation stuff has me curious, I'm wondering if it really is just ALU tessellation, and Fermi just happens to be able to run that code a lot better, or if there's been some grand improvement in setup rate.

 

[Davros]

whats the prob with 8x msaa ?

 

[Ailuros]

whats the prob with 8x msaa ?

Since G80/R600 all GFs lose more performance than Radeons with 8xMSAA.

 

[LordEC911]

As silent_guy often explained, a metal spin will not miraculously increase clocks. Yields seem to be a problem yes, but I'm not sure that is the cause for what neliz is saying, if it's true of course

My interpretation was that it doesn't help increase max clocks... but it can increase the number of parts in a potential bin.

Why did he quote you from another forum Neliz, are you being unfaithful?

He decided to give me some extra help over at XS with all the BS/non-info.
Or he might have just been fed up with me repeating almost all info he gives here, over at XS, and just decided to post there as well.