View Full Version : Jen-Hsun Talks NV4x
wow this is right from the Jen Hsun
http://beyond3d.com/#news10574
rets:afro2:
Edit: Can u not be so rude thanks
Jen-Hsun commented that the rumours are suggesting the announcement of NV4x in "a couple of short months" and that these are "likely to be correct".
E3 is the launch time for NV40? Damn. I wanted a new card soon. :o
Nothing too much that hasn't been said over the NV3X line. It was much speculated that the NV35/38 manufacturing had yield problems. This time they say they have erm designed the chip/die to increase yields. That remains to be seen.
What is interesting is that they say its also designed to be scable. How scable? or de-scable? In terms of speed? or modularity? as seen with nv3X line. Ie nv36 and nv34 etc
I sure hope the nv4X mainstream delivers performance akin to real 8 pipelined gpus. About time we had them although 9500pro was there but that got canned soon.
Well Hsun mentions 2X the performance...so we shall just have to see where the "2X" will be applied.
Maverick123w
03-02-04, 11:32 PM
He mentioned far more than 2x the performance :D
Originally posted by Maverick123w
He mentioned far more than 2x the performance :D
I hope for more in terms of AA and AF combined performance in all games:D
what intalectual propery is jen talking about...maybe this has to do with xbox 2 and backwords compatibility ?????
Maverick123w
03-02-04, 11:51 PM
Originally posted by MUYA
I hope for more in terms of AA and AF combined performance in all games:D
Preach on :)
Originally posted by Maverick123w
Preach on :)
How about the abilty to make me toast in the morning too? ;)
Well it should run hot enough :D
Jen Hsun.... I stopped listening to him after he made his famous hallucinogenic comment :rofl
EMunEeE
03-03-04, 01:58 AM
Jen-Hsun makes me want to get an NV40 now...LOL.
Yea, hopefully its not the same way he wanted us to all get a NV30:)
sorry guys about the rude comment, sounded funny at the time :spank:
rets:D
-=DVS=-
03-03-04, 04:58 AM
I think Jen-Hsun does all Nvidias PR work by him self :D
jen-hsun makes me want to go out and buy the stock if the card is 16x2 @500 mhz then i think i will go out and buy the stock
rets
:afro2:
mikechai
03-03-04, 06:01 AM
I respect Jen Hsun.... ;)
zakelwe
03-03-04, 08:43 AM
About 8 months ago I was wondering whether nv40 and r420 would be quick enough for Doom 3 and games that might be made for that engine ( SOF3 ? RTRTCW ( can you have two returns ? ) ) but it all seems like they will have more than enough firepower and instead we have to worry about cpu power to go with it and nv40's new AA, or lack of it.
What happens if nv40 is faster but the AA is still crap ? What would people do here ?
I think I'd be tempted with the speed still, as 1600x1200 could save the day. I'm semi blind from too many erotic downloads anyhow.
Regards
Andy
Originally posted by zakelwe
What happens if nv40 is faster but the AA is still crap ? What would people do here ?
It has been speculated that the NV40 will have a new AA unit as they were playing 2nd fiddle to ATI's AA implementation. It would be shooting themselves on the foot if they carried on the gf4 AA unit which was used in the nv3x to the nv40. There have been rumours that NV will have all sorts of new AA modes. Frankly with the speculated bandwidth of NV40, you might just wanna forget 4XAA and start of with 8XAA!!! thats me hoping rather than fact ;) RGMSAA is probably the order of the day in the new NV4X line which we all know is what ATI implemented in their r3XX line and what was missing in the NV3X line (except 2XAA). Of course you can expect to have hybrid AA techniques with all that fillrate and bandwidth.
PS- you got a FTP server up? ;)
zakelwe
03-03-04, 09:03 AM
Originally posted by MUYA
It has been speculated that the NV40 will have a new AA unit as they were playing 2nd fiddle to ATI's AA implementation. It would be shooting themselves on the foot if they carried on the gf4 AA unit which was used in the nv3x to the nv40. There have been rumours that NV will have all sorts of new AA modes. Frankly with the speculated bandwidth of NV40, you might just wanna forget 4XAA and start of with 8XAA!!! thats me hoping rather than fact ;) RGMSAA is probably the order of the day in the new NV4X line which we all know is what ATI implemented in their r3XX line and what was missing in the NV3X line (except 2XAA). Of course you can expect to have hybrid AA techniques with all that fillrate and bandwidth.
PS- you got a FTP server up? ;)
LOL, no I've got a very large hard drive disguised as
A: Floppy 3.5 ( 1.44MB)
in case people snoop around ;)
Well hopefully that will be the case, with 51GB of bandwidth there should be more than one way of bashing it into shape anyhow.
Regards
Andy
nForceMan
03-03-04, 09:42 AM
Originally posted by zakelwe
What happens if nv40 is faster but the AA is still crap?
I have a better/another question: What happens if r420 is slower (than NV40) and its AF (http://www.3dcenter.de/artikel/2003/11-21_a_english.php) and alpha blending (http://www.anandtech.com/video/showdoc.html?i=1931&p=9) are still crap? :rofl
Mariner
03-03-04, 10:45 AM
Ha! With those pictures, it seems obvious that NVs alpha blending has some problems. Otherwise, how is it possible for the 'glow' to show through the character's head? Is his head not solid? Something's going wrong somewhere there. In fact, you couldn't have picked a worse example for anti-ATI FUD! :D
Instead of all of these silly games, let's just wait until the two new products (which we know virtually nothing about, despite all the rumour and innuendo) are available before making decisions.
Use the intervening time to do something useful. :p
Crap! I just realised I'm wasting my own time making this post! Ta ta.
nForceMan
03-03-04, 11:17 AM
Originally posted by Mariner
Ha! With those pictures, it seems obvious that NVs alpha blending has some problems. Otherwise, how is it possible for the 'glow' to show through the character's head? Is his head not solid? Something's going wrong somewhere there.
It seems you have absolutely no clue what alpha blending is.
EMunEeE
03-03-04, 11:51 AM
Originally posted by nForceMan
I have a better/another question: What happens if r420 is slower (than NV40) and its AF (http://www.3dcenter.de/artikel/2003/11-21_a_english.php) and alpha blending (http://www.anandtech.com/video/showdoc.html?i=1931&p=9) are still crap? :rofl
nForceMan makes me laugh...:lol2:
I'm just waiting for both cards to be released. The NV40 though sounds like it will be a great card.
Mariner
03-03-04, 11:55 AM
It seems you have absolutely no clue what alpha blending is.
Ooh, very snippy. I could retort that it seem you have absolutely no idea about reading comprehension, but I won't as that would be rude.
I assume you linked the Anandtech page to indicate that the degree of alphablending on the lightsabers was different on ATI and NVidia cards. This does, indeed, appear to be the case.
However, my point was that on the image from the NV card, the glow (alpha blending, if you must) from the lightsaber is visible through the body of the character holding it. This is not correct and indicates some problems somewhere whether within the software itself (probable) or in NVidia's alpha blending implementation (less likely). If, indeed, this problem lies with the software then perhaps it is possible that most of the difference in rendering between the cards lies in the implementation in this particular piece of software? This is why the example you provided was a poor one.
IIRC, there was some discussion on B3D's boards which indicated that the way ATI (R3X0) hardware calculates alpha blending is very slightly less accurate than NVidia's, but only to one level of intensity out of 255. You'd have to search for it though because I really don't have the time to look.
nForceMan
03-03-04, 12:19 PM
Originally posted by Mariner
I assume you linked the Anandtech page to indicate that the degree of alphablending on the lightsabers was different on ATI and NVidia cards. This does, indeed, appear to be the case.
However, my point was that on the image from the NV card, the glow (alpha blending, if you must) from the lightsaber is visible through the body of the character holding it. This is not correct and indicates some problems somewhere whether within the software itself (probable) or in NVidia's alpha blending implementation (less likely). If, indeed, this problem lies with the software then perhaps it is possible that most of the difference in rendering between the cards lies in the implementation in this particular piece of software? This is why the example you provided was a poor one.
IIRC, there was some discussion on B3D's boards which indicated that the way ATI (R3X0) hardware calculates alpha blending is very slightly less accurate than NVidia's, but only to one level of intensity out of 255. You'd have to search for it though because I really don't have the time to look.
I'm so sorry dude, but after this, I only have to say:
It seems you have absolutely no clue what alpha blending is. :angel:
EMunEeE
03-03-04, 12:21 PM
Originally posted by nForceMan
I'm so sorry dude, but after this, I only have to say:
It seems you have absolutely no clue what alpha blending is. :angel:
nForceMan, tell us what alpha blending is. You keep saying he doesn't know, but you haven't made me believe that you know. Right now you are wasting posts or teh spammin!!111!!!11!!
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.