PDA

View Full Version : 16 headed beast?


Pages : [1] 2 3 4 5 6

Skinner
02-26-04, 11:55 AM
Nvidia NV40 specifications confirmed

Too late for ATI to respond


By Paul Dutton: donderdag 26 februari 2004, 14:52

NVIDIA REALLY needed to pull the cat out of the bag in the next round of its increasingly epic battle with ATI Technologies and from where we’re sitting we’d be more than surprised if the next round isn’t, now, a foregone conclusion.
US sources close to Nvidia have confirmed that their next generation GPU will feature a full sixteen pipelines – not as previously speculated an 8x2 arrangement - and this is reflected in the increased transistor count of circa 205-210 million, up from a previously speculated 175 million.

In very recent times ATI have been extremely confident in saying that their R420 - set to launch not long after, Nvidia show their hand – would, immaterial of architecture, thoroughly outgun the NV40.

Indeed today, Richard Huddy - ATI's European Developer Relations Manager maintained that their expectation was for ATi’s R420 to be, faster, on balance, across a suite of 10 common games than Nvidia NV40.

However we think that ATI's earlier confidence was based on internal intelligence that Nvidia’s NV40 feature set would be limited to an architecture built on 175 million transistors and that Nvidia would deliver, as anticipated, on the first day of CeBIT in March.

Nvidia's launch date for NV40 seems to have shifted backwards, but even a month or so is not going to give ATi enough time for a hardware response.

Knowing what we know now about NV40 having a full sixteen pipelines, we sensed that despite some fighting talk, and indeed some intelligent and rationalised arguments, ATi are now hoping, at best for a close fight…

NV40 0wnage of ATi R420 coming soon to a store near you?

We sure think so… µ


http://www.theinquirer.net/?article=14373

Are this 16 real pipes, or NV3x pipes ;)

Paul
02-26-04, 11:58 AM
Indeed :cool:

I would hold off on assuming this is correct, though. Everything that has been before points to it being wrong - although maybe that's the point.

DMA
02-26-04, 12:22 PM
Wasn't NV40 set to launch in Nov last year or something like that? Maybe they trashed that design and did a little makeover? :)
The rumors about NV40 being a 175mil/8x2 might be the "old" NV40? Just a thought. :D

Anyway, i'll believe it when i see it.

:afro:

Malfunction
02-26-04, 12:24 PM
Can't wait to hear what 3D Center will say if it is true. Wonder what new goodies will emerge as a result of not just going 175 million transistors, but up to possibly 205 to 210 million! Hopefully if this is true, the people who do enjoy nvidia products should have that New AA algorithm as well. :thumbsup:

I'll celebrate now till it is unconfirmed. :D

Peace,

:afro:

Paul
02-26-04, 12:43 PM
Originally posted by DMA
Wasn't NV40 set to launch in Nov last year or something like that? Maybe they trashed that design and did a little makeover? :)
The rumors about NV40 being a 175mil/8x2 might be the "old" NV40? Just a thought. :D

Well, the NV45 is what will be released now, effectively. So there is an "old" NV40, yeah.

retsam
02-26-04, 12:57 PM
if these rumors turn out to be true nvidia would turn out to be the most amazing engineering company ever !!!! how could any company pull off somthing like this is beyond me ..... hmmmm now that i think of it, this has to be a whole lot of bs ..... no company is that good .....



rets

Ady
02-26-04, 01:18 PM
I smell PR BS. Isn't the nv3x something like 130 million Transistors? And that is only really a 4 pipeline card. I can't see that happening. It's probably 16pipes the same as nv30 was 8 pipes. :rolleyes:

retsam
02-26-04, 01:22 PM
ok now that i have had time to read the artical i smell nvda pr bull ........mooooooooooo

SurfMonkey
02-26-04, 02:15 PM
I think just judging on the number of transistors is misleading. After all the R300 is a full 8x1 design and has a lower transistor count than the NV30. How many parts of the NV30 were foobar? And which parts will they have ditched entirely to make way for something new and improved?

Still a 16x1 design is going to be a monsterous GPU, hot and heavy. NVidia may have payed a canny game this time by not boasting about what they have and, in fact, have done just the opposite. The art of mis-direction. Then again maybe they're just telling huge porkies and it's going to be a 16x1 zixels\8x2 pixel design.

Skinner
02-26-04, 03:39 PM
Originally posted by SurfMonkey
I think just judging on the number of transistors is misleading. .

Yep, and the way of counting too, I read on b3b:

When I spoke to engineers at both ATI and Nvidia, their response was to not really understand what management at Nvidia were smoking, but the best explanation I got later was from someone senior at ATI who told me that, as near as they could tell, Nvidia were counting gates in things like caches and including that in the transistor count, while ATI only counted transistors directly involved in logic. Apples to oranges.]

CapsLock
02-26-04, 04:40 PM
Originally posted by Skinner

quote:
--------------------------------------------------------------------------------
Originally posted by SurfMonkey
I think just judging on the number of transistors is misleading. .
--------------------------------------------------------------------------------



Yep, and the way of counting too, I read on b3b:


quote:
--------------------------------------------------------------------------------
When I spoke to engineers at both ATI and Nvidia, their response was to not really understand what management at Nvidia were smoking, but the best explanation I got later was from someone senior at ATI who told me that, as near as they could tell, Nvidia were counting gates in things like caches and including that in the transistor count, while ATI only counted transistors directly involved in logic. Apples to oranges.]


And don't forget NV like to count the transistors that thier engineers dream about too. :rolleyes:

Looks like old tricks again, imaginary pipes, imaginary transistors, imaginary benchmarks, what next?: HEY! NV! GROW UP!

Caps

Nutty
02-26-04, 05:26 PM
Looks like old tricks again, imaginary pipes, imaginary transistors, imaginary benchmarks, what next?: HEY! NV! GROW UP!


I think your BS allegations are getting a little ahead of themselves. NV is a great company, they ruled the performance crown for many years, is it soo hard to believe they put their mistakes with the NV3X architecture behind them?

Lets wait until they hit the reviews before ridiculing them eh?

PreservedSwine
02-26-04, 06:20 PM
Originally posted by Nutty
I think your BS allegations are getting a little ahead of themselves. NV is a great company, they ruled the performance crown for many years, is it soo hard to believe they put their mistakes with the NV3X architecture behind them?

Lets wait until they hit the reviews before ridiculing them eh?

I agree, but then let us also wait until the reviews hit before we beleive they actually have 200+ million trans and 16 pipes....

Ninja Prime
02-26-04, 07:37 PM
First, there is no way its 210 million. The die would be the size of texas since its on 130nm. I mean, NV35 has a massive die, and its only 130 million. Second, that 16x0 wackness Uttar was talking about a while ago could still be claimed to be "16 pipes", so don't get too excited.

We will see soon enough.

MUYA
02-26-04, 07:50 PM
Originally posted by Ninja Prime
First, there is no way its 210 million. The die would be the size of texas since its on 130nm. I mean, NV35 has a massive die, and its only 130 million. Second, that 16x0 wackness Uttar was talking about a while ago could still be claimed to be "16 pipes", so don't get too excited.

We will see soon enough.

I dunno the dies for previous GPUs/CPUS on 150nm, 180nm or 250 m, were massive too. It just means, NV with a huge die like that will cost more to produce ie less dies per wafer, not imposible but maybe the benefits (or not) maybe related to thermal characteristics, ie more area to spread the heat generated evenly allowing for greater clocks or, it could be the best way to achieve nulcear fusion.

bkswaney
02-26-04, 07:54 PM
If you look at past inside info on cards from the Inq they have
always hit the nail on the head.
They have d@mn good inside info on some things.
Go back and look at old stuff. Most come to pass. ;)

So I say nvidia just gave out bad info to throw ATI off
what they were really doing.
If this is true The R420 is dead before it even hits the market.
A 16 pipe NV40 will have ati for lunch.

I'm sure Nvidia is sick of being second to ATI.
Since the R3xx come out ATI has ruled.

It's time for nvidia to be on top again.

Razor04
02-26-04, 08:04 PM
Originally posted by bkswaney
If you look at past inside info on cards from the Inq they have
always hit the nail on the head.
They have d@mn good inside info on some things.
Go back and look at old stuff. Most come to pass. ;)

So I say nvidia just gave out bad info to throw ATI off
what they were really doing.
If this is true The R420 is dead before it even hits the market.
A 16 pipe NV40 will have ati for lunch.

I'm sure Nvidia is sick of being second to ATI.
Since the R3xx come out ATI has ruled.

It's time for nvidia to be on top again.
Off topic... With over 1500 posts why oh why haven't you figured out what text wrapping is...hell even NV40 figured it out.

On topic... My personal feeling is that this is marketing BS. I will wait to make a final decision on which card is better after they are both out and places like Beyond3D have dissected them and figured out what they actually are and thrown out the PR BS.

sandeep
02-26-04, 08:42 PM
NVIDIA Corporation (Nasdaq: NVDA) today confirmed that it will be one of the first semiconductor companies to manufacture select up-coming graphics processing units (GPUs) at Taiwan Semiconductor Manufacturing Company’s (TSMC’s) (TAIEX: 2330, NYSE: TSMC) 0.11 ěm (micron) process technology. NVIDIA will combine TSMC’s 0.11 micron process with its own innovative engineering designs, to deliver high-performance and low-power consumption in a graphics processor.

CapsLock
02-26-04, 08:51 PM
Originally posted by CapsLock
And don't forget NV like to count the transistors that thier engineers dream about too.

Looks like old tricks again, imaginary pipes, imaginary transistors, imaginary benchmarks, what next?: HEY! NV! GROW UP!

Caps

Originally posted by Nutty
I think your BS allegations are getting a little ahead of themselves. NV is a great company, they ruled the performance crown for many years, is it soo hard to believe they put their mistakes with the NV3X architecture behind them?

Lets wait until they hit the reviews before ridiculing them eh?

BS allegations? :wtf:

What planet have you been living on for the past year?

NV may have been a great company in the past, but they certainly haven't been a great company since they made consumer fraud thier most popular marketing strategy a while ago. Frankly, my animosity towards them has nothing to do with the NV30 architecture.

OH, sorry. I should have added imaginary trilinear too. :rofl:

CapsLock

MuFu
02-26-04, 08:59 PM
16 pixel pipes (16x1)
32 z-pixel only/clk :eek:
~475/800MHz
210 million transistors
F*ckin' stupid cooling
Even more ridiculous price tag.

'tis what I believe ATPIT, thanks to various mutterings. Sounds like NV40 was scrapped and reborn as an AGP version of NV45. :)

MuFu.

CapsLock
02-26-04, 09:14 PM
Originally posted by MuFu
16 pixel pipes (16x1)
32 z-pixel only/clk :eek:
~475/800MHz
210 million transistors
F*ckin' stupid cooling
Even more ridiculous price tag.

'tis what I believe ATPIT, thanks to various mutterings. Sounds like NV40 was scrapped and reborn as an AGP version of NV45. :)

MuFu.

As I just pointed out in the NV40 vs. R420 poll thread - 210 million transistors is going to be damn hard to clock high. We're talking 75 million MORE transistors on one chip. IF its true I would think 400 Mhz is doing good. Which makes a rather big difference when it was originally touted at 500 - 600 MHz.

I guess thats just a guess though. :confused: (Although Mariner at B3D also used that number.)

Caps

MuFu
02-26-04, 09:24 PM
Sounds like they're aiming for 475-500Mhz.

Originally posted by CapsLock
Which makes a rather big difference when it was originally touted at 500 - 600 MHz.

Yes, but we suspected 8x1 and a lower transistor count (which is probably what NV40 originally was). 16x1@475MHz has 160% the pixel fillrate of 8x1@600MHz.

MuFu.

MUYA
02-26-04, 09:32 PM
Originally posted by MuFu

Even more ridiculous price tag.


Would u have to mortage your house or something? ;)

$499? $599? $699? - NV arranges flexi pay scheme with $40 payments over 2 years?


man..... i sure Hope NV subsidizes the top range even after supposedly including a free copy of Doom III :D

CapsLock
02-26-04, 09:52 PM
Originally posted by MuFu
Sounds like they're aiming for 475-500Mhz.



Yes, but we suspected 8x1 and a lower transistor count (which is probably what NV40 originally was). 16x1@475MHz has 160% the pixel fillrate of 8x1@600MHz.

MuFu.

I won't believe it till I see it. 75 million more transistors AND the same clock as the NV35. pffft, ya right.

Caps

MUYA
02-26-04, 10:00 PM
Does Intel's Itanic have filty amounts of transistors and yet they run quite high? 1GHz+??

400-500Mhz seems very good ball park figure.

Sure the die will be big, but thats not neccesarily a bad thing. They can arrange the transistors in the die to minimise heat spots. That will lead to maybe better thermal characteristics. Ie be able to tranfer heat off directly to the heat spreader etc? I dunno just a thought. The cooling for it will probably be like the dustbsuter though. And OCing is proably not an option? who knows....not long left