Post your breaking NV30 news links here!

I WANT TO SEE WHAT ALEXSOK HAS TO SAY ABOUT THIS ...

 

I would like to re-iterate this:
according to some sources..nvidia is not saying all the story...
It's quite clear nvidia has given very few architectural infos at this time.

 

We do know "something" about performance. Anand's article finished with that nice relative performance in Doom 3 graph

46 fps vs 33 fps for a 9700 on a 3 GHz Pentium IV

Although we have no ides of the methodology that produced those numbers, those are the ones people are going to be remembering.

I guess they are fairly plausible given the difference in the clock speeds.

I wonder if ATI will consider selling speed-binned R300's with an equivalent cooling solution.

 

Just read Toms - the only bandwidth advantage Tom gives the NV30 over the R300 is unknown LMA III effectiveness & colour compression (especially for AA).

But the R300 has AA colour compression on higher raw bandwidth so how is that an advantage to NV30?

 

The Geforce FX name is explained to come from NV30 being the fusion of 3dfx and nvidia tech - WTF?? Nvidia cannot be such a loser to rely on the whole "magical 3dfx technology" legend...

 

Anand has really sunk to new lows with his preview. Cutting and pasting Nvidia conceptual benchmarks into a preivew? I never thought I'd see the day.
Total hack job, and a total disgrace.

 

Exactly, while not totally sure, I'm still confident that many people will be eating crow - I think we need to wait to see what falls under the guise of their LMA-III.

So, take a picture of these responces

PS. Perhaps their even waiting to see the results of their respins. ATI launched without Multisampling support before, nVidia may just be keeping their options open incase their backed into a temporal corner with more metal errors.

 

Here is one point where ATi may be able to make a dent in the GeforceFX armour.....

"Although the GeForceFX's overall shader performance is higher than the Radeon's, it can only achieve this through much higher clock speeds. If the Radeon were clocked only 8% higher (350MHz), it would once again be even up with the GeForceFX."

http://www.tomshardware.com/graphic/02q4/021118/geforcefx-06.html

 

... what should the idea by holding back info be?

You would want to paint as good a picture as you possible can at this point to stop people from buying the Radeon 9700 Pro.

If you had a key tech to improve performance would you hold it back and let previewers go on saying: yeah... sure nice core speed, but it doesn't have the memory bandwidth to support it and is thus unbalanced in comparasion to ATI's offering? Would you?

I'm not trying to downplay the NV30, but I just don't think they would omit such crucial info, if they want to say: Hey, this is so awesome that you have to wait to February.

 

What bothers me is 2 things at the moment.

Aniso only goes upto 8x in that screenshot in either anands or toms preview.

FSAA - Every preview I've seen so far say "in addition to the current modes there are the new 6xS blah blah". This rather worryingly suggests that the lower FSAA modes are exactly the same as they are in the gf3/4 ie poo OGSS (or 'skewed' grid for 4xS)

Oh and we get given a doom3 benchmark probably taken on the same leaked alpha e3 version that I have sitting on my harddrive thats buggier than a termite mound.

 

Perhaps I'm Jesus....

 

Dunno..but do you remember gf1's register combiners?

Umh..I won't say anything at this time, I would like to double check my sources, but if what I was told it's true I do think nv30 will have no prob with bandwith at all..
And even if the things I know are untrue it seems they can do pretty neat things with less bandwith.

ciao,
Marco

 

Someone should teach Anand a little bit of math. I mean, I thought 16 * 4 is 64, not 48.
Also, he seems to forget that the Radeon 9700 has color compression for FSAA just as well.

It's a shame how low that site has fallen...

 

Precisely. I couldn't agree more. Nvidia is playing the cards that it has to lay down here and rather then the hand being a Royal flush it appears to be a bit more like a full house.... They don't need ATi to garner any more support then they have already managed so the logical response is to play what you have...(Although in this case it is something that nvidia won't actually have for a number of months yet.)

 

Can't be the leaked Alpha look at the bottom left of the picture:


It says 'nvdemo3' and we probably know what that means

 

I've just got back from the launch - I've not had a look at all this thread but here are 4 things:

1.) Its a 128bit bus.

2.) Its 8x1, not 8x2.

3.) There are no GigaPixel deferred rendering / geometry caching elements. ZCULL can reject more pixels per clock than GF4.

4.) FSAA is similar to GF4 and there's no programmable grids etc.

I interviewed one of the tech guys and assuming my tape caught it all I'll have it up as soon as I can.

 

Well, that would be a first. Marketing not shouting about every possible advantage (real or conceived) during a product launch?

Not saying it's not possible, but if there is something significant, there is only one possible explanation I can think of for not mentioning it: that whether or not the feature is actually enabled, depends on how the re-spin comes back.

 

You wouldn't think that NVidia has a rather new version of Doom3 ?

Not that it matters that much since i could care less about benchmarks coming from Nvidia using their own so called "nvdemo3".

Another thing, they claim 30-50 % performance advantage across the board. Will be rather interesting to see how those claims hold up.

 

So what we know?

* 1 vertex shader (or a FP array ala P10)?
* 8 pixel pipes 1 TMU per pipe
* 400-500 MHz (and the last one using what seems an overcloking large cooling solution)
* 128bit bus with DDRII at 500 MHz.
* more or less the same compression techniques than the R300
* nothing really new in terms of FSAA or AF either
* and the support VS 2.0+, PS 2.0+ that we already knew about

We don't know yet about that LMAIII (if it really means something) and about the benefit (or penalty) of using FP16 rather than FP32 in the pixel shaders.

Do I miss something?