John Carmack's official comments on NV40 (GeForce 6800 family)

"John Reynolds" wrote in message
...

The #s speak for themselves, but against what? Last generation's
products?
Well, I would certainly expect a next gen. product to outperform last gen.
parts, but I'm just funny that way.

LOL. ;-)

--
Paul Smith,
Yeovil, UK.
http://www.smirnov.demon.co.uk/
http://www.doom3portal.com/ A Doom 3 fansite.

*Replace nospam with smirnov to reply by e-mail*

Hey John, FYI a purist troll doesn't reply to the original post.

"John Lewis" wrote in message
...
On Thu, 15 Apr 2004 22:59:53 GMT, "John Reynolds"
wrote:

Your post would be. . .hmmm, what's the word. . .more legit if you
weren't
coming off as a nVidia fanboy flaming away at Valve, John. Newell simply
voiced what every developer knew about the FX parts: they sucked at
running
DX9 code at floating point precision. Hell, these NV40 previews show
that
more clearly than anything else. And what do Carmack and Newell have in
common? Their companies' new engines both required special code paths to
get good performance out of FX boards? Think about that, John. Oh, and
for
Far Cry whether those new screenshots require SM 3.0 support is still up
in
the air. I've heard they're created using offset mapping, not vertex
texturing; this was written by Democoder, the guy who got that Unreal 3
engine movie and some Far Cry shots from yesterday (he's a regular poster
at
B3D).

Anyways, the 6800U looks like a very impressive part. The only real
negatives are the power consumption/heat

the NV40 GPU consumes ~ 25 watts more than the NV35 or R350.
The whole board consumes a max of 110 watts. Compare the
Prescott 3.4 CPU @ 103 watts max. ( Northwood 3.4, 89 watts )


and the fact that both AA and AF
could be better.

In what way... please be specific...

It'll be interesting to see if the R420 from ATI can
compete

Let's hope that they have a VPU on board that is competitive with
that in the NV40. For professional video applications, that feature
is almost as important as the graphics-engine features.


Anthony "Reverend" Tan just quoted Tim Sweeney on B3D's board and
Tim said, about R420, that "It rocks!". This next generation is
definitely
going to be much more interesting than last year's, that's for sure.

John "fanboys suck" Reynolds

aka John "Ati fanboy now, past- 3dfx and nVidia fan-boy" Reynolds.




John Lewis

Mike P wrote:
the NV40 GPU consumes ~ 25 watts more than the NV35 or R350.
The whole board consumes a max of 110 watts. Compare the
Prescott 3.4 CPU @ 103 watts max. ( Northwood 3.4, 89 watts )

Holy ****.

1000 Watt PSU here we come.


--

^..^
Bernard
www.cs.uwa.edu.au/~langhb01

"bp" wrote in message
...


So how much are these new super cards going to cost ?
Twice the cost of the PC I install it in ?

EST prices: $500 for Ultra version. $300 for non-Ultra (12 pipelines vs
16, I think). oh, and there's suppose to be a 512mb version coming out
later on.

regards
Jonathan

"K" wrote in message
news
On Fri, 16 Apr 2004 16:35:08 +0000, Pluvious wrote:


Actually I was rather amused to see the preview at HardOCP
(http://www.hardocp.com/article.html?art=NjA2). The performance is
now-where near where I expected. 10-15 fps better then the ATI 9800 XT
is a joke. Of course I expect the 3rd party companies to juice up the
card a bit. We'll see.. but my money is on ATI for retaining the crown
this next generation.



It's a bad review that's typical of those ****wits at HardOCP. It's not
that they're biased, just that most of their reviews have more holes in
them than a Swiss cheese. What they are trying, and totally failing to do,
is set each card at it's maximum settings that will produce similar
results and get a feel for 'the overall gaming experience'. For example
they set the GF6800 to 1600x1200 with x8AF and 4xAA and compare it to the
9800XT at 1280x1024 with no AA or AF. So what happens is that many people
like yourself will quickly scan over the graphs and think 'hmm, the GF
isn't much faster than the Radeon.' I suggest you check out Anandtech if
you want credible reviews.

K

Hardcore does a WARNING right up front about their benchmarks not being
apples to apples. But then why do the test then? So that the reader could
find the highest possible rez to play the game on.

Sort of like taking two pickup trucks, dumping 16,000lb load in one, and
only 2,000lb in the other, and having both of them race to the finish line.
The one with the 16,000lb load get's there 2 msecs ahead of the other.
2msecs doesn't seem like much until you realize the enormous weight that it
had to haul. The test isn't trying to find the fastest truck, but rather
which truck can carry the most and still get across the finish line in
acceptable time.

regards
Jonathan