16 pipes for R420 too?

[whql]

http://www.pcgames.de/?article_id=303000

[Hellbinder]

*snicker* :twisted:

[digitalwanderer]

What a totally crappy time to be trying to stay low-key to conserve bandwidth and to have a stinking splint on the middle finger of me right hand that makes typing a BIATCH! :(

I'm abstaining on the couch eating popcorn for this one I'm afraid. ;)

EDITED BITS: I'll contribute the babbleation of it:

08.04.2004 17:42 o'clock - like our editorship out well informed circles, Ati experienced still in the spring a NEXT generation diagram map with 16 pipelines on the market will bring. Thus the Canadians put a high measuring pole for the NEXT generation diagram map of Nvidia, which is introduced on 14 April (we reported). However are no further details admits (exact erscheinungstermin, organization of the pipelines, Shaderleistungen). Ati could not be induced to a statement.

[Evildeus]

Hehe this battle will be interesting 8)

[digitalwanderer]

Hehe this battle will be interesting 8)

Methinks that is an extreme understatement! :shock:

[Dave Baumann]

"A reliable industry source" has confirmed this to be the case.

[Hellbinder]

*Snicker* :twisted:

[I.S.T.]

"A reliable industry source" has confirmed this to be the case.


YAY! :D :D :D

I hope the R420 does have 16 pipes. It'll make this round more interesting.... ;)

[Mindriot]

Is it actually possible that those damn canadian have been "lying in the weeds" so to speak?

:o :?: :shock:

The anticipation is killing me!

And Binder: knock that off...don't make me come over there and pummel you with potatoes. :wink:

[Hellbinder]

hehehe Sorry... ;)

Dig, Remember what i mentioned at Nvnews waaay back in the day? ;)

[PaulS]

This is only new news if the initial release is 16 pipes, rather than 12. A 16 pipe chip was always on the cards once they had enough chips capable of it, as has been noted in numerous threads and on several websites.

[T2k]

"A reliable industry source" has confirmed this to be the case.

Dave! I told you don't tell anyone! :lol:

[kemosabe]

This is only new news if the initial release is 16 pipes, rather than 12. A 16 pipe chip was always on the cards once they had enough chips capable of it, as has been noted in numerous threads and on several websites.

Indeed. Dave, what, if any, indication has the "reliable industry source" given in this regard? :twisted:

[digitalwanderer]

hehehe Sorry... ;)

Dig, Remember what i mentioned at Nvnews waaay back in the day? ;)

I remember that post like it's etched in words of fire on my brain Hellbinder, I sensed the truth in it that day and have been pinning my hopes on it ever since.

Good, COOL BEANS!!!! 8)

Is it actually possible that those damn canadian have been "lying in the weeds" so to speak?

Yes, big time. ATi has been pushing FUD just as bad as nVidia has, but it's been of a different kind of misleading.

I get it, my hunches are right...this is going to be good! http://instagiber.net/smiliesdotcom/otn/happy/ylsmoke.gif

It's so damned good in fact that I'm going to go celebrate and take some time off and go out and play with my son. 8)

[I.S.T.]

Um, what post in Hellbinder refering to? :oops:

[kemosabe]

Here's some more juice.

http://www.xbitlabs.com/news/video/display/20040408100025.html

[Joe DeFuria]

I've said all along that if R420 is designed for 4 quads with "one turned off" to get yields for a "typical" high end product, that ATI should trot out a full R420 RE ("Reviewer edition" ;) ), just to put their best foot forward in terms of the PR battle.

Assuming the 16 pipeline rumor is indeed correct, it will be most interesting to see

1) If the X800 "XT" is 16 pipe, vs. X800 "Pro" being only 12
2) What the price / availability will be of the "XT".

If both the Pro and the XT are 16 pipe cards (just different clocks), and are as readily available as today's high end cards are...then ATI would have succeeded in something pretty remarkable....keeping it secret for so long. :)

[MuFu]

Hehe, sounds good. Can't ever remember this being confirmed though...

A 16 pipe chip was always on the cards once they had enough chips capable of it...

Thought that was still a point of uncertainty until just recently. :?:

[Joe DeFuria]

Here's some more juice.

http://www.xbitlabs.com/news/video/display/20040408100025.html

Gotta love how they claim the "specifications have been boosted", when specifications have never been released. :)

[mreman4k]

Here's some more juice.

http://www.xbitlabs.com/news/video/display/20040408100025.html

Article says ATI could have competed with NVidia''s part with only 8 pipes.... :o :shock:

Initially the company believed that 8 pixel conveyers would be enough to stay competitive against the rivaling NVIDIA NV40 chip, however, at this point ATI wants to expose all capabilities of the RADEON X800-series chips to fight with the competitor.


:shock:

[digitalwanderer]

The rumors are true; ATi is not going conservative this time.

This time they're going balls out, this is going to be fun! :D

[Joe DeFuria]

Article says ATI could have comptered with NVidia''s part with only 8 pipes.... :o :shock:

??

It doesn't say that at all, IMO.

I read that as "ATI would have liked to be able to compete with NV40 using fewer than 16 pipes, but have since decided to use all 16 from the get-go."

There's no real direct indication if ATI believes it would have compared one way or the other.

That being said, I expect one ofthe following to be true:

1) ATI believes they need all 16 pipes enabled to compete with the initial NV40.

or

2) They think a 12 or 8 pipe R420 would have been competitive, and planned to "refresh" it with a "16 pipe" R420 this fall. However, If the R500 is in fact looking to be ready for the fall, (as some rumors of a "R500 sooner than expected" are starting to circulate) then ATI has little incentive to "save" 16 pipe performance, and might as well go balls-out now.

[Joe DeFuria]

The rumors are true; ATi is not going conservative this time.

This time they're going balls out, this is going to be fun! :D

I just hope that this holds true for memory clocks as well :!:

And yes, was it Tank or Dozer who said:

"Mmmmm....This is an exciting time!"

[tEd]

Could it be that

16pipes=XT
12pipes=pro

[Evildeus]

Interesting to see this "close source to Ati" giving this information to every web site in the world. Is Ati afraid? Some FUD? Thought that wasn't Ati's way of doing things :roll:
http://www.clubic.com/n/n12244.html

[PatrickL]

Lol evildeus, i knew ATI should be guilty for something :P

You prefectly know the value you can give to clubic claims...

[Nick Spolec]

Interesting to see this "close source to Ati" giving this information to every web site in the world. Is Ati afraid? Some FUD? Thought that wasn't Ati's way of doing things :roll:
http://www.clubic.com/n/n12244.html

I wouldn't say afraid, I would just say concerned. They really have no choice. They have to be getting the information out.

Nvidia has been spreading around 16x1 (and 32x0), and there is still the old rumor that R420 is 12x1. Everyone is jumping on Nvidia's bangwagon because of it. Neither product has launched, no numbers have been confirmed, and already ATI is losing ground.

[kemosabe]

Could it be that

16pipes=XT
12pipes=pro

Remains in the realm of possibilities, but would that not introduce a far more significant performance gap between Pro and XT models than has been the case for the R3XX generation? A less-than-16-pipeline R420 configuration might just redefine the upper midrange segment, however, and diminish the upcoming huge performance gap between the high end and mainstream.

RV280 and mostly RV370 = low end
RV380 = mainstream
R420 (8 or 12-pipe?) = performance
R420 (16-pipe) Pro and XT = enthusiast

:?:

[Evildeus]

Lol evildeus, i knew ATI should be guilty for something :P
It's not directed to Ati but to people saying that Ati is not doing that.

You prefectly know the value you can give to clubic claims...
No i don't. Clubic has got some good track on scoops, and really bad on tests ;)

[John Reynolds]

"Mmmmm....This is an exciting time!"

Yeah, but that was right before he asked for too much money and got his ass cut. :twisted:

[Joe DeFuria]

"Mmmmm....This is an exciting time!"

Yeah, but that was right before he asked for too much money and got his ass cut. :twisted:

Hmmm...we appear to not be thinking of the same Tank / Dozer...The Matrix. ;)

BTW...still (sincerly) awaiting your apology.

[Ragemare]

Seems to good to be true, although I think it would make sense if it was, there's no proper evidense of that it is so I'm going to remain skepticle instead of risk being wrong. :)

[Joe DeFuria]

It's not directed to Ati but to people saying that Ati is not doing that.

ATI is Not doing what? Keeping stuff close to their vest until right before launch?

I'll tell you what, you can criticize ATI "for doing what nVidia does" if they:

* Supply web sites and allow them to publish results with a "Secret card with non confired specs" with "secret, not public drivers" before said card is officially launched, to benchmark against a publically launched NV40.

[Sxotty]

*snicker* :twisted:
The only thing that bothers be is that Nvidia at is core is still a bunch of megalomaniac Evil doers, Scoundrels and liars. Which has basically been my beef with them for a few years now.
Hellbinder you are at heart a cheerleader it is to bad you were not born a girl.

Still there were many that thought that a 16pipe nv40 would have trouble beating an 8 pipe r420, so the answer remains I don't care how many pipes it has I care how well it plays games and how much it costs.

[Fodder]

Hmmm...we appear to not be thinking of the same Tank / Dozer...The Matrix. ;)
One of them got killed in the first, the other mysteriously vanished before the sequel after a pay dispute.

[John Reynolds]

Hmmm...we appear to not be thinking of the same Tank / Dozer...The Matrix. ;)

I was referring to the actor who played the operator (Tank or Dozer, can't remember which was his name). He reportedly asked for a rather substantial raise and thereby got himself removed from the sequels.

[Joe DeFuria]

Hmmm...we appear to not be thinking of the same Tank / Dozer...The Matrix. ;)

I was referring to the actor who played the operator (Tank or Dozer, can't remember which was his name). He reportedly asked for a rather substantial raise and thereby got himself removed from the sequels.

Ahh...OK.

[MistaPi]

Remains in the realm of possibilities, but would that not introduce a far more significant performance gap between Pro and XT models than has been the case for the R3XX generation?

My thoughts exactly. It would be quite a difference if the 12 pipeline Pro is clocked 450 while the 16 pipeline XT is clocked 500Mhz.

[Doomtrooper]

Need to update the rumour mill thread, now all the fun is gone :( :wink:

[Joe DeFuria]

Remains in the realm of possibilities, but would that not introduce a far more significant performance gap between Pro and XT models than has been the case for the R3XX generation?

My thoughts exactly. It would be quite a difference if the 12 pipeline Pro is clocked 450 while the 16 pipeline XT is clocked 500Mhz.

Whether or not there is a huge difference all depends on the specific clocks of both the core and memory.

What if the clocks are the other way: a supposed 12 pipeline core is clocked at 500, and the 16 pipe at 400?

[plat]

something i find interesting is.

According to previously released information, the ATI code-named R420/R423 graphics processors will contain about 160 million of transistors and will be produced using low-k 0.13 micron technology at TSMC.

when nvidia said they had a 16 pipeline card and the number of transistors was only 175 million people said "NO WAY!" and even when it came out that the nv40 had 210 million people still said that wasn't enough. now the r420 only has 160 million and has 16 pipelines?

[anaqer]

something i find interesting is.

Good point... if you don't take into account the idea of nV and ATi counting transistors differently.

[PaulS]

That being said, I expect one ofthe following to be true:

1) ATI believes they need all 16 pipes enabled to compete with the initial NV40.

or

2) They think a 12 or 8 pipe R420 would have been competitive, and planned to "refresh" it with a "16 pipe" R420 this fall. However, If the R500 is in fact looking to be ready for the fall, (as some rumors of a "R500 sooner than expected" are starting to circulate) then ATI has little incentive to "save" 16 pipe performance, and might as well go balls-out now.

Almost certainly the first - if a 12 pipe card would have been enough, they wouldn't lower their yields by needlessly releasing 16 pipe cards. They're releasing what they need to release in order to compete/win.

Hehe, sounds good. Can't ever remember this being confirmed though...

A 16 pipe chip was always on the cards once they had enough chips capable of it...

Thought that was still a point of uncertainty until just recently. :?:

Well, it's certainly been pretty clear since the 12 out of 16 pipes rumour came about, which was a little while ago now. The assumption, however, was that the last quad would only be activated in a refresh.

[kemosabe]

Remains in the realm of possibilities, but would that not introduce a far more significant performance gap between Pro and XT models than has been the case for the R3XX generation?

My thoughts exactly. It would be quite a difference if the 12 pipeline Pro is clocked 450 while the 16 pipeline XT is clocked 500Mhz.

Whether or not there is a huge difference all depends on the specific clocks of both the core and memory.

What if the clocks are the other way: a supposed 12 pipeline core is clocked at 500, and the 16 pipe at 400?

Joe, can we not draw some inferences from the 9500 Pro and 9600 Pro? The former had 4 more pipelines, but was clocked quite a bit lower (275/540 if I recall) than the latter (400/600). Yet the performance difference remained quite significant if memory serves me. Then again, that might provide more justification for $100 price difference (compared to 9800 Pro vs. 9800XT, for instance). Moreover, the difference between 12 and 16 pipes might not correlate as well as 8 vs. 4. I still can't see a balanced product line if the high-end part has 16 pipes and then it drops to 4 (RV380), regardless of core/memory clocks. Seems like a huge hole to fill there.

[MistaPi]

Whether or not there is a huge difference all depends on the specific clocks of both the core and memory.

What if the clocks are the other way: a supposed 12 pipeline core is clocked at 500, and the 16 pipe at 400?

Yes ofcourse, but I doubt the XT will be clocked lower than the Pro version.

[Fodder]

I still can't see a balanced product line if the high-end part has 16 pipes and then it drops to 4 (RV380), regardless of core/memory clocks. Seems like a huge hole to fill there.
Keep the 9800 around until RV410 is ready, and hope NV41 doesn't land early.

[Joe DeFuria]

something i find interesting is.

[quote]when nvidia said they had a 16 pipeline card and the number of transistors was only 175 million people said "NO WAY!"

Not interesting at all.

The nVidia situation would be nVidia going from a 4x2 design, to a 16x1 design. In other words, thats more than "double" the pipelines, and probably closer to quadrupling them.

[Joe DeFuria]

Whether or not there is a huge difference all depends on the specific clocks of both the core and memory.

What if the clocks are the other way: a supposed 12 pipeline core is clocked at 500, and the 16 pipe at 400?

Yes ofcourse, but I doubt the XT will be clocked lower than the Pro version.

Why not? IF one of the differences between the XT and the PRO is 16 vs. 12 pipelines, why should one expect the clock rate to be higher?

[surfhurleydude]

Well, if this isn't a surprising bit of information...

If this rumor is true, then it also must confirm another thing - that the NV40 isn't simply the load of FUD some people thought it to be. Otherwise, ATI wouldn't need to pull out a 16 pipe card.

Nevertheless, great news... It seems to look more and more and more like whoever takes the crown won't necessarily mean the card will be the "best" card. From these latest rumors, it looks like performance between the NV40 and R420 is going to be VERY VERY close... And because the typical most important factor is going to be even, I believe some of the smaller, less important things will now play a bigger part in which card someone decides to get...
I'm sure some people didn't have a problem with NV40's rumored power requirements, as it was also rumored to have enough of a speed advantage over the R420 to warrant the power... Now things look about even and things like molex connectors, heat added to case temperature, stability, and other factors will contribute to purchasing instead of raw speed and image quality.

[surfhurleydude]

Whether or not there is a huge difference all depends on the specific clocks of both the core and memory.

What if the clocks are the other way: a supposed 12 pipeline core is clocked at 500, and the 16 pipe at 400?

Yes ofcourse, but I doubt the XT will be clocked lower than the Pro version.

Why not? IF one of the differences between the XT and the PRO is 16 vs. 12 pipelines, why should one expect the clock rate to be higher?

Agreed, 12 pipelines vs 16 can be compared to GeForce FX 5800 Ultra vs 5900 Ultra... lower clockspeeds, but because of architectural advantages, more speed overall.

[Joe DeFuria]

I still can't see a balanced product line if the high-end part has 16 pipes and then it drops to 4 (RV380), regardless of core/memory clocks. Seems like a huge hole to fill there.

Which could be nicely filled in the interim by remaining R350 and R360 stock, no?

4 pipelines at <$179 and lower. 8 pipelines (R360) at $180-$250, and 12/16 pipelines (Xt/Pro at $300+)

The question is, how much stock of R350/R360 boards is still out there, and did this factor into any decision on how to "configure" the R420 for initial sale?

I can easily envision the R420 "SE" (perhaps R420 with TWO quads disabled, and on 128 bit bus) replacing the R350/R360...but only after the OEMS have depleted the R350/R360 stock to some pre-determined level.

[surfhurleydude]

I still can't see a balanced product line if the high-end part has 16 pipes and then it drops to 4 (RV380), regardless of core/memory clocks. Seems like a huge hole to fill there.

Which could be nicely filled in the interim by remaining R350 and R360 stock, no?

4 pipelines at <$179 and lower. 8 pipelines (R360) at $180-$250, and 12/16 pipelines (Xt/Pro at $300+)

The question is, how much stock of R350/R360 boards is still out there, and did this factor into any decision on how to "configure" the R420 for initial sale?

I can easily envision the R420 "SE" (perhaps R420 with TWO quads disabled, and on 128 bit bus) replacing the R350/R360...but only after the OEMS have depleted the R350/R360 stock to some pre-determined level.

I really can't see this happening, as ATI tried it once with the Radeon 9500 and 9500 Pro... While they WERE excellent cards, we all saw what ATi did with them... ditched them for a card specifically designed for the mid to low end market. I honestly can't see them returning to that old method once again, I believe the reason they made the 9600 series in the first place was to make a larger profit than selling disabled R300 cores, correct?

[MistaPi]

Yes ofcourse, but I doubt the XT will be clocked lower than the Pro version.

Why not? IF one of the differences between the XT and the PRO is 16 vs. 12 pipelines, why should one expect the clock rate to be higher?

I think it will be confusing for the public if the XT is clocked lower. Perhaps it will not be clocked higher than the PRO version but I do doubt it will be clocked lower. This is at least my personal opinion. I am not saying it WONT be clocked lower (IF the XT=16 pipelines and PRO=12 pipelines is the case).

[surfhurleydude]

Yes ofcourse, but I doubt the XT will be clocked lower than the Pro version.

Why not? IF one of the differences between the XT and the PRO is 16 vs. 12 pipelines, why should one expect the clock rate to be higher?

I think it will be confusing for the public if the XT is clocked lower. Perhaps it will not be clocked higher than the PRO version but I do doubt it will be clocked lower. This is at least my personal opinion. I am not saying it WONT be clocked lower (IF the XT=16 pipelines and PRO=12 pipelines is the case).

It will not be confusing whatsoever... Remember these cards are not intended to be sold to the "average joe" in the first place (although that does not say the average joe doesn't purchase them). By your logic, the GeForce FX 5800 Ultra should have been confused as a better card than the GeForce FX 5900 Ultra because its clockspeeds were higher... And I don't believe ANYONE confused the two cards.

[digitalwanderer]

Still there were many that thought that a 16pipe nv40 would have trouble beating an 8 pipe r420, so the answer remains I don't care how many pipes it has I care how well it plays games and how much it costs.
Yup, after all is said and done that's gonna be all it ends up being in the end for most of us. :)

Well, if this isn't a surprising bit of information...

If this rumor is true, then it also must confirm another thing - that the NV40 isn't simply the load of FUD some people thought it to be. Otherwise, ATI wouldn't need to pull out a 16 pipe card.
I don't think that's necessarily the case. My usual "reliable" sources have all been going thru some tizzies them ownselves having trouble with changing rumors/stats/design features on the nVidia side of things, I'm really thinking this is a case of both IHV's not quite being sure up until the last possible second of what they are releasing...and I fear nVidia might have once again went a bit overboard on the FUD which caused ATi to get more aggressive which in turn boxed nVidia into going even a bit more since they'd talked themselves into a corner. 8)

For my proof of that, I offer the latest "dustbuster2" rumor going around this morning in lieu of the elegant single-slot solution being rumored earlier in the week. ;)

[surfhurleydude]

Still there were many that thought that a 16pipe nv40 would have trouble beating an 8 pipe r420, so the answer remains I don't care how many pipes it has I care how well it plays games and how much it costs.
Yup, after all is said and done that's gonna be all it ends up being in the end for most of us. :)

Well, if this isn't a surprising bit of information...

If this rumor is true, then it also must confirm another thing - that the NV40 isn't simply the load of FUD some people thought it to be. Otherwise, ATI wouldn't need to pull out a 16 pipe card.
I don't think that's necessarily the case. My usual "reliable" sources have all been going thru some tizzies them ownselves having trouble with changing rumors/stats/design features on the nVidia side of things, I'm really thinking this is a case of both IHV's not quite being sure up until the last possible second of what they are releasing...and I fear nVidia might have once again went a bit overboard on the FUD which caused ATi to get more aggressive which in turn boxed nVidia into going even a bit more since they'd talked themselves into a corner. 8)

For my proof of that, I offer the latest "dustbuster2" rumor going around this morning in lieu of the elegant single-slot solution being rumored earlier in the week. ;)

Well you may very well be right, as mine was just a theory with no real solid evidence behind it. Your theory might well be true either, I really have no clue :) Either way, I have no reason to believe that either of these cards will be a disappointment.

Would ya mind offering your dustbuster2 rumor again for those that haven't had time to decipher good rumors from constant fanboy flamings/off topic nonsense? :)

[kemosabe]

I still can't see a balanced product line if the high-end part has 16 pipes and then it drops to 4 (RV380), regardless of core/memory clocks. Seems like a huge hole to fill there.

Which could be nicely filled in the interim by remaining R350 and R360 stock, no?

4 pipelines at <$179 and lower. 8 pipelines (R360) at $180-$250, and 12/16 pipelines (Xt/Pro at $300+)

The question is, how much stock of R350/R360 boards is still out there, and did this factor into any decision on how to "configure" the R420 for initial sale?

I can easily envision the R420 "SE" (perhaps R420 with TWO quads disabled, and on 128 bit bus) replacing the R350/R360...but only after the OEMS have depleted the R350/R360 stock to some pre-determined level.

Agreed. Squeezing the last drop of nectar from R350/R360 is obviously in the interests of both ATI and its AIB partners. If such were the plan, then RV380 will be the only PCI-E mid-range product available for the OEM refresh, since Orton left no doubts about a "top-to-bottom" PCI-E lineup for the spring. An eventual R420 SE part to replace R350/360 would then be more of a retail product, I presume.

[Joe DeFuria]

I really can't see this happening, as ATI tried it once with the Radeon 9500 and 9500 Pro... While they WERE excellent cards, we all saw what ATi did with them... ditched them for a card specifically designed for the mid to low end market.

Right. I didn't say it would be a permanent solution, but it could be an interim solution until they get a chip specifically designed for the market in place.

ATI did whatever "economic calculation" (considering market share, margins, etc.), and decided that the 9500 was the way to go back then.

It may not be ideal, but that doesn't mean it might not still be the best approach to take this time around.

I honestly can't see them returning to that old method once again, I believe the reason they made the 9600 series in the first place was to make a larger profit than selling disabled R300 cores, correct?

Sort of correct. I don't know if you intended it, but you make it sound like the 9600 was a "recation" to the margins on the 9500. It was a planned part from the beginning, knowing that the 9500 part would have low margins.

In other words, while I'm sure ATI would not plan on a 2 quad disabled part to be some long-term solution to fill the gap between a 4 pipe low-end and a 16 pipe high end, I would not rule out the possibility of such a part materializing as a stop-gap measure.

Again, it all depends (IMO), on how much stock of R350/R360 is still out there, how fast that sells-through once the R420 is released, and how soon a "true 8 pipe" (no disabled quads) part that's meant for the sub $200 market will be ready.

I'd love to see basically a R360 on 0.11u to be that part...and it may be in the works, but I've not seen anything like that on any published road-map to date. (Though I'd be expecting something like that for perhaps this fall.)

[MistaPi]

It will not be confusing whatsoever... Remember these cards are not intended to be sold to the "average joe" in the first place (although that does not say the average joe doesn't purchase them). By your logic, the GeForce FX 5800 Ultra should have been confused as a better card than the GeForce FX 5900 Ultra because its clockspeeds were higher... And I don't believe ANYONE confused the two cards.

FX5800 and FX5900 have difference model numbers, I belive this isnt the case here.

[Psikotiko]

Is this a confirmation?


ATI Pipelines:
Rumors are flying, but here is a bit of fact for you. ATI's next-gen part, codenamed "R420," will in fact be a full 16 pipeline VPU. You can take that to the bank, put it in your pipe and smoke it, or do whatever you want to do with it.


http://www.hardocp.com/

[SsP45]

It would be rather interesting if Nvidia was only talking FUD about the NV40, and it is an 8x2/16x0 design. And ATI will have brought out their big guns against a much weaker opponent than they thought.

[Joe DeFuria]

FX5800 and FX5900 have difference model numbers, I belive this isnt the case here.

That implying that the 5900 is "better" than the 5800.

ATI always markets the "XT" as "better than" a pro or non pro. As long as performance of the XT is faster than the same model pro and non pro (who cares how many piplines or clock rates), how could anyone possibly be deceived or mislead?

[PaulS]

Can't wait until next week when the Dual NV40 solution is revealed :twisted:

[surfhurleydude]

Right. I didn't say it would be a permanent solution, but it could be an interim solution until they get a chip specifically designed for the market in place.

ATI did whatever "economic calculation" (considering market share, margins, etc.), and decided that the 9500 was the way to go back then.

It may not be ideal, but that doesn't mean it might not still be the best approach to take this time around.

I gotcha... I'd say whether or not they release a halved version of the R420 may depend on how well the value chip is coming along?



I'd love to see basically a R360 on 0.11u to be that part...and it may be in the works, but I've not seen anything like that on any published road-map to date. (Though I'd be expecting something like that for perhaps this fall.)

Hrm... I can't decide whether I would or would not yet. I'm getting pretty excited over the speculation for new AA and AF methods that the R420 is offering, and a .11u R360 would seem a lot less appealing than a R420 based derivative offering virtually the same speed as a .11u R360, but with the featureset of the R420... or was I reading it wrong and that's what you were trying to say in the first place? :wink:

[Bjorn]

Is this a confirmation?


ATI Pipelines:
Rumors are flying, but here is a bit of fact for you. ATI's next-gen part, codenamed "R420," will in fact be a full 16 pipeline VPU. You can take that to the bank, put it in your pipe and smoke it, or do whatever you want to do with it.


http://www.hardocp.com/

What happened to the 8 Extreme pipes then ?

Anyway, great news. The more pipes, the merrier :-)

[digitalwanderer]

Is this a confirmation?


ATI Pipelines:
Rumors are flying, but here is a bit of fact for you. ATI's next-gen part, codenamed "R420," will in fact be a full 16 pipeline VPU. You can take that to the bank, put it in your pipe and smoke it, or do whatever you want to do with it.


http://www.hardocp.com/
No, that just means he heard one of us say it! ;)

[Psikotiko]

Is this a confirmation?


ATI Pipelines:
Rumors are flying, but here is a bit of fact for you. ATI's next-gen part, codenamed "R420," will in fact be a full 16 pipeline VPU. You can take that to the bank, put it in your pipe and smoke it, or do whatever you want to do with it.


http://www.hardocp.com/
No, that just means he heard one of us say it! ;)

http://www.anandtech.com/video/showdoc.html?i=1966

That's why i asked....... :D

[PaulS]

Anyone seen this pic before? Fake?

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

From Hard OCP (http://www.hardforum.com/showthread.php?s=8568674d4bc527f5fb5dc7a6febab4c7&threadid=752237)

[MuFu]

That's a genuine marketing pic, AFAIK.

Hehe, sounds good. Can't ever remember this being confirmed though...

A 16 pipe chip was always on the cards once they had enough chips capable of it...

Thought that was still a point of uncertainty until just recently. :?:

Well, it's certainly been pretty clear since the 12 out of 16 pipes rumour came about, which was a little while ago now. The assumption, however, was that the last quad would only be activated in a refresh.

I meant the discussion as to whether it was a 12 pipe design with "spare" functionality to account for yield shortcomings (and hence could only ever work with <4 quad-pipes enabled, pending a redesign) or potentially a full 16-pipe architecture in the case of good dice. The issue's moot now, of course.

Re. dustbuster 2, there are samples floating about with VERY noisy cooling solutions. There were a few HS/fan designs in the works though (as always) so I have no idea whether it will ship in tinnitus-inducing form or not. Still waiting for the 1kg "copper sandwich" to appear. :lol:

[UPO]

What if 12pipe configuration of R420 is competitive with NV40 but Ati realised they have to minimise the fact they don't have PS3.0 support?
"hey, we haven't got shaders 3.0 - so what? - we are faster!"

[digitalwanderer]

Anyone seen this pic before? Fake?

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

From Hard OCP (http://www.hardforum.com/showthread.php?s=8568674d4bc527f5fb5dc7a6febab4c7& threadid=752237)
I believe that is the picture that was going around to IHV's last week to show what the "original" cooler on it was going to look like...I think that's the picture my friend was telling me about that made him talk so highly about the nV40's cooling, 'til the actual reference boards went out this morning with loud dustbusters on 'em! :twisted:

[Doomtrooper]

There is those two molex connectors again.... :lol:

[Joe DeFuria]

I gotcha... I'd say whether or not they release a halved version of the R420 may depend on how well the value chip is coming along?

Yes, at least in part. Though I'd say that ATI never planned on having a 8 pipeline "value" part to be available at the same time as R420 launch. So it would be more accurate to charaterisze my viewpoint as "Whether or not they release a halved version of the R420 depends primarily on the stock situation of R350/R360, and if that stock would last until the new 8 pipe part is ready."

Hrm... I can't decide whether I would or would not yet. I'm getting pretty excited over the speculation for new AA and AF methods that the R420 is offering, and a .11u R360 would seem a lot less appealing than a R420 based derivative offering virtually the same speed as a .11u R360, but with the featureset of the R420... or was I reading it wrong and that's what you were trying to say in the first place? :wink:

Well, sure, we'd all prefer (as consumers) the latest tech at every price point. ;) At sub $200 price points though, I personally wouldn't care too much to quibble over AA and Aniso modes above and beyond that which R3xx already supports, which is damn good.

The big question in my mind for this market, is if ATI and / or nVidia will introduce a 256 bit memory bus part at sub $200 MSRP.

[Joe DeFuria]

Anyone seen this pic before? Fake?

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

From Hard OCP (http://www.hardforum.com/showthread.php?s=8568674d4bc527f5fb5dc7a6febab4c7& threadid=752237)

At least it's got Dual DVI. :) I hope both ATI and nVidia start to make this standard on at last their high end offerings.

[surfhurleydude]

Anyone seen this pic before? Fake?

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

From Hard OCP (http://www.hardforum.com/showthread.php?s=8568674d4bc527f5fb5dc7a6febab4c7&threadid=752237)
I believe that is the picture that was going around to IHV's last week to show what the "original" cooler on it was going to look like...I think that's the picture my friend was telling me about that made him talk so highly about the nV40's cooling, 'til the actual reference boards went out this morning with loud dustbusters on 'em! :twisted:

Considering how much of a shame it would be to have the NV40 ship with another dustbuster, I'm going to just be optimistic and say that nVidia might just be having problems with cooling units, while the actual cards themselves are coming along just fine, and in order to appease review sites (I guess they're the ones getting the boards?), the cards were shipped on time, just with an interim cooling solution.

[digitalwanderer]

Considering how much of a shame it would be to have the NV40 ship with another dustbuster, I'm going to just be optimistic and say that nVidia might just be having problems with cooling units, while the actual cards themselves are coming along just fine, and in order to appease review sites (I guess they're the ones getting the boards?), the cards were shipped on time, just with an interim cooling solution.

That could be, and I hope you're right just for the poor hardware's sake!

Don't you think they would have mentioned that to the people they're giving the dustbuster2s to though? I sure would say something like, "BTW, that isn't the final cooling solution...it's just a temp so you can see the card" to make sure no icky rumors got started.

[John Reynolds]

Considering how much of a shame it would be to have the NV40 ship with another dustbuster, I'm going to just be optimistic and say that nVidia might just be having problems with cooling units, while the actual cards themselves are coming along just fine, and in order to appease review sites (I guess they're the ones getting the boards?), the cards were shipped on time, just with an interim cooling solution.

With retail availability supposedly a mere 2.5 weeks away, that's being awfully optimistic IMO.

[Joe DeFuria]

Re. dustbuster 2, there are samples floating about with VERY noisy cooling solutions. There were a few HS/fan designs in the works though (as always) so I have no idea whether it will ship in tinnitus-inducing form or not. Still waiting for the 1kg "copper sandwich" to appear. :lol:

Sounds like we're getting set-up for NV30 reduex:

Every single preview on the planet will have a similar statemet: "nVidia tells us these these are early engineering samples...they promise the shipping products will have tweaked solutions that will be quieter. Honest."

;)

[surfhurleydude]

Considering how much of a shame it would be to have the NV40 ship with another dustbuster, I'm going to just be optimistic and say that nVidia might just be having problems with cooling units, while the actual cards themselves are coming along just fine, and in order to appease review sites (I guess they're the ones getting the boards?), the cards were shipped on time, just with an interim cooling solution.

That could be, and I hope you're right just for the poor hardware's sake!

Don't you think they would have mentioned that to the people they're giving the dustbuster2s to though? I sure would say something like, "BTW, that isn't the final cooling solution...it's just a temp so you can see the card" to make sure no icky rumors got started.

Well, let's just both hope I'm right and not even think about the second paragraph you wrote. In fact, I'm going to forget I even read it :)

[thatdude90210]

Or, everyone in the cooling design department are already hearing impaired from the last time.

[Mark]

something i find interesting is.

According to previously released information, the ATI code-named R420/R423 graphics processors will contain about 160 million of transistors and will be produced using low-k 0.13 micron technology at TSMC.

when nvidia said they had a 16 pipeline card and the number of transistors was only 175 million people said "NO WAY!" and even when it came out that the nv40 had 210 million people still said that wasn't enough. now the r420 only has 160 million and has 16 pipelines?
Remember that the NV40 has full support for PS3.0 while the R420 does not.

[shanehudson]

Anyone seen this pic before? Fake?
http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

I like that black gradient they got going all around the sink cover. Or... maybe that's where the chip charred the board? :P

[digitalwanderer]

AND I JUST KEEP GETTING HAPPY PMS ABOUT THE R420!!! :D :D :D

Ok, I'm getting too damned giddy and gotta step away from the boards for a bit before I get silly.

Just one thing; the match up between the X800 XT and the 6800 Ultra is going to be a good match up and I don't think ANYONE knows who is going to walk away the winner from the match up yet. 8)

[surfhurleydude]

AND I JUST KEEP GETTING HAPPY PMS ABOUT THE R420!!! :D :D :D

Ok, I'm getting too damned giddy and gotta step away from the boards for a bit before I get silly.

Just one thing; the match up between the X800 XT and the 6800 Ultra is going to be a good match up and I don't think ANYONE knows who is going to walk away the winner from the match up yet. 8)

I know who's going to walk away as the winner - the consumer :)

[MistaPi]

FX5800 and FX5900 have difference model numbers, I belive this isnt the case here.

That implying that the 5900 is "better" than the 5800.

ATI always markets the "XT" as "better than" a pro or non pro. As long as performance of the XT is faster than the same model pro and non pro (who cares how many piplines or clock rates), how could anyone possibly be deceived or mislead?

I believe most people is more familar with clockrates than the number of pipelines. I think this can be somewhat confusing dispite the higher price of the XT version. And it doesnt help that nVidia use the XT name on their slower parts. Im not saying its gonna be a huge problem, and the XT version might very well be clocked lower. All im saying is I doudt that will be the case.

[Dave Baumann]

What happened to the 8 Extreme pipes then ?

That specification was given to partners. Partners are known to be leaky.

[MuFu]

Hmm... I am £1101 overdrawn, leaving £399 available. It's a sign. :cool:

[tEd]

What happened to the 8 Extreme pipes then ?

That specification was given to partners. Partners are known to be leaky.

8)

[mreman4k]

AND I JUST KEEP GETTING HAPPY PMS ABOUT THE R420!!! :D :D :D

Ok, I'm getting too damned giddy and gotta step away from the boards for a bit before I get silly.

Just one thing; the match up between the X800 XT and the 6800 Ultra is going to be a good match up and I don't think ANYONE knows who is going to walk away the winner from the match up yet. 8)

I know who's going to walk away as the winner - the consumer :)

Yea right, both cards will be $400-$500 easy. :(
At least that is out of my price range, maybe ATI/NV will send me a sample. :?

[Hellbinder]

something i find interesting is.

According to previously released information, the ATI code-named R420/R423 graphics processors will contain about 160 million of transistors and will be produced using low-k 0.13 micron technology at TSMC.

when nvidia said they had a 16 pipeline card and the number of transistors was only 175 million people said "NO WAY!" and even when it came out that the nv40 had 210 million people still said that wasn't enough. now the r420 only has 160 million and has 16 pipelines?
Remember that the NV40 has full support for PS3.0 while the R420 does not.
Also From what I understand the Nv40 only has 32bit registers. OR Pure 32bit through adn through.. although it does still gain some advantage when using 16bit.. or Split 32.. or however they are specifically doing it now..

Meaning that their Transistor count for all that massive 32bit stuff ois a lot larger.

They may still take a performance hit due to register limitations although its not as bad now.

[martrox]

Hmmm....wonder where DemoCoder and Chalnoth are? The "defenders of the Faith" seem to have lost their voices......... :twisted:

[surfhurleydude]

AND I JUST KEEP GETTING HAPPY PMS ABOUT THE R420!!! :D :D :D

Ok, I'm getting too damned giddy and gotta step away from the boards for a bit before I get silly.

Just one thing; the match up between the X800 XT and the 6800 Ultra is going to be a good match up and I don't think ANYONE knows who is going to walk away the winner from the match up yet. 8)

I know who's going to walk away as the winner - the consumer :)

Yea right, both cards will be $400-$500 easy. :(
At least that is out of my price range, maybe ATI/NV will send me a sample. :?

Yes, they may be 400-500 easy, but as I mentioned on another board, if this comes as any surprise to you, you are a nutcase and you really arne't the type of person that should be buying one of these cards in the first place...

Anyhow, we definitely are walking away as the winner, because we are most likely getting A LOT more for our 400-500 bucks than we would if there wasn't such a fierce competition between the two companies.

[Arun]

What happened to the 8 Extreme pipes then ?
That specification was given to partners. Partners are known to be leaky.
Woah. So NVIDIA's doing it to developers, and ATI's doing to to partners? Not bad, eh... What a business! :lol:

Uttar

[mreman4k]

AND I JUST KEEP GETTING HAPPY PMS ABOUT THE R420!!! :D :D :D

Ok, I'm getting too damned giddy and gotta step away from the boards for a bit before I get silly.

Just one thing; the match up between the X800 XT and the 6800 Ultra is going to be a good match up and I don't think ANYONE knows who is going to walk away the winner from the match up yet. 8)

I know who's going to walk away as the winner - the consumer :)

Yea right, both cards will be $400-$500 easy. :(
At least that is out of my price range, maybe ATI/NV will send me a sample. :?

Yes, they may be 400-500 easy, but as I mentioned on another board, if this comes as any surprise to you, you are a nutcase and you really arne't the type of person that should be buying one of these cards in the first place...

Anyhow, we definitely are walking away as the winner, because we are most likely getting A LOT more for our 400-500 bucks than we would if there wasn't such a fierce competition between the two companies.

True, I was looking at it from a purely financial standpoint. You are definately getting more for your $400-$500. But why am I not the type of person to buy one of those cards? :?:

[digitalwanderer]

Hmmm....wonder where DemoCoder and Chalnoth are? The "defenders of the Faith" seem to have lost their voices......... :twisted:
They could still be reading the thread trying to catch up, this one has grown FAST! :shock:

[surfhurleydude]

True, I was looking at it from a purely financial standpoint. You are definately getting more for your $400-$500. But why am I not the type of person to buy one of those cards? :?:

Because these cards are intended for the ENTHUSIAST market. An enthusiast by definition is someone who is extremely interested in the GPU market, and is into what they purchase and read/research all about the GPUs and the market they belong to... They also want the best product their money can buy. Thus, it should be no surprise to any enthusiast (the type of customers the R420 and NV40 are INTENDED to be sold to) that these cards are going to be between 400 and 500 bucks.

[Hellbinder]

Some other things to consider...

ATi = 160m Transistors on LOW-k

Nvidia = 210m Transistors not on low-k...

Nvidia is not likely to achieve the high core speeds ATi is without hooking a sump-pump to the Nv40.

Thus as well as architecural differences.. Core speed and memory will be a big deciding factor.

It is Very likely that if the cards are close Ati may be able to take the Crown simply by boosting the core to some extreme level via overdrive etc...

pro in the 400mhz range

XT in the 500mhz range

Overdrive = Unknown...

It may be even higher than this... only the shadow knows.... ;)

[Evildeus]

Hmmm....wonder where DemoCoder and Chalnoth are? The "defenders of the Faith" seem to have lost their voices......... :twisted:Let me think of a response. Well the wield will be so low that there will be as much 16 pipelines R420 as NV30 :lol:

Well, it seems that HB was wrong at least, the R420 will be a 16 pipelines beast, and the competition will be great once more. Ati's winning once more? :?

[mreman4k]

True, I was looking at it from a purely financial standpoint. You are definately getting more for your $400-$500. But why am I not the type of person to buy one of those cards? :?:

Because these cards are intended for the ENTHUSIAST market. An enthusiast by definition is someone who is extremely interested in the GPU market, and is into what they purchase and read/research all about the GPUs and the market they belong to... They also want the best product their money can buy. Thus, it should be no surprise to any enthusiast (the type of customers the R420 and NV40 are INTENDED to be sold to) that these cards are going to be between 400 and 500 bucks.

LOL...Don't want to keep this goin, but its possible to be a 'broke' enthusiast like myself at the present time. Or perhaps I'm waiting for something... :wink:

[Doomtrooper]

In about 4 days I want to revisit the 'rumour thread' and how quickly some select individuals wrote off the R420.....just like a little over a year ago.

[Evildeus]

pro in the 400mhz range

XT in the 500mhz range
That seems odd, because yield will effectively be much lower on a XT with 16 pipes @500 than a pro with 12 pipes @400. But, what do i know? ;)

[surfhurleydude]

True, I was looking at it from a purely financial standpoint. You are definately getting more for your $400-$500. But why am I not the type of person to buy one of those cards? :?:

Because these cards are intended for the ENTHUSIAST market. An enthusiast by definition is someone who is extremely interested in the GPU market, and is into what they purchase and read/research all about the GPUs and the market they belong to... They also want the best product their money can buy. Thus, it should be no surprise to any enthusiast (the type of customers the R420 and NV40 are INTENDED to be sold to) that these cards are going to be between 400 and 500 bucks.

LOL...Don't want to keep this goin, but its possible to be a 'broke' enthusiast like myself at the present time. Or perhaps I'm waiting for something... :wink:

Oh, welcome to the club then. :lol:

[Tim Murray]

Anyone seen this pic before? Fake?

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

From Hard OCP (http://www.hardforum.com/showthread.php?s=8568674d4bc527f5fb5dc7a6febab4c7& threadid=752237)
Pic is real. Cooler might be real, but it isn't yet. (And I'm very annoyed about that.)

[Hellbinder]

Hmmm....wonder where DemoCoder and Chalnoth are? The "defenders of the Faith" seem to have lost their voices......... :twisted:Let me think of a response. Well the wield will be so low that there will be as much 16 pipelines R420 as NV30 :lol:

Well, it seems that HB was wrong at least, the R420 will be a 16 pipelines beast, and the competition will be great once more. Ati's winning once more? :?
I have comented privately to a number of people for a long time now that the R420 *could* be considered 8 12 or 16 pipeline card depending on how ATi decided to market it.

there have been *games afoot* with the public information for Quite some time. ;)

[Joe DeFuria]

pro in the 400mhz range

XT in the 500mhz range
That seems odd, because yield will effectively be much lower on a XT with 16 pipes @500 than a pro with 12 pipes @400. But, what do i know? ;)

We have no idea if the Pro is 12 pipes or not. Assuming 16 pipes (for at least the XT) is true, BOTH the pro and the xt may be 16 pipes. If that's the case, the surely, the Pro core and memory clock would be lower.

[Bjorn]

Some other things to consider...

ATi = 160m Transistors on LOW-k

Nvidia = 210m Transistors not on low-k...


I thought that you couldn't compare them since Ati and Nvidia doesnt count transistors the same way.

[Sazar]

is the nv40 utilising SOI ?

btw... that is one MASSIVE card you lads posted a pic of :shock:

I am hoping that the r420 is a shorter card (shorter and taller I can live with...

[Hellbinder]

pro in the 400mhz range

XT in the 500mhz range
That seems odd, because yield will effectively be much lower on a XT with 16 pipes @500 than a pro with 12 pipes @400. But, what do i know? ;)
Ati stated a long tine ago that they would be willing to sacrifice yield for performance this time around.

There will not be truckloads of X800 XT's floating around initially. Just like the Nv40 ultra (as i am told privately) will be as rare as a four leaf clover and should be considered a "Reviewers card".

The X800Xt will have much greater availability.

[anaqer]

| EDIT : I'm SO fscking slow. :oops: |

[Bjorn]

I have comented privately to a number of people for a long time now that the R420 *could* be considered 8 12 or 16 pipeline card depending on how ATi decided to market it.


"Depending on how they decide to market it" ?

Sounds awfully vague to me.

[Joe DeFuria]

I have comented privately to a number of people for a long time now that the R420 *could* be considered 8 12 or 16 pipeline card depending on how ATi decided to market it.

there have been *games afoot* with the public information for Quite some time. ;)

I just hope that ATI "deciding to market it" as 16 pipes, doesn't mean that 16 pipe operation doesn't come with a big, huge, asterik of "conditions." ;)

[Hellbinder]

Some other things to consider...

ATi = 160m Transistors on LOW-k

Nvidia = 210m Transistors not on low-k...


I thought that you couldn't compare them since Ati and Nvidia doesnt count transistors the same way.
In this case we are talking about heat generation and energy consumption. Not architectural performance.

[Tim Murray]

HB, I think that, whoever the winner is, this generation's battle will be won through efficiency (we're already hitting memory bandwidth limitations). And that's where we can't begin to speculate until we have cards from both.

It wouldn't surprise me if the Pro is 12 pipes, and the XT is 16, they'll be clocked the same. Memory bandwidth will ultimately determine the winner, I'll guess.

Oh, you meant heat? Gawd, that battle was over before it began. NVIDIA is always ahead on heat and energy consumption. They got bit in the ass with .13u and the NV30, so now they won't do anything for heat or power consumption that would make the card far more appetizing to consumers. Doh.

[Evildeus]

I have comented privately to a number of people for a long time now that the R420 *could* be considered 8 12 or 16 pipeline card depending on how ATi decided to market it.

there have been *games afoot* with the public information for Quite some time. ;)Well, i can't comment on private communication, but at least it wasn't public ;). What interest me the more is to see Ati going for 16 pipes, meaning that the NV40 isn't a bad thing. Then if Ati's going to stay ahead, well, great job is all i can say :)

[DemoCoder]

Hmmm....wonder where DemoCoder and Chalnoth are? The "defenders of the Faith" seem to have lost their voices......... :twisted:

What's to defend? I don't get it Martox. I'm giddy as a school girl that there are two uber cards on the horizon and ATI an NVidia are climbing over one another trying to for an upset. It's the ATI boys that were talking trash about the NV40 and hyping the R420. Remember HB using words like "Owned"


I only get involved in threads where there is some obvious nonsense involved. I challenge you to find a single post of mine where I talked down R420 performance or said NV40 was going to own them. Step off.

[Evildeus]

pro in the 400mhz range

XT in the 500mhz range
That seems odd, because yield will effectively be much lower on a XT with 16 pipes @500 than a pro with 12 pipes @400. But, what do i know? ;)

We have no idea if the Pro is 12 pipes or not. Assuming 16 pipes (for at least the XT) is true, BOTH the pro and the xt may be 16 pipes. If that's the case, the surely, the Pro core and memory clock would be lower.Why get the XT one month later then? :?:

[Hellbinder]

I have comented privately to a number of people for a long time now that the R420 *could* be considered 8 12 or 16 pipeline card depending on how ATi decided to market it.

there have been *games afoot* with the public information for Quite some time. ;)Well, i can't comment on private communication, but at least it wasn't public ;). What interest me the more is to see Ati going for 16 pipes, meaning that the NV40 isn't a bad thing. Then if Ati's going to stay ahead, well, great job is all i can say :)
You are right about that. The Nv40 Ultra will be some stiff competition. With each card offering something the buyer should consider.

However... I think that AA and AF (both are improved on the R420) will go to ATi.

[Evildeus]

I have comented privately to a number of people for a long time now that the R420 *could* be considered 8 12 or 16 pipeline card depending on how ATi decided to market it.

there have been *games afoot* with the public information for Quite some time. ;)

I just hope that ATI "deciding to market it" as 16 pipes, doesn't mean that 16 pipe operation doesn't come with a big, huge, asterik of "conditions." ;)Hope so too, remind me of some cards from.... Nv :roll:

[jb]

HB, I think that, whoever the winner is, this generation's battle will be won through efficiency (we're already hitting memory bandwidth limitations). And that's where we can't begin to speculate until we have cards from both.

I think you have a good point...but like the good ol days of the NV30 stuff...have efficent bandwidth saving does help but at the end of the day you only have what you have :)

[Joe DeFuria]

We have no idea if the Pro is 12 pipes or not. Assuming 16 pipes (for at least the XT) is true, BOTH the pro and the xt may be 16 pipes. If that's the case, the surely, the Pro core and memory clock would be lower.Why get the XT one month later then? :?:

:?:

Because the XT would be clocked higher and therefore perform better than the Pro?

[Evildeus]

The X800Xt will have much greater availability.Well, we will find out later. Why Ati's launching this part later? :?:

[demalion]

Are we sure the RV380 is just the RV360 with PCI Express? What is the RV370?

RV370 AGP? Doesn't make sense with top to bottom being proposed and existing AGP design stock.
RV370 just using a 64-bit memory interface? Why bother with a 64-bit only design...yields and cost? How much margin increase would that be over a chip that could be used for both 128-bit and 64-bit?
An ultra dirt cheap absolute bottom of the line chip, perhaps dropping vertex shaders or something (perhaps in conjunction with 64-bit interface), and heading for reincarnation as an IGP?
The last makes a great deal of sense for aggressive profit seeking opportunities, what with PCI express allowing better performance from this type of part, but I don't recall confirmation of the RV370 being this.

So, I wonder, if it isn't more likely that RV370 is the PCI express RV360, and RV380 is a significant enhancement (perhaps 8 pipes, which makes sense to me what with the Cadence tool set discussions, or perhaps a far more drastic clock speed increase than I recall being mentioned)?
Makes sense with the R420 expectations and aggressive mid range performance leadership efforts, and looks like it could be a profitable strategy...the only thing arguing strongly against it (besides specific information I don't happen to recall being confirmed) would be if an 0.11 4 pipe card fairly identical to the RV360 would be a profit margin problem for the low end, and insufficient improvement over the current 9600SE profitability.

[Dave Baumann]

Probably memory.

[martrox]

The 5 stages of grief......
1-Denial
2-Anger
3-Bargaining
4-Depression
5-Acceptance

A few here have spent the last couple of years between #'s 1 and 3 :twisted: :wink:

(j/k)

[Evildeus]

However... I think that AA and AF (both are improved on the R420) will go to ATi.Hope to see some great improvements from Nv also, otherwise, it's a pity.
I don't know, but if the R420 is really a 4 time Rv360, then i don't see much more IQ improvement from Ati, but surely some on effeciency.

[Evildeus]

Probably memory. :shock: Not sure that 1 month will be that much to got some super cool memory parts, but well, you have more information on this than me :)

[Bjorn]

In this case we are talking about heat generation and energy consumption. Not architectural performance.

I know.

[Sxotty]

For my proof of that, I offer the latest "dustbuster2" rumor going around this morning in lieu of the elegant single-slot solution being rumored earlier in the week. ;)

I think dual slot solutions are more "elegant" :) Really I like them better, but I like it best when a single slot would do, but a dual is used instead to be silent, a la the zalmann9800pro, or the vga silencers for 9800's.

LOL at the sump pump from hellbinder... is that for the leaking coolant in the case then?

[Mark]

And people call me a fan-boy because of my email address. I say look at this thread if you want some examples of fanboys. Wow.

Anyway, 16-pipe vs 16-pipe seems like a fair fight to me. I wouldn't go around automatically claiming any pre-mature victories just yet. Personally I expect them to compare roughly the same as the R360 and NV38 compare today. That is, about on par with each other.

[Joe DeFuria]

Are we sure the RV380 is just the RV360 with PCI Express? What is the RV370?

General concensus is that RV380 is in fact "RV360 but with PCI-Express native support". (Although shipping cards core and memory clock rates may be a bit different.)

I don't think anyone is "really sure" what the RV370 is, but it sounds like (read: speculation) to me it would be almost identical to RV380 feature and architecture wise, but on 0.11. Target clocks likely to be designed to be significantly lower than RV380 though, making it much easier to design a high yield part. Combination of high yield with smaller size (0.11) should equate to cheap part.

ATI has promised "top to bottom" native PCI-E support when PCI-E becomes available, so I'm guessing the RV370 is actually native PCI-E...though I'm also guessing the chip will appear on AGP cards with a "reverse bridge."

[Hellbinder]

However... I think that AA and AF (both are improved on the R420) will go to ATi.Hope to see some great improvements from Nv also, otherwise, it's a pity.
I don't know, but if the R420 is really a 4 time Rv360, then i don't see much more IQ improvement from Ati, but surely some on effeciency.
Its not merely a 4x Rv360.. But a Quad 4x1 with 2 vertex shader pipelines each. The vertex and Pixel shader engines are Quite enhanced and it has other additional Changes, enhancements and Features.

One being improved AF.

[martrox]

[quote="Sxotty"
I think dual slot solutions are more "elegant" :) Really I like them better, but I like it best when a single slot would do, but a dual is used instead to be silent, a la the zalmann9800pro, or the vga silencers for 9800's.[/quote]

What I like about some dual setups are the ones that vent out of the computer......like the vga sliencer...... with "very hot" processors, the last thing anyone needs is more heat rising up from the videocard right into the path of the CPU cooler......

[Tim Murray]

if it's not angle-independent, I don't care. :)

[Evildeus]

Its not merely a 4x Rv360.. But a Quad 4x1 with 2 vertex shader pipelines each. The vertex and Pixel shader engines are Quite enhanced and it has other additional Changes, enhancements and Features.

One being improved AF.On AF i can see that happening, PS/VS ok, but AA?

[Bjorn]

However... I think that AA and AF (both are improved on the R420) will go to ATi.

I would agree when it comes to AA. Not so sure about AF quality though, at least when it comes to what the hardware is capable of :-)

What i do expect from Ati is a "no angle problem" mode when it comes to AF.

[digitalwanderer]

I think dual slot solutions are more "elegant" :) Really I like them better, but I like it best when a single slot would do, but a dual is used instead to be silent, a la the zalmann9800pro, or the vga silencers for 9800's.
Personally I'm with you on that. My 9600 & 9700 both only take up a single-slot, but that's only 'cause I haven't modded out the 9600 yet and I ain't done with the 9700 yet. ;)

I'll always be willing to give up a slot for a faster/cooler card, but some others seem to feel differently. :)

Hey HB, is that "aniso improvement" mayhaps in the form of non-angle dependant aniso? ;)

[DemoCoder]

Put up or shutup matrox. Find a post of mine where I talk down R420 performance, disbelieve 16 pipes, or even get involved in the pipeline discussion at all. In fact, you will find that in all of the rumor mill threads, I have not even once commented on the 12 or 16 pipeline issue with the 420 or future R420 PS2.0 performance. I primarily get in threads which have to do with software or hardware features and architecture, such as compilers and drivers. And although YOU think that criticizing people's attack posts on say, Cg, constitute a uncritical defense of Nvidia, they do not.

So don't lump me in with your war against Chalnoth. I expect an apology.

[Joe DeFuria]

I think dual slot solutions are more "elegant" :) Really I like them better, but I like it best when a single slot would do, but a dual is used instead to be silent, a la the zalmann9800pro, or the vga silencers for 9800's.

Well, I wouldn't call dual slot more "elegant", ;) but I do agree completely that I'm happy to make the trade off of a slot / elegance for more silent operation.

I use a VGA silencer on my 9800.

I prefer the silencer to the zalman, because even though the zalman could be "completely silent" in and of itself...it's still dumping all the heat directly into the case, which will require higher speed case and / or PSU fans to circulate it out.

[kemosabe]

We have no idea if the Pro is 12 pipes or not. Assuming 16 pipes (for at least the XT) is true, BOTH the pro and the xt may be 16 pipes. If that's the case, the surely, the Pro core and memory clock would be lower.Why get the XT one month later then? :?:

:?:

Because the XT would be clocked higher and therefore perform better than the Pro?

But that would in turn depend on whether the Pro itself outperformed the highest configuration of NV40 available at that time, no? Otherwise, the XT would be competing with its sibling and force its price down prematurely. In that case, the XT could be saved for later, unless they want to introduce a new price point at $599. :lol:

[digitalwanderer]

Hmmm....wonder where DemoCoder and Chalnoth are? The "defenders of the Faith" seem to have lost their voices......... :twisted:

What's to defend? I don't get it Martox. I'm giddy as a school girl that there are two uber cards on the horizon and ATI an NVidia are climbing over one another trying to for an upset.

Yup, I'm feeling the same way.

(I'm still rooting for ATi though, but that's 'cause I got a lot more friends who work there. ;) )

[Joe DeFuria]

So don't lump me in with your war against Chalnoth. I expect an apology.

I hope you have more luck getting one than I am with John. ;)

(And I agree, I don't see why you should or shouldn't be saying anything in this thread...)

[Tim Murray]

I don't care. I want good AA AND good AF AND good performance AND a quiet card. If it's ATI, cool. If it's NVIDIA (how? dunno, maybe someone will be awesome and make a quiet cooler), cool. I don't care.

[Bouncing Zabaglione Bros.]

Woah. So NVIDIA's doing it to developers, and ATI's doing to to partners? Not bad, eh... What a business! :lol:

Uttar

Well if the rumours are to be believed, Nvidia told developers the NV40 is 16x1 when we think it's really 8x2, ATI told partners R420 was 8x1 when it turns out to be 16x1.

I suspect you can get away with lying to your partners/customers if you then produce something better than you promised, rather than worse than you promised. Especially if you keep insisting something is 16x1 when it can only be used that way under certain specific conditions.

[Joe DeFuria]

I don't care. I want good AA AND good AF AND good performance AND a quiet card. If it's ATI, cool. If it's NVIDIA (how? dunno, maybe someone will be awesome and make a quiet cooler), cool. I don't care.

You forgot, "And I want it for FREE, damnit!" :)

[Evildeus]

:?:
Because the XT would be clocked higher and therefore perform better than the Pro? It still doesn't make sense :/ Of course the XT is clocked higher and perform better, that's why the XT is there. Perhaps they should release the SE one month before because the pro part performe better and is clocked higher? :twisted:

[digitalwanderer]

I don't care. I want good AA AND good AF AND good performance AND a quiet card. If it's ATI, cool. If it's NVIDIA (how? dunno, maybe someone will be awesome and make a quiet cooler), cool. I don't care.

You forgot, "And I want it for FREE, damnit!" :)
And "NOW!!!!", can't forget that. 8)

[PaulS]

Hmmm....wonder where DemoCoder and Chalnoth are? The "defenders of the Faith" seem to have lost their voices......... :twisted:

What's to defend? I don't get it Martox. I'm giddy as a school girl that there are two uber cards on the horizon and ATI an NVidia are climbing over one another trying to for an upset.

Yup, I'm feeling the same way.

(I'm still rooting for ATi though, but that's 'cause I got a lot more friends who work there. ;) )

I'm more of an NVIDIA fan, so I guess I'm rooting for them to win. Best case scenario would be a dead heat between them, however 8) I suspect ATi will probably get it overall, though, and that's been the same since the 12 pipelines. Their whole architecture is just geared so perfectly for DX9, it's hard to imagine NV stealing that from them, regardless of the insane performance from NV40 at times.


Woah. So NVIDIA's doing it to developers, and ATI's doing to to partners? Not bad, eh... What a business! :lol:

Uttar

Well if the rumours are to be believed, Nvidia told developers the NV40 is 16x1 when we think it's really 8x2, ATI told partners R420 was 8x1 when it turns out to be 16x1.

No, developers were told NV40 was 8x2.

[Joe DeFuria]

:?:
Because the XT would be clocked higher and therefore perform better than the Pro? It still doesn't make sense :/ Of course the XT is clocked higher and perform better, that's why the XT is there. Perhaps they should release the SE one month before because the pro part performe better and is clocked higher? :twisted:

I guess I honestly just don't understand your question then. (Why should people wait another month for a better performing card?) Some will, some won't...same as always?

[kemosabe]

:?:
Because the XT would be clocked higher and therefore perform better than the Pro? It still doesn't make sense :/ Of course the XT is clocked higher and perform better, that's why the XT is there. Perhaps they should release the SE one month before because the pro part performe better and is clocked higher? :twisted:

I suppose it will all depend on the general consensus in the hardware review community. If R420 Pro has indeed just been upgraded from 12 to 16 pipelines and competes effectively with whatever NV40 turns out to be, then the XT is not an imperative launch in the near term. Perhaps NV40 has its own Ultra version to be launched in case this very scenario materialized, and will then force the XT launch.

C'est un putain de jeu d'échec, ça c'est sûr. ;-)

[Lecram25]

Anyone seen this pic before? Fake?

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

From Hard OCP (http://www.hardforum.com/showthread.php?s=8568674d4bc527f5fb5dc7a6febab4c7&threadid=752237)

Yes, it is a real pic. Though I've been told a few things about the card:

- the fan covers 3/4 of the card (the pic confirms this)
- there's supposedly a heat pipe on the card too (i couldn't make out a heatpipe though)

[Evildeus]

I guess I honestly just don't understand your question then. (Why should people wait another month for a better performing card?) Some will, some won't...same as always?I can understand that, of course, but since now the higher part was released before and not the contrary. But well, i understand what you mean.

[Sxotty]

BTW I agree with you all that dumping the heat out is better.

And remember the dustbuster 1 did do that. It was just horrbily loud and annoying. If they would have used a quiet fan and stuff (had it clocked at say 400Mhz) it would have been a great cooling solution in my mind. Yet the card would have still been terrible :).

[Bjorn]

Well, I wouldn't call dual slot more "elegant", ;) but I do agree completely that I'm happy to make the trade off of a slot / elegance for more silent operation.


I agree as well. Although i will probably get a water cooling system for my next PC and then i guess that a single slot solution will be better since i woould'nt have to use a potential dustbuster anyway.

[Evildeus]

:?:
Because the XT would be clocked higher and therefore perform better than the Pro? It still doesn't make sense :/ Of course the XT is clocked higher and perform better, that's why the XT is there. Perhaps they should release the SE one month before because the pro part performe better and is clocked higher? :twisted:

I suppose it will all depend on the general consensus in the hardware review community. If R420 Pro has indeed just been upgraded from 12 to 16 pipelines and competes effectively with whatever NV40 turns out to be, then the XT is not an imperative launch in the near term. Perhaps NV40 has its own Ultra version to be launched in case this very scenario materialized, and will then force the XT launch.
Yes, of course that would make sense. But Ati (and Nv surely) must have a good view of what each part will be able to deliver.

C'est un putain de jeu d'échec, ça c'est sûr. ;-)Oui, un peu de theorie des jeux ;)

[Evildeus]

Anyone seen this pic before? Fake?

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

From Hard OCP (http://www.hardforum.com/showthread.php?s=8568674d4bc527f5fb5dc7a6febab4c7&threadid=752237)

Yes, it is a real pic. Though I've been told a few things about the card:

- the fan covers 3/4 of the card (the pic confirms this)
- there's supposedly a heat pipe on the card too (i couldn't make out a heatpipe though)Yeah seems to be a real thing 8)

[digitalwanderer]

It was just horrbily loud and annoying. If they would have used a quiet fan and stuff (had it clocked at say 400Mhz) it would have been a great cooling solution in my mind.

Yup. I never mind fan noise, but I HATE annoying fan noise. (There is a difference, if I have to explain it you just wouldn't understand. ;) )

[Hellbinder]

However... I think that AA and AF (both are improved on the R420) will go to ATi.

I would agree when it comes to AA. Not so sure about AF quality though, at least when it comes to what the hardware is capable of :-)

What i do expect from Ati is a "no angle problem" mode when it comes to AF.
Oh ye of little faith ;)

I said improved, not totally different :)

like the removal of thiose pesky mip lines etc..etc due to some new enhancements :)

[kemosabe]

Yes, of course that would make sense. But Ati (and Nv surely) must have a good view of what each part will be able to deliver.

I don't know about you, but recent events would suggest that that's not at all the case. :lol:

[Dave Baumann]

if it's not angle-independent, I don't care. :)

You might be in for a bit of a surprise (and I'm not talking about ATI).

[Humus]

8 pages in 45 minutes? FFS! Stop typing! :D

[Joe DeFuria]

if it's not angle-independent, I don't care. :)

You might be in for a bit of a surprise (and I'm not talking about ATI).

:shock:

This "NV40 vs. R420" debate should be quite a spectacle! :)

[Joe DeFuria]

8 pages in 45 minutes? FFS! Stop typing! :D

The quick brown fox jumped over the lazy dog... :)

[Sxotty]

It was just horrbily loud and annoying. If they would have used a quiet fan and stuff (had it clocked at say 400Mhz) it would have been a great cooling solution in my mind.

Yup. I never mind fan noise, but I HATE annoying fan noise. (There is a difference, if I have to explain it you just wouldn't understand. ;) )

I assume you mean that high pitched.. hellacious noise, I had a delta fan @ 7500rpm (maybe more I forget) on my cpu and I almost crushed the damn thing. I thought I got a deal on the heatsink cause people said good things about how much air it could move and stuff the dB wasn't that high either but it drove me mad.

[McDusty]

if it's not angle-independent, I don't care. :)

You might be in for a bit of a surprise (and I'm not talking about ATI).

It was just a matter of time before Nvidia copied ATI on this. They can't really afford to continue taking a beating on AF benchmarks anymore.

[Evildeus]

Yes, of course that would make sense. But Ati (and Nv surely) must have a good view of what each part will be able to deliver.

I don't know about you, but recent events would suggest that that's not at all the case. :lol:Hmmm, i'm sure they are. If Ati/Nv wants the card to quicly comes to retail channels they need to give all information needed by the marketing departments of card makers quite early :)

[gurgi]

Hehe, it would be pretty funny if nVidia went to an ATI implementation of AF after all the stink over it.

[Bouncing Zabaglione Bros.]

And people call me a fan-boy because of my email address. I say look at this thread if you want some examples of fanboys. Wow.

Anyway, 16-pipe vs 16-pipe seems like a fair fight to me. I wouldn't go around automatically claiming any pre-mature victories just yet. Personally I expect them to compare roughly the same as the R360 and NV38 compare today. That is, about on par with each other.

If they are matched on performance, it will come down to IQ (for those of us that care), heat and noise generation.

I was going to buy an NV3x (before we found out how poor they were), and the single biggest thing that drove me to ATI was the noise of the NV3x dustbuster. No way was I going to pay that much money and live with that much noise. If NV4x has a dustbuster 2, that will be a big, big strike against it in my opinion.

[Evildeus]

if it's not angle-independent, I don't care. :)

You might be in for a bit of a surprise (and I'm not talking about ATI).NV40 ;) seems to be the rumor (and more it seems) ATM :)

[Bjorn]

if it's not angle-independent, I don't care. :)

You might be in for a bit of a surprise (and I'm not talking about ATI).

As long as you can choose modes that don't have it then it's great. Just like brilinear.

[digitalwanderer]

8 pages in 45 minutes? FFS! Stop typing! :D

The quick brown fox jumped over the lazy dog... :)
Hah! And I'm even typing with a splint on the middle finger of me right hand! 8)

[Joe DeFuria]

As long as you can choose modes that don't have it then it's great. Just like brilinear.

Yup...having the option to trade image qaulity for performance is great...though that's unlike brilinear...at least on the NV3x with current drivers.

[PaulS]

I've got to assume Angle Dependant AF is a mere option on NV40, else what I've heard about AF IQ wouldn't make sense.

[Bjorn]

As long as you can choose modes that don't have it then it's great. Just like brilinear.

Yup...having the option to trade image qaulity for performance is great...though that's unlike brilinear...at least on the NV3x with current drivers.

I guess i should have said "just like brilinear if it could be turned off". But i thought that was obvious :-)

[Evildeus]

I've got to assume Angle Dependant AF is a mere option on NV40, else what I've heard about AF IQ wouldn't make sense.Give us all :P

[Moose]

I want performance and IQ with IQ being more important to me since I suspect the performance will be nearly the same or at least high enough on either card not to matter. Noise and heat don't bother me too much with all the fans I have running in my case but I am concerned about power requirements. I'm not in the mood to have to buy another power supply so soon.

[Tahir2]

if it's not angle-independent, I don't care. :)

You might be in for a bit of a surprise (and I'm not talking about ATI).

Tell me Dave did you have to upgrade your Power Supply Unit when you put that NV40 in your system by any chance? ;)

[Dave Baumann]

Tell me Dave did you have to upgrade your Power Supply Unit when you put that NV40 in your system by any chance? ;)

Fortunatly I'd already upgraded during an abortive attempt to run a P4EE review.

[digitalwanderer]

if it's not angle-independent, I don't care. :)

You might be in for a bit of a surprise (and I'm not talking about ATI).
STOP IT!!! :lol:

Seriously, that would be just too freaking funny! :lol:

[Tahir2]

Tell me Dave did you have to upgrade your Power Supply Unit when you put that NV40 in your system by any chance? ;)

Fortunatly I'd already upgraded during an abortive attempt to run a P4EE review.

If you have any more problems like that give me a call.. I owe you one as it is :)

[Seiko]

Tell me Dave did you have to upgrade your Power Supply Unit when you put that NV40 in your system by any chance? ;)

Fortunatly I'd already upgraded during an abortive attempt to run a P4EE review.

Well I'll take that as you actually have a sample board then ;)

[Tahir2]

Tell me Dave did you have to upgrade your Power Supply Unit when you put that NV40 in your system by any chance? ;)

Fortunatly I'd already upgraded during an abortive attempt to run a P4EE review.

Well I'll take that as you actually have a sample board then ;)

I dont think he went to San Hose just to see the sites :twisted:

[Doomtrooper]

Hehe, it would be pretty funny if nVidia went to an ATI implementation of AF after all the stink over it.

I can't wait :D

[nelg]

Maybe the change from 12 to 16 pipelines is due to yields not competitive pressure.The chat around here suggest that ATI has been very successful in achieving high yields.

[Nick Spolec]

:shock:

I guess this "R420" vs. "NV40" is pretty interesting, huh?

[Doomtrooper]

Maybe the change from 12 to 16 pipelines is due to yields not competitive pressure.The chat around here suggest that ATI has been very successful in achieving high yields.

That simply can't be possible because Chalnoth says Nvidias engineers are better http://www.artnskins.com/e107_images/emoticons/glasses2.png

[Seiko]

Tell me Dave did you have to upgrade your Power Supply Unit when you put that NV40 in your system by any chance? ;)

Fortunatly I'd already upgraded during an abortive attempt to run a P4EE review.

Well I'll take that as you actually have a sample board then ;)

I dont think he went to San Hose just to see the sites :twisted:

Well I wouldn't know what Dave got up to in San Hose :)
It's just great to know that B3D is finally getting a chance to review boards on the initial rounds.

It's been a long time coming but I think Dave and the crew have finally established themselves as one of the leading sources for informative and technically accurate reviews. My only complaint of old was you usually had to wait a good number of weeks before you'd see the reviews from B3D that the likes of Toms and Anands had already blurted out in a generally poor manor.

Of course getting confirmation from Dave when he'll be in a position to publish his review will be like getting blood out the stone with no doubt a huge number of teases!

:)

edit = typo

[Tahir2]

I would rather Dave would do a thorough job than a 'me too' job by trying to be the first out of the blocks.. I am pretty sure most people would agree. Now I am left wondering what the status is with the R420.. Dave? ;)

[McDusty]

What I want to know is .... why is Dave posting here right now!

If I'd just received an N40 sample, I don't think i'd be loitering in here

[Seiko]

I would rather Dave would do a thorough job than a 'me too' job by trying to be the first out of the blocks.. I am pretty sure most people would agree. Now I am left wondering what the status is with the R420.. Dave? ;)

You bet although I think the delays of old where simply due to lack of samples as oppsoed to a longer scrutiny period!

And if you get that "I am left wondering what the status is with the R420.. Dave?" out of Dave I'll eat my hat!

[Tahir2]

What I want to know is .... why is Dave posting here right now!

If I?d just received an N40 sample, I don't think id be loitering in here

He has cloned himself as "V2" and can be in two places at once. I thought that was common knowledge..?

(Apologies for going OT)

[PaulS]

I was under the impression no one had any R420 samples (not the same as not having any samples full stop), although I may be wrong.

[MasterBaiter]

Such a good time to be a consumer.. :P

This is the first time that I've actually thought about switching back to Nvidia since my 4600. I've been limping (hehe) along with a 9700Pro for quite a while now. I think the big problem might be if I should upgrade now or wait for a PCI-X board. Granted we haven't seen anything really that tangible about these cards yet, but it really does seem to be a win-win time right now. Now if Nvidia can only resist the temptation to get more creative with their drivers.... :oops:

[Joe DeFuria]

Presumably ATI, is getting their samples and press documents out on the 13th. (With NDAs being lifted on the 26th). Don't know that for 100% fact, but that's what seems to be the case from various reports / hints.

[Mark]

presumptions are often wrong

[Joe DeFuria]

presumptions are often wrong

Yes, they certainly are!

[kemosabe]

Of course getting confirmation from Dave when he'll be in a position to publish his review will be like getting blood out the stone with no doubt a huge number of teases!

:)

Indeed......teaser extraordinaire. But you can just sense that as the pieces of the puzzle start leaking through at an accelerating rate from various sources, he's been fighting a mounting urge to explode at the seams. :wink:

But I'm curious, Dave, did B3D not get a call from ATI giving you the green light to spill this 16-pipe story yourself? Or did you just figure it was going to be discussed to death anyway? :lol:

[Seiko]

Maybe the change from 12 to 16 pipelines is due to yields not competitive pressure.The chat around here suggest that ATI has been very successful in achieving high yields.

Well this is a much more aggressive ATI of late and although I know Joe won't be too impressed as a share holder I honestly think it's time for ATi to lower margins a little and put out another stunning card for us consumers and gamers. The whole 8500/9200 relaunch really wound me up at the time and from rumours of late even ATI realised they may have missed a trick by allowing Nvidia in on the DX9 budget section. Sure they got rid of old stock but lost a great deal of the lower end market share.

I'm hoping for a good dominant position this time around too to really ensure an even and balanced mindset for future game developments. I'm not sure if it's my imagination but I still feel many more developers use Nvidia as a base line for features no doubt due to the exceptional GF3 series and consider the R300 as a one off?

Any developers care to offer an opinion?

[Doomtrooper]

http://www.techreport.com/etc/2003q3/valve/image13.jpg

[Joe DeFuria]

Well this is a much more aggressive ATI of late and although I know Joe won't be too impressed as a share holder I honestly think it's time for ATi to lower margins a little and put out another stunning card for us consumers and gamers.

On the contrary...there is good PR value to be had in being the perceived market leader, and if lowering margins helps that, it's not necessarily a bad thing, and can ultimately be a good thing.

There's nothing wrong with giving margins now and then, As long as there is a calculated return on that investment. The bottom line is long term growth and profitabilty.

Something is "wrong", if you end up having to sacrifice margins continuously, just to be competitive. That's when I'd get concerned.

[Raystream]

What ever happen to the whole hype concernining how the "420" in the project name r420 stands for in April?

Raystream

[Quitch]

I'm just surprised that when the rumours of the nv40 specs came out, people had already declared it the winner. Talk about learning nothing from the past...

This should certainly be an interesting round, though I see the usual assumptions of nVidia "superior" AF, which I think, as Brent from [H]ard|Ocp has shown in some of his reviews, is often more myth than fact. Anyway, both companies use adaptive AF to some extent already, it's just that ATI are better known for it.

[kemosabe]

Well this is a much more aggressive ATI of late and although I know Joe won't be too impressed as a share holder I honestly think it's time for ATi to lower margins a little and put out another stunning card for us consumers and gamers.

On the contrary...there is good PR value to be had in being the perceived market leader, and if lowering margins helps that, it's not necessarily a bad thing, and can ultimately be a good thing.

There's nothing wrong with giving margins now and then, As long as there is a calculated return on that investment. The bottom line is long term growth and profitabilty.

Something is "wrong", if you end up having to sacrifice margins continuously, just to be competitive. That's when I'd get concerned.

As a shareholder, I expect to see overall margins drop somewhat over the next quarter or two if ATI is successful in implementing its strategy. Assuming NVDA takes back some of the high-end market share and ATI earns the favour of OEMs with RV370 and its RS300 (and improved RS350) chipset, it will be selling proportionally more low-margin silicon than in previous quarters. Then again, growing Imageon revenues might help compensate.

[Seiko]

Well this is a much more aggressive ATI of late and although I know Joe won't be too impressed as a share holder I honestly think it's time for ATi to lower margins a little and put out another stunning card for us consumers and gamers.

On the contrary...there is good PR value to be had in being the perceived market leader, and if lowering margins helps that, it's not necessarily a bad thing, and can ultimately be a good thing.

There's nothing wrong with giving margins now and then, As long as there is a calculated return on that investment. The bottom line is long term growth and profitabilty.

Something is "wrong", if you end up having to sacrifice margins continuously, just to be competitive. That's when I'd get concerned.

As a shareholder, I expect to see overall margins drop somewhat over the next quarter or two if ATI is successful in implementing its strategy. Assuming NVDA takes back some of the high-end market share and ATI earns the favour of OEMs with RV370 and its RS300 (and improved RS350) chipset, it will be selling proportionally more low-margin silicon than in previous quarters. Then again, growing Imageon revenues might help compensate.

Is it fair to assume now then that ATI will consider the 9800PRO as it's main stream card? With prices as low as £135.00 I think it naturally falls into that bracket anyway but are ATI planning on replacing it with another 8 pipline card with greater margins but again no new features for the user?

[Natoma]

Yea the 9800 Pro 128MB is selling for $200 on pricewatch while the 9600XT is selling for $150. When the X800 launches, I'm sure the 9800 Pros will drop in price to the $150 range while the 9600s will drop to sub-$100. That's a great mainstream card offering if you ask me. :)

[aaronspink]

Anyone seen this pic before? Fake?

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

From Hard OCP (http://www.hardforum.com/showthread.php?s=8568674d4bc527f5fb5dc7a6febab4c7&threadid=752237)
Pic is real. Cooler might be real, but it isn't yet. (And I'm very annoyed about that.)

Anyone know what that connector at the end of the heat sink is for? Is it just on the proto boards?

Aaron Spink
speaking for myself inc.

[Bouncing Zabaglione Bros.]

On the contrary...there is good PR value to be had in being the perceived market leader, and if lowering margins helps that, it's not necessarily a bad thing, and can ultimately be a good thing.


Yes, ATI has had one sucessful product cycle. The public perception of ATI now being a leader instead of an also-ran has just filtered into the mainstream. Another cycle of success in the market will consolidate that leadership position amongst the mainstream as well as the early adopters.

[Seiko]

Yea the 9800 Pro 128MB is selling for $200 on pricewatch while the 9600XT is selling for $150. When the X800 launches, I'm sure the 9800 Pros will drop in price to the $150 range while the 9600s will drop to sub-$100. That's a great mainstream card offering if you ask me. :)

Well yes, but considering it's taken 20+ months to get the R3xx series of chips and their respective performance and feature set to the mainstream market segment it's become obvious I'm far too impatient for this market place ;) I still can't help feel that we've stalled somewhere along the line since the 9700Pro? Maybe it was simply the 9700Pro pushed further up the curve than any other previous offering but long gone are the days where by you can see a new and very large improvement every year.

With 16 pipelines going to dominate the high end, 8 in the middle and 4 with the low end I wonder how the range of FPS will pan out (well don't we all) ?

Game devlopers will face some complex dsign decisions.

edit = typo

[kemosabe]

Yea the 9800 Pro 128MB is selling for $200 on pricewatch while the 9600XT is selling for $150. When the X800 launches, I'm sure the 9800 Pros will drop in price to the $150 range while the 9600s will drop to sub-$100. That's a great mainstream card offering if you ask me. :)


I just can't see the 9800 Pro going that low. Right now it's a reaction to the 5900XT that NVDA desperately introduced to rescue the 5700 Ultra and keep some market share in that segment, otherwise the 9800 Pro would still be over $250. Even the 9700 non-Pro hasn't made it down to $150 territory, and they'll all probably be discontinued before getting there. Don't forget RV380 will probably occupy the $150 slot, and I agree with Joe that R420-based (SE?) versions will replace R350/360 in the $200-$300 segment.

[Ailuros]

After reading all of the 11 pages of this thread (yes I am a masochist), I've decided that I've grown a definite allergy against numbers like 8,12,16 and the word pipeline. :x

Other than that I do have that weird feeling that ATI will sustain it's MSAA quality lead, NVIDIA lose it's AF quality lead. Which of the two carries the performance crown I can't possibly foresee, yet the more details float in my direction, the more I think that ATI hasn't that much to worry about after all.

And no it's definitely NOT because of the "p- word" crap or the prementioned numbers.

[digitalwanderer]

What ever happen to the whole hype concernining how the "420" in the project name r420 stands for in April?

Raystream
Nope, wrong reference. It's definately meant as a tea-time joke which I still find mega-cool. 8)

[Megadrive1988]

here is what I posted to the Rage3D console forum on the subject of R420 having 16 pipes, and what that might mean for ATI's 2 console VPUs (for Nintendo and MS) that are still in development.

quote me

excuse me if this post gets somewhat long winded..


If ATI's new highend PC VPU, the R420, does indeed have 16 pixel pipelines, what does that *likely* mean for the VPUs ATI is designing for the Xbox 2 and Nintendo's N5 console, in terms of amount of pipelines?

I'd say it means that both of those console VPUs will have at least 16 pipes--being that they should both be at least as advanced, if not more advanced and more powerful than the R420. because those console VPUs are still in development unlike the R420 which was probably finished many months ago.

Although I am hesitant to jump to the conclusion that the Xbox 2 VPU and the N5 VPU (which btw are not the same exact chip) might have 32 pipelines each, well they might, especially if R500 has 32, I think it is obvious the console chips will have 16 pipes, minimum.

would you guys not agree?



ok, both the ArtX/ATI Flipper in Gamecube and the Nvidia NV2A in Xbox have 4 pixel pipelines. (although the NV2A has more TMUs per pipe than Flipper)

for comparisons sake, Nintendo 64's SGI-designed RCP (Reality Co-Processor) GPU had 1 pixel pipeline, as far as I know.
(the Reality Display Processor within the RCP)

So going from Nintendo 64 to Gamecube, gave us a 4x increase in number of pixel pipes. from 1 to 4.

in a few years, we should be going from 4 pipes (in Gamecube's Flipper and Xbox's NV2A) to at least 16 pipes in Nintendo's N5 and Xbox 2. again, at least. so that is at least another 4x increase in number of pipes. understand ?

now obviously if N5's VPU or Xbox 2's VPU has more than 16 pipes, the increase over Gamecube/Xbox will be more than the increase from N64 to Gamecube.

see what I'm saying?

and yeah, I know that the number of pixel pipelines is FAR from everything that determines how good a GPU/VPU is. I am simply keeping this focused on raw pixel power/pipes :D

since a lot of people seemed to be obsessed with number of pipes, I can be too, so this is my thread on the subject, for consoles. :lol:

The other major factor, as far as raw fillrate is concerned, is the clockspeed of the GPU/VPU. but that's for another thread. :D




it hasnt got much response. maybe the ppl in that forum arent as much into technology :(

[Dave Baumann]

Console demands are very different from PC demands. Expect the configuration to be very different.

Consoles deal with lower resolutions and new titles only, hence they can be designed specifically for the era (i.e. expect lots of "high quality" pixels in a console). PC's have high resolutions to deal with and a greater variety of functionality use (we are still seeing games with very little shader adoption coming out now) so PC graphics hardware needs to be a little more flexible in how they render.

[ChrisRay]

Well if Nvidia Goes Angle Dependent AF, it'll be probably the last thing which has held me to using there hardware, I absolutely detest Angel Dependent AF *sigh*

God I hope it aint true .

[Megadrive1988]

Console demands are very different from PC demands. Expect the configuration to be very different.

Consoles deal with lower resolutions and new titles only, hence they can be designed specifically for the era (i.e. expect lots of "high quality" pixels in a console). PC's have high resolutions to deal with and a greater variety of functionality use (we are still seeing games with very little shader adoption coming out now) so PC graphics hardware needs to be a little more flexible in how they render.

All true, but still, the Gamecube and Xbox GPUs both had 4 pixel pipelines which was the same amount of pipes as most PC GPUs of the time, fall 2001, when we had Nvidia NV20/GeForce 3/Ti200/Ti500 and ATI's R200 / Radeon 8500.

[plat]

Console demands are very different from PC demands. Expect the configuration to be very different.

Consoles deal with lower resolutions and new titles only, hence they can be designed specifically for the era (i.e. expect lots of "high quality" pixels in a console). PC's have high resolutions to deal with and a greater variety of functionality use (we are still seeing games with very little shader adoption coming out now) so PC graphics hardware needs to be a little more flexible in how they render.

next gen consoles are going to need almost 3x the resolution of todays consoles

[McDusty]

... and the PlayStation 2's GS has 16 pipelines.

[tEd]

Well if Nvidia Goes Angle Dependent AF, it'll be probably the last thing which has held me to using there hardware, I absolutely detest Angel Dependent AF *sigh*

God I hope it aint true .

I don't think that will be the case. At least not from the implementation. It may be so that they will have angle dependencie as an optimization but still be able to unlock the full quality somehow. I think i read a recent interview with nvidia where they say that they want to give the user the ability to unlock full the quality settings via coolbit. It seems that both ati and nvidia are not be able to give the full quality options via CP per default unfortunately because of the massive benchmark war going on

[digitalwanderer]

It seems that both ati and nvidia are not be able to give the full quality options via CP per default unfortunately because of the massive benchmark war going on
Uhm, WHAT?!?!

When is the last time ATi hasn't had full quality options available by the control panel? I know nVidia forces image reduction on their users, but what are you accusing ATi of? :|

[tEd]

It seems that both ati and nvidia are not be able to give the full quality options via CP per default unfortunately because of the massive benchmark war going on
Uhm, WHAT?!?!

When is the last time ATi hasn't had full quality options available by the control panel? I know nVidia forces image reduction on their users, but what are you accusing ATi of? :|

well even with the quality option you get trilinear only on the first texture stage and bilinear on all others. Recent problems with AF in some games like the farcry demo and battlefield vietnam(there are others)on ati hardware occuring because of this kind of optimizations.

add mipmap banding in some games like ut2003,mafia,vietnam,postal2 and others and you see that you get not the full quality on what the hardware would be able to do

[BRiT]

Would ya mind offering your dustbuster2 rumor again for those that haven't had time to decipher good rumors from constant <bleep> flamings/off topic nonsense? :)

This one: "waterblock on the card with a little pump/radiator you need a drivebay free for" ?

[Dave Baumann]

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

Did anyone notice the little component at the top right of the image?

[indio]

isn't that a alarm for overheating ?
edit :: or is buzzer the correct terminology?

[Joe DeFuria]

Did anyone notice the little component at the top right of the image?

Looks like a sound "buzzer"?

Perhaps it's an early warning siren to alert the user to evacuate if temps get too high? Or maybe it's for "sound cancellation" technology? :twisted:

Seriously...hadn't noticed it...and am now of course curious as to what the hell it is.

[digitalwanderer]

Did anyone notice the little component at the top right of the image?

Looks like a sound "buzzer"?

Perhaps it's an early warning siren to alert the user to evacuate if temps get too high? Or maybe it's for "sound cancellation" technology? :twisted:

Nope, it's in case the onboard reactor goes critical... ;)

[Joe DeFuria]

http://www.kprf.ru/clipart/misc/nuclear_explosion3.jpg

[Raystream]

http://www.kprf.ru/clipart/misc/nuclear_explosion3.jpg
Nice... Something to look forward too! :lol:

Raystream

[Megadrive1988]

http://www.kprf.ru/clipart/misc/nuclear_explosion3.jpg


16 megapipe yeild 8)

[indio]

http://www.kprf.ru/clipart/misc/nuclear_explosion3.jpg


16 megapipe yeild 8)

That's the only kind of mushroom I wouldn't eat ...

[OpenGL guy]

It seems that both ati and nvidia are not be able to give the full quality options via CP per default unfortunately because of the massive benchmark war going on
Uhm, WHAT?!?!

When is the last time ATi hasn't had full quality options available by the control panel? I know nVidia forces image reduction on their users, but what are you accusing ATi of? :|
well even with the quality option you get trilinear only on the first texture stage and bilinear on all others. Recent problems with AF in some games like the farcry demo and battlefield vietnam(there are others)on ati hardware occuring because of this kind of optimizations.
This is incorrect. The forced AF problems with BFV were caused by a bug. Forced AF problems with Farcry? Wasn't this same problem found on other vendor's cards as well with forced AF?
add mipmap banding in some games like ut2003,mafia,vietnam,postal2 and others and you see that you get not the full quality on what the hardware would be able to do
If you enable AF in games that support it, there is no problem.

[ninelven]

nm, heh.

[demalion]

tEd,

On the one hand, your statement is wrong, because there is the control panel option of application preference that allows you to avoid that issue by delivering what it says it does.

On the other, you could say that ATi isn't doing all they could do in offering functionality to the user for applications that don't set their own aniso levels, and have good cause to be disappointed in ATi (for benchmark wars, as you state) and developers (some of them, for varying degrees of laziness, though I don't think Far Cry belongs on the list) for that.

But you also don't seem aware, by your discussion, that there has been a "coolbits" type setting to unlock trilinear filtering on texture stages for ATI cards via the control panel. AFAIK, it has been there for as long as this issue has been there. I don't think a "coolbits" type solution that obscures such options is sufficient, though better than nothing.

In any case, I see application profiles as, hopefully, a way to address such issues, because it can allow an avenue to serve consumers with a full set of features/tradeoffs in profiles while still allowing an easy "default" behavior for "benchmark wars" (or, perhaps more accurately, "the propogation of lazy/incompetent benchmarking comparisons"). I suppose we'll have to wait and see.


...

On a completely unrelated pet peeve note, reactors go "critical" all the time, and on purpose. "Meltdown" is still as bad as you think, though :shock: :P. Wait, I'm actually not familiar with terminology for some of the stranger (to me) coolant materials...I'd hope they'd pick something less likely to encourage nervousness in people familiar with other reactors, though.

[tEd]

This is incorrect. The forced AF problems with BFV were caused by a bug. That may well be. Still i could get around this problem with a 3rd party tool which means there was a solution right away

Forced AF problems with Farcry? Wasn't this same problem found on other vendor's cards as well with forced AF?
no and i could get around the problem with a 3rd party tool which means there was a solution right away

If you enable AF in games that support it, there is no problem.

i know that. The point is that there would be a solution to all the little problems i mentioned if IHV's would expose the full functionality to the user which is on the table

I've seen so many discussions in message boards about AF doesn't work here , i get artefacts with AF there and so on and i knew there is a solution but people couldn't use it because it's not exposed in the CP. It's just sometimes alittle frustrating.

[tEd]

tEd,

On the one hand, your statement is wrong, because there is the control panel option of application preference that allows you to avoid that issue by delivering what it says it does.

On the other, you could say that ATi isn't doing all they could do in offering functionality to the user for applications that don't set their own aniso levels, and have good cause to be disappointed in ATi (for benchmark wars, as you state) and developers (some of them, for varying degrees of laziness, though I don't think Far Cry belongs on the list) for that.

well i mentioned farcry because there were problems with forced AF in the demo. The full game did have a ingame AF option but only up to 4xAF

But you also don't seem aware, by your discussion, that there has been a "coolbits" type setting to unlock trilinear filtering on texture stages for ATI cards via the control panel. AFAIK, it has been there for as long as this issue has been there. I don't think a "coolbits" type solution that obscures such options is sufficient, though better than nothing.

you're right i'm unaware of such coolbits and for now i would say it doesn't exist because if it really does exist and i don't know of it then i'm pretty sure that it must be really hard to get to that information in the first place which makes it kind of useless again

Anyway there is a mode in the driver which is exposed by rtool where the driver forces the AF degree but let the application choose the mipmap filter. It's a great mode because you get the full quality plus compatibility. The most problems doesn't happen because you force bilinear they happen because a mipmap filter gets forced. If it would be exposed in the CP in some way or another then that would have helped to get around alot of litte AF problems in many games and that's all i want as a customer[/quote]

[MuFu]

Anyone seen this pic before? Fake?

http://www.angelfire.com/theforce/barubal/6800_Ultra_F_3qtr_vert.jpg

From Hard OCP (http://www.hardforum.com/showthread.php?s=8568674d4bc527f5fb5dc7a6febab4c7&threadid=752237)
Pic is real. Cooler might be real, but it isn't yet. (And I'm very annoyed about that.)

Anyone know what that connector at the end of the heat sink is for? Is it just on the proto boards?

Aaron Spink
speaking for myself inc.

A daughtercard perhaps - likely for something with through-board standoffs and hefty pins. Whatever it is probably requires those 5 SMT components as they're not present in the earlier leaked photos either.

Maybe you can add a DC with another GPU? :D :roll:

MuFu.

[T2k]

Maybe the change from 12 to 16 pipelines is due to yields not competitive pressure.The chat around here suggest that ATI has been very successful in achieving high yields.

That simply can't be possible because Chalnoth says Nvidias engineers are better http://www.artnskins.com/e107_images/emoticons/glasses2.png

Yeah, it's clear, just check their card size: bigger is better for Chalnoth....http://forum.hwsw.hu/html/emoticons/nevet.gifhttp://forum.hwsw.hu/html/emoticons/nevet.gifhttp://forum.hwsw.hu/html/emoticons/nevet.gifhttp://forum.hwsw.hu/html/emoticons/nevet.gifhttp://forum.hwsw.hu/html/emoticons/nevet.gif

[Dave Baumann]

Can we calm this stuff down a little please. Less of the gloat posts and less targetting of individuals please.

[DemoCoder]

Muahahah, my corporation is going to OWN your corporation. Muahaha! Hahaha. Feel it! I win! I am vindicated. Ahahahah! 16 pipes, yah---hooo!! Eat it beotch. Your mother wears combat boots with huge heatsinks! OWNED and PAWNED!


:) :) :)

[John Reynolds]

Yeah, well, your mother sucks louder than Dustbuster 2. 8)

[BRiT]

Did anyone notice the little component at the top right of the image?

Are you sure that's a component and not a place for a tube connector for the watercooler? :twisted:

[T2k]

Yeah, well, your mother sucks louder than Dustbuster 2. 8)

:shock:

PS: :lol:

[digitalwanderer]

On a serious note, WHY does it have a buzzer on it?

[mikechai]

Can't wait until next week when the Dual NV40 solution is revealed :twisted:

:shock: :shock: :shock:

[AlphaWolf]

On a serious note, WHY does it have a buzzer on it?

Perhaps related to the asterisks that come with the dual molex's.

[OpenGL guy]

This is incorrect. The forced AF problems with BFV were caused by a bug. That may well be. Still i could get around this problem with a 3rd party tool which means there was a solution right away
And you completely missed my point which was that there was a bug. The "third party tool" doesn't do what many people think it does.
i know that. The point is that there would be a solution to all the little problems i mentioned if IHV's would expose the full functionality to the user which is on the table
A bold, and incorrect, statement.
I've seen so many discussions in message boards about AF doesn't work here , i get artefacts with AF there and so on and i knew there is a solution but people couldn't use it because it's not exposed in the CP. It's just sometimes alittle frustrating.
The control panel is not meant to be a panacea. The best option is that games start supporting AA/AF so that it gets done the way the developer intended.

[Razor04]

On a serious note, WHY does it have a buzzer on it?
So you can have fun trying to figure out if it is the Dust Buster or your motherboard beeping when your system gets FUBARed... :lol: Ok that wasn't so serious but I couldn't resist. Most likely it is to warn users about the power requirements the beast needs.

[Joe DeFuria]

On a serious note, WHY does it have a buzzer on it?
So you can have fun trying to figure out if it is the Dust Buster or your motherboard beeping when your system gets FUBARed... :lol: Ok that wasn't so serious but I couldn't resist. Most likely it is to warn users about the power requirements the beast needs.

Is it really a sound buzzer of some type? Hell, I was just joking...

[Mark]

On a serious note, WHY does it have a buzzer on it?
So you can have fun trying to figure out if it is the Dust Buster or your motherboard beeping when your system gets FUBARed... :lol: Ok that wasn't so serious but I couldn't resist. Most likely it is to warn users about the power requirements the beast needs.

Is it really a sound buzzer of some type? Hell, I was just joking...it looks like one of those mini speakers you see on some motherboards.

[flick556]

Often system crashes occur do to gpu related problems it would be nice if your graphics card could issue some error codes on reboot.

[BRiT]

Is it really a sound buzzer of some type? Hell, I was just joking...

If it's not a buzzer or a water-cooler connector then it must be a flux-capacitor... 8)

Or maybe you put your weed in there...

[elroy]

I'm on the wrong side of the world. I checked the boards before I went to bed, nothing. 8 hours later, wake up and check the boards, and there's 13 FRICKIN' PAGES of posts to read!!!!!!!



I think nVidia should buy out Zalman, and, like MuFu said, slap 1 kg of copper on NV40 with a few heatpipes for good measure. I sure don't want to see/hear DB2! I thought nVidia would have learnt by now that smaller fan usually = louder fan.