Full NV35 official feature list inside

[MuFu]

 Accelerated pixel shaders allow for up to 12 pixel shader operations/clock

The way that is worded, bearing in mind the wording of the NV31 "leaks" prior to it's launch it sounds like the logical equivalent of 2 x NV31. The same can probably be said of the memory interface (Dave, you were right mate...). 2(2x64bit).

 Dual DVI ports for interfacing to external TMDS transmitters and external TV encoder
 Dual internal TMDS encoders (one single and one Dual link) able to drive next-generation flat panel displays with resolutions greater than 1600 × 1200

Hmm... =\

Sounds like two 165MHz transmitters integrated with the option of an external chip (Sil 164 etc). That should be "DVO" ports, btw.

MuFu.

 

[Mark]

No mention of improvements in Anti-Aliasing and Anisotropic? Have the managed to dump the FlowFX?

 

[MuFu]

The 12 PS ops/clock thing is rather vague, because you could actually describe the NV30 that way too. It's still interesting to see nVidia won't market it as a "8 pipelines" architecture, though.

Of course. That's because it's a 12 pipe architecture!

BTW, one of the few things I know *definitely* about NV35 (source is nV@board-level) is that some of John Carmack's ideas were incorporated. I doubt that just refers to getting Z-trickery working with stencil ops.

MuFu.

 

[LeStoffer]

BTW I like this bit:

* Full 128-bit, studio-quality floating point precision through the entire rendering pipeline with native hardware support for 32 bpp, 64 bpp and 128 bpp rendering modes

No need to fear that those state-of-the-art int12 Register Combiners are going anywhere. This is also nice because we now don't have to fear that the CineFX whitepaper from July 2002 would be rendered obsolete.

 

[stevem]

Sounds like two 165MHz transmitters integrated with the option of an external chip (Sil 164 etc). That should be "DVO" ports, btw.

Appears to be two units, one standard and the other dual link capable which is needed for greater bandwidth @ >1600x1200.

 

[Luminescent]

Well, supposedly Nvidia stated NV30 was capable of 8 pixel ops per clock, however, it was capable of 12 (8 FX + 4 fp). I wonder what Nvidia counted to obtain the new 12 number.

What about Open Gl 2.0 support?

 

[Bambers]

 IntellisampleTM performance technology, a Hi-Res compression technology (HCT), increases performance at higher resolutions through advances in compression and anti-aliasing technology.



:|

They had really better fix the AA for nv40.

 

[Tridam]

Uttar -> Sounds pretty like what I'm trying to explain here

I think the NV35 will rock and that NVIDIA's marketing will be right with regard to real specs this time !

 

[Arun]

Hmm, well, there's ONE thing the NV30 does at 8 ops/clock: LRP when running in Integer mode.
http://www.beyond3d.com/forum/viewtopic.php?t=5124

Do we know if it's done at 6 ops/clock on the NV31 & 4 ops/clock on the NV34? If it was the case, then it would make sense for the NV35 to have 12 INT LRP ops/clock...


Uttar

 

[Luminescent]

Just for reference, R300 and R350 can compute LRP at 8 ops/clock in fp mode. I have confirmed this with a reliable source and so has Dave.

 

[Hellbinder]

We'll see how it really stacks up against the 9800pro ultra high end models (if they exsist). Probobly will be pretty close except for all the Driver hacks they did forcing FP12 which im sure they will continue to use....

One thing this does.. is continue to make me feel like ATi really sold themselves short with the 9800pro basic specs. It just makes no sense that they went with such pitiful clock speeds at the start. The current 9800pro specs should be the 9800. The 9800pro specs should be at a minimum 400/400. Let the OEMs sell 9800. or actually better yet. have 3 9800 models.

-9800 Eliete 425/4xx
-9800pro 380/340
-9800 325/275 (or whatever)

Let the pro's be for the oem crowd and the elietes be for the enthusiast crowd. Its not like this is rocket science. But to release a card with such pitiful Ram and untapped Core speeds so close to the Nv35 (which everyone knows about) is just a pretty complacent attitude imo.

 

[MuFu]

Sounds like two 165MHz transmitters integrated with the option of an external chip (Sil 164 etc). That should be "DVO" ports, btw.

Appears to be two units, one standard and the other dual link capable which is needed for greater bandwidth @ >1600x1200.

That would be three units, though (165Mbps x 3). That part of the spec seems incorrect to me. I am sure it just has dual integrated 165MHz transmitters + DVO outputs for external devices.

MuFu.

 

[LeStoffer]

Uttar -> Sounds pretty like what I'm trying to explain

But what exactly did you hint at, if we assume 12 shaders ops per clock?

a) 4 traditional pipelines with 4 FP and 8 FX?
b) 4 traditional pipelines with 8 FP and 4 FX?
c) 6 traditional pipelines with 6 FP and 6 FX?
d) 8 traditional pipelines with 4 FP and 8 FX?
e) 8 traditional pipelines with 8 FP and 4 FX?

Add to that they might (hopefully!) make the FP shader and FP texture ALU separate and I'm not really sure what you suggest the case will be. So even if you're under NDA I still need a hint or two.

 

[JD]

No 2+ shaders?

 

[Tridam]

I think a but I hope b and I dream e

I think that in NV30, NV31 and NV34, the Floating Point ALU and the Floating Point adress processor share the same unit and it's why these chips can't do texturing and FP calculation at the same time. Maybe NVIDIA could split them up...

 

[Reverend]

Dunno 'bout you guys but this ranks as the best NVIDIA "refresh" in my books if Uttar is correct.

Of course, best refresh still doesn't mean best product at time of introduction!

 

[BoardBonobo]

I see this has all leaked to x-bit labs now... Do they ever collect their own stories?

 

[Hellbinder]

Im telliong you. Its really sad. You know damn well where they got this info. Yet they take credit for it.

B3D and the powers that be should take action of soime kind. how can a website be so unabashedly unethical?

 

[MuFu]

]Im telliong you. Its really sad. You know damn well where they got this info. Yet they take credit for it.

Have they changed the article since you read it? Looks like they give credit where credit's due to me... :?

Did you spam them with abusive e-mails?

MuFu.

 

[Evildeus]

]Im telliong you. Its really sad. You know damn well where they got this info. Yet they take credit for it.

Have they changed the article since you read it? Looks like they give credit where credit's due to me... :?

To me also :?