NV30 FINALLY Being launched!!

[sancheuz]

Well, even nvnews moderators have confirmed nvidias introduction of the nv30 into the public world this november. Isnt that great? It looks as if they might just make the december deadline. We will finally know what the nv30 has.
(Worthless, but fun)Predictions:
As far as Clockfeatures I predict
-400 MHz
-200 MHz seperate t & L chip speed
-700 MHz DDR 1 memory
-256 MB Capabilities
-16x AA with no hit on frames{yeah right.. :) }

[Bigus Dickus]

More worthless but fun predictions:

350 to 400MHz core clock
8x FSAA offered
16x AF offered

Late January availability

[SlyBoots]

bah, paper....I suppose ATI will use the opportunity to announce the R9700 Ultra:p

[Nutty]

Late January pah..

http://www.xbitlabs.com/news/story.html?id=1033601507

I should also remind you that according to the unofficial information, NV30 based graphics cards will be available for sale almost right after the announcement.

Please god, let that be true!!

[volt]

Originally posted by Bigus Dickus
More worthless but fun predictions:

350 to 400MHz core clock
8x FSAA offered
16x AF offered

Late January availability

They stated that the card should be widely available (within few weeks) after a launch.

[monkeydust]

In a related note, I'll have a Gainward Geforce 4 4600 for sale cheap shortly after Comdex. :D

[thcdru2k]

i'd like to imagine nv30 coming up with full expectations, ddr-II, etc. ehh might as well be positive about the release. we want better cards don't we.

[StealthHawk]

Originally posted by volt


They stated that the card should be widely available (within few weeks) after a launch.

not if you believe the rumors :(

i hope nvidia's card also doesn't turn out to be bandwidth limited due to slow memory or slow bus. i'm looking forward to seeing 8x FSAA in action :)

[Bigus Dickus]

Originally posted by volt


They stated that the card should be widely available (within few weeks) after a launch.

They (Xbit labs) stated that... not NVIDIA. Read the news blurb again. It's nothing but baseless speculation from Xbit, just like mine is basesless speculation. However, my speculation is supported by plenty of rumors, and some information that comes from individuals having connections with TSMC and telling of manufacturing problems with the NV30 that put it's volume ramp firmly in the January timeframe... if they're lucky.

[***CENSORED***]

:p :p :p omigod,omigod,ojoy,orapture
yeaw:) :p

[borntosoul]

yeh but we might have to wait till next year for the ddr11 and the higher clocked one ,i still think we getting the vanilla one first then ultra a few months later then mid/value market version 6 months after that

[nutball]

Originally posted by sancheuz

-200 MHz seperate t & L chip speed


Now you already know that NV30 won't be multi-chip. Why say something silly like that?

[Kruno]

I don't think he means 2 seperate main chips maybe he could be reffering to the built in units and there could be 2 of them. That will be of course much less efficient on top of that the T&L unit runs at the speed of the core so if you would have 1 T and 1 L unit running seperatly at 200MHz and constantly having to seperately send geometry info across the pipe it would end up being a bottleneck.

Ask Chalnoth for a full explanation.

[nutball]

Isn't the concept of a T&L unit somewhat redundant these days?

If I recall correctly the T&L units introduced (in whichever generation TNT/GeForce it was) were basically hardware-accelerated implementations of the standard OpenGL pipeline. Previously on consumer-grade graphics solutions the transformation + lighting had been done in software by the CPU. Of course high-end solutions had hardware right through the OpenGL pipeline, which was why they cost so much money.

The fixed-function pipeline is becoming increasingly less relevant in these days of programmability, programmability, programmability.

The role previously played by the T&L unit is now handled by the vertex processor surely? Transformation and lighting calculations aren't seperable in this model.

There may well be multiple vertex pipelines on NV30 (in fact I'm sure I've seen quotes from NVIDIA people that there will be). But why would these run at a speed lower than the rest of the core? That would be daft.

[Kruno]

The role previously played by the T&L unit is now handled by the vertex processor surely? Transformation and lighting calculations aren't seperable in this model.

Not in really old games, the GF3 afaik has a fixed function as well as reprogrammable pipeline. Guess it is used to play DX 7 T&L games.
Correct me if I am wrong.

[nutball]

No, you're right as far as I understand it the fixed function pipeline is implemented by different transistors than the programmable sections.

Point is it seems to me that the fixed function parts must surely occupy much less real-estate than the programmable parts? (I guess one could esimate this from the increased transistor count from GF2 -> GF3 -> NV30???)

What's the point spending precious transistors adding a second fixed function pipeline in an architecture which is all about programmability (if the NVIDIA PR is anything to go by).

I get the impression reading between the lines of what NVIDIA people have said that there are already features which have been omitted from the programmable pipeline due to limitations of the transistor budget.

Why waste good trannies on speeding up Duke Nukem 3D??? Especially if you're going to use them to implement two T&L units which you then underclock?

[Stavros]

I think the NV30 is going to be 1000 separate chips running at 1Ghz each, with one huge heatsink covering all the chips with a fan thats modified from helicopter tailfin rotor, and it will require a small nuclear powerplant to generate the power needed to run it (AGP port doesn't produce the power needed) and it will come with an interface to connect the NV30 to your brain. The NV30 will then use the unused memory areas in the brain as fast high bandwidth memory.

This is purely speculative

[SavagePaladin]

I've heard so many rumors I safely disregard them all....but for one thing
NVidia has never disappointed me wrt engineering and support.
I don't expect them to now.

[Kruno]

:eek:

"Why waste good trannies on speeding up Duke Nukem 3D??? Especially if you're going to use them to implement two T&L units which you then underclock?"

[SavagePaladin]

hey K.I.L.E.R
...wierdo (thats all I can think when I see your name, nevermind)

animevamp I think has a FF7 WinXP patch that lets you use FSAA without artifacts...

i dunno, it looked kinda cool, except that i don't like FF7 anymore

[Kruno]

That's not the case with that patch.

[sancheuz]

Originally posted by nutball


Now you already know that NV30 won't be multi-chip. Why say something silly like that?

It wont? I thought it was going to be multi chip. In that case never mind that about the t & l engine clock speed I said. However, if you go to that picture on the anand report on the nvidia plant tour, and look at the picture where they are simulating the nv30 with IKOS, that have 3 boxes there with LEDs showing three numbers:
-400
-200
-700
We know the main chip will be 400, and that the memory will be 700, so what is the 200? That is where i came to the theory that the 200 represents a seperate t & l engine. I could be wrong though.
Here is the pic: http://www.anandtech.com/video/showimage.html?u=http://images.anandtech.com/reviews/video/insidegraphics/NV30running.jpg

[Nv40]

This is Great news! :)

there will be so many ATi fans dissapointed ,to see Nv30 is
On schedule and not a February 2003 release ,like many
beyondfans members here ->wished :)

ati must be working overtime to sell every radeon9700 while they can
,in inventory ,before Nv30 hits the shelves .
The issue is not if Nv30 will be faster ,
(we have the word of Creative Europe and anand who saw the Nv30 real specs):)
It is when in November or december it will be released ,because as ANand have say , NV30 IS ON TIME !

Ati Must hurry up with the tapeout of the R350 ,,, hehe
because they will need alot more than Luck to end this year
as the leader in 3dgraphics. :D

however a big kudos to ATI ,thanks to them ,Nv30 will be
even more powerfull than Nvidia original Nv30 desing plans .

WHat a world ,great competition ,but at the end ,the
Stronger and smarter COmpany will lead the Industry.

Roll on Nv30 ;)

[jbirney]

Bigus Dickus,

do you want the honor of trying to provide logic facts that counter point out some of NV40s statements or should I?

[StealthHawk]

Originally posted by jbirney
Bigus Dickus,

do you want the honor of trying to provide logic facts that counter point out some of NV40s statements or should I?

please, please just ignore him already :D