NV40 INQ News

[nvnews-reader]

Originally posted by GlowStick
[B]New Generation, no LOL.

Infact, R420,nv38,nv40,r360 are not being promoted at all, they arnet out yet.


Well, Hell has been pimping Luci for ages now. Same goes for Uttar with the nv40.. although not as much pimpage.

:condom:

[GlowStick]

Originally posted by nvnews-reader
Well, Hell has been pimping Luci for ages now. Same goes for Uttar with the nv40.. although not as much pimpage.

:condom:

Yeah but its not even offical, just word of mouth, very intresting to speculate abought though.

[Paul]

Er, seems like people are mixing their codenames up. Neither the R420 nor the NV40 are mere "trivial" speed boosts (such as those to be seen in the NV38 and R360).

[nvnews-reader]

Originally posted by Paul
Er, seems like people are mixing their codenames up. Neither the R420 nor the NV40 are mere "trivial" speed boosts (such as those to be seen in the NV38 and R360).

I consider 30% improvement borderline.....

R420 is sounding like a decent refresh to me.

[MuFu]

http://smilies.xibase.com/sleep.gif

MuFu.

[StealthHawk]

Originally posted by reever2
Using that mentality, why is eveyrbody so excited about the NV35 and R350? No games use dx9 yet, right?

His mentality is wrong and stems from past generations of hardware.

Star Wars Galaxies supposedly is using DX9, and so does Tomb Raider Angel of Darkness. Not to mention Half-Life2, which will undoubtedly spawn game-class mods just like the original HL did.

[Raiken3712]

are any of those games out right now though?? not that i know of

[Paul]

SWG and Tomb Raider are both out.

[Geforce4ti4200]

well ati and Nvidia better go .11 microns for their nv40 and r420. .13 is not going to cut it much longer unless they stay with 8x1 and 600MHz or lower core clocks. This would make it like a 9700 to 9800 type of upgrade on .13 microns. If both go .11 perhaps they can do 12x2 and clock them at 500-600MHz core then later on go .09 for 16x2 and 800MHz core clock :eek:
the radeon9900 pro isnt exciting except itll cause older cards to drop a bit in price. The day when a 9800 becomes $100 will happen eventurally, perhaps in early 2007

[GlowStick]

Originally posted by Geforce4ti4200
well ati and Nvidia better go .11 microns for their nv40 and r420. .13 is not going to cut it much longer unless they stay with 8x1 and 600MHz or lower core clocks. This would make it like a 9700 to 9800 type of upgrade on .13 microns. If both go .11 perhaps they can do 12x2 and clock them at 500-600MHz core then later on go .09 for 16x2 and 800MHz core clock :eek:
the radeon9900 pro isnt exciting except itll cause older cards to drop a bit in price. The day when a 9800 becomes $100 will happen eventurally, perhaps in early 2007

.13 is plenty of technology for at least 3 years easy, ATi pritty much proved that .15 is still plenty of technology to go around. And mostly, nvidia and ati are still about profit, they are not gonna jump in a low yeild (witch means less profit) .13 is just starting to get more profitible.

[Geforce4ti4200]

Then explain why Intel is gonna go .09 for their pentium5 prescotts and amd will do that soon after with their later a64's? and why would .09 cause low yields? They better work on that then, its the same except smaller

[incurable]

Originally posted by Geforce4ti4200
Then explain why Intel is gonna go .09 for their pentium5 prescotts and amd will do that soon after with their later a64's? and why would .09 cause low yields? They better work on that then, its the same except smaller
Intel and AMD have their own FAB(s), allowing them to quickly tune their processes and products.
Intel (and AMD to an extend) develop their own processes, ending up with ample of knowledge before the first product wafers are run.
Intel and AMD both produce simpler processors using fewer transistors most of which are used up by huge on-die caches w/ built-in redundancy, higher yields are expected w/ such products.

cu

incurable

[Lezmaka]

I think the next step is gonna be .09u, only thing I know of that uses .11u is ram.

Just because Intel/AMD are going to transition to .09u this and next year sure doesn't mean nvidia/ati need to. What indications do you have that nvidia/ati NEED to move to a smaller process for NV40/R420? If it's based on transistor count, then you obviously don't remember what ati was able to do with .15u. If you're going by NV30 then don't. NV30 may not have been a good design (talking about physical design/layout) not to mention it was originally designed with low-k in mind.

They better work on that then, its the same except smaller

I don't know much compared to others here, but I do know they definately aren't the same. Design rules can be totally different, it's not a simple matter of just resizing a design.

Intel/AMD are moving to .09u because they think they need to. I think it's obvious nvidia/ati don't think they need to and I don't either.

[Uttar]

Originally posted by Lezmaka
I think the next step is gonna be .09u, only thing I know of that uses .11u is ram.

I thought that too... But a certain source managed to make me think otherwise ;) I'm not as sure for ATI, though...

The NV40 & R420 are both 0.13u, though.


Uttar

[Geforce4ti4200]

well ok then .13 on their next cards will mean tons of heat, will we be seeing cards that take an extra pci slot and leaf blower fans like the fx5800u?

[Paul]

Heat shouldn't be that much more of an issue than it is now... Perhaps less in some cases. Double height cards are still a possibility, though.

As for 0.11, it's possible you could see that on NV45. I doubt nVidia would push for it with the new generation NV50, and NV55 is far enough away to make 0.09 a reality there. Also, nVidia have always traditionally changed process on the refresh parts, along with other changes, to get the required speed boost.

I thought they'd stick with 0.13 for longer to be honest, but it's worth thinking about.

EDIT: Typo

[MuFu]

Originally posted by Paul
As for 0.11, it's possible you could see than on NV45.

Good guess. ;)

MuFu.

[Ninja Prime]

well ok then .13 on their next cards will mean tons of heat, will we be seeing cards that take an extra pci slot and leaf blower fans like the fx5800u?

I think it's more likely we will see better/more efficent HSF designs first, which the vid card market really needs a anyway IMO. I'm thinking of all copper sinks with some kind of design similar to the Vantec Aeroflow, with dual fans or just larger fans.

[Deimos]

Originally posted by Ninja Prime
I think it's more likely we will see better/more efficent HSF designs first, which the vid card market really needs a anyway IMO. I'm thinking of all copper sinks with some kind of design similar to the Vantec Aeroflow, with dual fans or just larger fans. I dunno.. if it was that "simple", then someone would have done it by now, because there really is a market for efficient low-noise cooling.

Process "maturing" (low-k, experience etc) is what I would guess will save them from a significant heat-problem for the next parts.

/Deimos

[GlowStick]

Originally posted by Geforce4ti4200
Then explain why Intel is gonna go .09 for their pentium5 prescotts and amd will do that soon after with their later a64's? and why would .09 cause low yields? They better work on that then, its the same except smaller

simple explination is Intel and AMD are the leaders in the simi-conductor industry. They naturally are the leaders foundry technology, they have BEEN on .13 for quite some time. Nvidia and ATi rely on other comapnys to make foundrys to make chips, TMSC has got .13 working properly now.