Post your breaking NV30 news links here!

Those caps on the NV30 look like the manhattan skyline! HUGE freaky things! And tons of them too! And the rear side of the card's loaded down with components too, and there's a big heatSINK (not just spreader) there too that looks copper in color. That will be ONE HEAVY vidcard, that I tell you, an AGP retention clip will be extremely handy when carting around a computer with that baby in it, too bad so few mobo manufacturers invest in those.

The heatsink must mean there's memory on both sides, either DDR2 is not very dense yet (meaning 2x many chips to reach 128MB as standard DDR), or they have 256 MB on those first boards! Looks like it runs way hot too despite BGA format, considering the size of the chunk of metal they have strapped onto there.

I wonder why the board has to be so bloody big (it seems to be even bigger than GF4Ti cards), and why the voltage regulators look so complicated. R300 regulators seem to occupy not even half the board area, and they supposedly deliver close to 50W!

NV30 can't be THAT much more demanding can it?

*G*

 

I'm very leary of the information that has been accumulating today..

"Intellisample" for antialiasing, "adaptive" anisotropic filtering, and totally customizable texture filtering... clever use of these can be a very destructive tool to aid in benchmark fabrication... in creating apples to oranges IQ then comparing FPS ratings or 3DMarks of drastically dissimilar IQ.

Now, more than ever before, will a keen eye on resultant/final image quality be important for comparing this card with other hardware. I'm confident the NV30 based videocards will trounce the GF4, but when all things are made equal (or as close as possible in terms of resultant finished render quality versus different methods of obtaining the same), I fear the spread will not be anywhere near the 4x or touted amounts in this pre-release marketing.

Flexibility in settings is an absolute godsend for gamers- having more tunable and configurable options to balance IQ for performance is wonderful. The only problem here is suggested "default" values websites may use, or how blind benchmarks may be when used to do flat FPS comparisons across product lines. Depending on the range of this configuration, this can lead to wildly inaccurate comparisons- so special care is going to be needed when depicting this product if it's anywhere as customizable as this pre-release information is suggesting.

 

Hrm I knew that.... why was I thinking a 10 layer PCB? I know the geforce cards are presently using 6 layer PCB.. argh.

 

Grall - I can't believe i am going to defend nvidia but that is a best a beta board compared to a fnished production board and so the lay out might change a bit .

 

just read this over at Firingsquad.com
http://firingsquad.gamers.com/hardware/gffx/page6.asp

 

It works outside of AA modes? That seems significant.

From what little I've heard, ATI's implementation appears to be more sub-sample compression than framebuffer color compression. Can ATI actually reduce the bandwidth required for an anti-aliased pixel below that required for a standard, single-sampled, uncompressed pixel? If they could, one would imagine that some low-sample anti-aliasing modes (say, 2x sampling) might even be faster than single-sampled (no compression) modes, under the right conditions.

 

When beyond3d do a review can they please put up links to high quality pictures (ie jpeg compression ratio of 1 in psp7)

 

With the way this thread is going they should have named it "WhippingBoy 2003" instead of merging the geforce and 3dfx names...

Oh well, just more of the same old, same old.

 

At least it shows people light into Nvidia just as much as ATi when they are late.

 

Currently I'm using a Leadtek 4400...

I sure hope the NV30 is as hot as Nvidia and the loyal acolytes say. It is sure sweet to have two wickedly fast optoins! Let's hope Nvidia is not stretching the truth too far.

THIS IS WHAT I WANT TO KNOW THOUGH:

I'm concerned that the NV30 will end up being a macro extension of the fabled speed increases of their Det drivers!

When the rubber really meets the road, I want one of two things to take place:

1 - The NV30 is vindicated and long live the King! This would be cool, because Nvidia is a hot company and makes great boards. They will have shown that when they do move to 256bit memory they will be even faster.

2 - The NV30 demonstrates it isn't as hot as marketed. If this is the case I WANT to have this made clear! I just hate how companies sew FUD and misleading statements and seemingly get away with it because no one connects the dots between early marketing statements (like we have now) and reality.

With what I've read, the NV30 seems like it is as fast as a card running at 500MHz with DDRII. I see no magic rabbits. I see no super pixie dust.

Let's all remember what we hear/read today and compare it to what what the performance really is like.


Steve

 

From what I have read, not really.

For this type of color compression to be effecitve, AFIAK, the "neighboring" pixels / samples must be the same color.

This happens with regularity when doing multisample AA, but not that much under "normal" circumstances. In other words, this type of color compression will have very limited returns in most non AA siutations.

 

look at my new thread for the pixie dust

 

yeah!

 

No company is going to come out and say its product sucks. Let's keep this discussion in reality. As much as marketing hype is annoying, it'd be suicidal for a company to not play up their products.

 

OK, our own preview (rather: wrap-up) is online, too:

http://www.3dcenter.de/artikel/2002/11-19_a.php

Leo did another great (and late-night) job, I may say. I don't agree with all in there, and it's in German, but well worth a read, probably.

ta,
-Sascha.rb

 

Obviously you have never stood next to a B52 bomber...then you will realize sometime bigger is better

Well I agree with Sharkfood. Its great to have options, fast AF/quailty AF and many different AA settings. I just hope reviewers try to do an apples to apples......

 

I just wonder how anyone will do an apples-to-apples comparison of NV30 and R300's FSAA modes. On the one hand, NV30 uses a combination of supersampling and multisampling, so it may have slightly better quality, but R300 has a much better "jittered" grid while NV30 tends to have ordered grid for their subsample placement.

FSAA is a major strength of both high end videocards. How are they to be compared, "apples to apples", when they seem much more like an orange and a pear?

 

OK. B-1 to B-52 Comparrison

Size:
B-52 - 505,000 lbs max at take off, 160' long, 185' wingspan
B-1 - 477,000 lbs max takeoff, 147' long, 136' max wingspan
The B-52 is bigger.

Speed:
B-52 - 509 mph
B-1 - 823 mph at high altitude, 600 mph at low altiude
the B-52 is slower

Bomb load:
B-52 - 50,000 lb convential bombs, 12 JDAM (the weapon most used in Afghanistan)
B-1 - 72,000 lb convential bombs, 24 JDAM
Despite being smaller, the B-1 can carry more bombs, up to 100% more.

Crew members: (fewer means fewer people at risk)
B-52 - 6
B-1 - 4
Fewer Airmen are at risk on B-1 missions

My analysis is that despite the B-1 is smaller than the B-52, the B-1 is in fact better. Hence, bigger is not always better.

 

Learning new things is so much fun.

 

The B-1 is substantially supersonic @ sea level.