PDA

View Full Version : "nv40 Exclusive"


Pages : [1] 2 3 4

Cotita
06-22-03, 09:26 PM
From the front page

To cut to the chase, the NV40 is currently being targeted for the following:

550MHz-600MHz Core
700MHz-800MHz Memory


Are you kidding me?

anyone could make that up.

For one thing. if the memory is 700-800mhz ddr its not a lot of bandwith for a 600mhz core. If on the other hand its 1.4-1.6ghz DDR (dreaming) it next to impossible since there won't ever be DDR memory that fast and we won't be seeing DDR2 at those speeds at least in a couple of years. Not even GDDR3 may reach that kind of speeds.

Ok I'll make my own numbers based on "reliable sources"
1 ghz core
2 ghz DDR3
keep on dreaming

Pafet
06-22-03, 09:32 PM
go go Uttar !! :clap:

sbp
06-22-03, 09:49 PM
Yummm...nothing like freshly baked hype. http://sbp777.homestead.com/files/drool.gif

fantomas
06-22-03, 11:03 PM
this 'leak' is nvidia's marketing strategy to a T...create some hype, create some free advertising.. anyway from xbitlabs:

Apparently, yields of NVIDIA’s products made using 0.13 micron technology are still lower than the company originally thought them to be. Furthermore, because of very high complexity of modern graphics processors, it takes more time to re-spin them. According to Mike Hara, it takes around 10 weeks to re-spin a 0.15 micron GPU, and now it takes around 14 weeks (NVIDIA’s original target was 12 weeks) to re-spin a 0.13 micron GPU. As a consequence, the yield issues are addressed slower than before and this may also impact the announcements of more advanced products.

Because the yields of the GeForce FX 5600 and GeForce FX 5900 parts are lower, NVIDIA cannot flood the market with such GPUs due to troubles with supply. They indicate relatively strong demand for the new graphics processors and express intentions to build more inventory. As 0.13 micron products represent a small part of the company’s business, its gross-margins should not be tangibly affected this quarter.

Mike Hara commented that he had not said which product would come out this Fall, but he said that this product had already been taped out at IBM and its commercial manufacturing will begin in the fourth quarter. I believe this was told about the code-named NV40 graphics processor that is set to emerge later this year. Note that in case it will require two re-designs after the initial tape out (and if every new redesign takes 14 weeks), the graphics chip will only be able to come here in very late 2003 (six or seven month from now), but it is more likely to massively appear in 2004. If NVIDIA requires only one re-spin of the NV40 to achieve proper performance and yield, NV40 may make it into the market this Fall.

MAY make it to the market this Fall. MAY make it.

digitalwanderer
06-22-03, 11:09 PM
They MAY make a few for reviewers, but that'll play against them bigtime after last year's fiasco. :(

Very limited availability by x-mas, quantities in Jan-Feb....IMHO.

The Baron
06-22-03, 11:14 PM
I think we'll see it this winter if the drivers come along quickly. Yields won't be a problem with IBM, so it should be interesting.

Ninja Prime
06-23-03, 03:36 AM
If, as Uttar suggests, it's a NV35 refresh, they must be pulling another BS 2 times faster than previous generation thing. With only 100-150MHZ core difference, only 20-33% or so, they won't be able to get actual games up to twice as fast as NV35. Maybe if they "optimize" drivers... :p

Edit: I also don't think they will have GDDR2 or 3 that fast ready for high volume production by 1stQ2004.

Edit2: Either the original NVnews post was edited or I just read it retarded, but I swear I read Uttar saying something about it being a NV35 refresh somewhere.... :confused:
However, with a new core and more transitors, I find it hard to believe that they will reach 600MHZ on .13 micron, unless they want to go back to the FX5800 jet-engine heatpipe. Maybe IBM is going to do it in .11 micron? That sounds unlikely to me too... Hmmm...

silence
06-23-03, 04:20 AM
i understood they'll have 2 products...nv35 refresh, with higher clock, which should be possible...and nv40 with completly new core and new features.

looks like that due to longer time needed for redesign of taped out product both NV and ATi are going to play more with clock speeds, releasing products clocked not_as_high_as _possible , so they can release "new" product later with higher clock..... making it look like they manage to keep 6 month cycle even if they can't.....more PR bull, just look at ATi lately....NV will prolly adopt same tactics.....

Dazz
06-23-03, 04:34 AM
At the current rate 600MHz an't possible due to low yields.

SurfMonkey
06-23-03, 04:51 AM
It'll probably be, in traditional terms, either an 8x2 core or a 16x1 core, and the memory clocks are inline with GDDR3, which will providing 40+GB\sec of bandwidth at those speeds.

volt
06-23-03, 05:17 AM
Originally posted by Cotita
Ok I'll make my own numbers based on "reliable sources"
1 ghz core
2 ghz DDR3
keep on dreaming

I won't comment on that, but I bet Uttar will.

Geforce4ti4200
06-23-03, 05:24 AM
550 core isnt unreasonable and 800 ram. This ram will be as good as on the fx5900u while the .13 microns will allow a 100MHz increase on the core. Nothing special, like 20% faster :confused:

silence
06-23-03, 05:30 AM
Originally posted by Dazz
At the current rate 600MHz an't possible due to low yields.


hmmmmm.......at TSCM it isn't.....but it says it taped out at IBM and they have much better process. we need to wait and see what is IBM capable of doing....they have so much more experience with .13 then TSCM.....maybe yields at IBM might finally bring top Nvidia chips in quantities to market?

Evildeus
06-23-03, 05:34 AM
GDDR3 should be available at those MHz around Q3, so that's not impossible.

Nv themself told us that the issue with the yield is not TSMC unique fault. So if Nv can do something about the receipe, then there's no issue to go to 550-600MHz for the core.

What i'm interesting in is:
-New improve AA quality
-Possibility to run any game @16*16*4/6/8AA+8/16AF on a 4GHz ring (well most of....)
-Ubber boost in PS2/3 like 5 times what it is now (meaning only 2-3 times faster than R350)
-FP32/16 and no F*cking FX12
-No F*cking cheat to take the lead
-And retake at last the performance and quality crown :afro:

Uttar
06-23-03, 06:16 AM
Cotita: Hmm, it IS 700Mhz-800Mhz RAW.

That means 44.8-48GB/s considering a 256-bit memory bus ( I must admit I got no info that it's a 256-bit memory bus, but it'd make no sense to use a 128-bit one, since overall, it'd cost even more than the NV35 and be less effective! )

And frankly, no problem if you don't trust me - but I certainly know I trust my source :) Never has been wrong before, if you know what I mean.

Now, AFAIK, and this comes from another source, also very reliable ( but the info is also less recent, so it *might* be outdated, doubt it though ) - the NV40 uses GDDR2.
I was very surprised by this myself - seemed like insane for GDDR2 figures.
But considering there's already 600Mhz GDDR2 IIRC, having 700Mhz in Q4 is not insane - and 800Mhz would then kinda be the best-case scenario.
As I said, the fact it uses GDDR2 might be outdated - or heck, the memory controller could support both ( GDDR2 for the $399 part and GDDR3 for the $499 one? Who knows - I don't ) - but it's still the most likely.

Also, the NV40 is *not* a NV35 refresh. It's a completely new core - just like the NV10, NV20 or NV30. The biggest feature additions over the NV30 are texture lookups in the Vertex Shader and branching in the Pixel Shader, making it PS3.0. / VS3.0. compliant.
A very cool feature is also the addition of a Programmable Primitive Processor. Few details on the implementation as of yet, though :(


So, to summarize, the targets are, I believe:
- 550+Mhz core clock on IBM's 0.13µ process ( no idea if it uses SOI or not )
- 44.8GB/s+ memory bandwidth using most likely GDDR2, maybe GDDR3

What's a given, too, is that Loki will also feature that type of memory - details are obviously hard to tell, because it's all targets right now ( and I don't even have Loki's targets, eh... )

Finally, a disclaimer I believe volt should have added, but oh well:

Those numbers are targets. Problems might arise, such as an overoptimistic look at IBM capabilities or too aggressive RAM decisions. In conclusion, the final clock speeds might be either higher or lower than those targets.
However, the info is still very accurate - I firmly believe those are nVidia's current targets.

BTW, for avaibility - the idea there is still "hopefully before year end". Speculation: Don't hope for more than limited avaibility though - more like reviews in early December and a few cards in stores several days before christmas, just to make sure Loki doesn't steal the spotlight.

Anyway, I hope you appreciated the info. Remember though that the most reliable piece are those targets - the rest is also very reliable, but still not as much :)

Finally, let's admit *both* NV40 and R420 are going to "own". For now at least, instead of saying "The R420 is gonna kill that anyway" or "Can't do better, in your face ATI!" ( not like I heard that yet, eh ) - let's just hope they'll deliver both on time ( that is, Comdex 2003 ) so we can have an interesting battle - okay?


Uttar

nutball
06-23-03, 06:57 AM
Originally posted by Uttar
Finally, let's admit *both* NV40 and R420 are going to "own".

A nice though Uttar, I don't think many people are that broad minded.

Hanners
06-23-03, 08:42 AM
Originally posted by Uttar
Finally, let's admit *both* NV40 and R420 are going to "own". For now at least, instead of saying "The R420 is gonna kill that anyway" or "Can't do better, in your face ATI!" ( not like I heard that yet, eh ) - let's just hope they'll deliver both on time ( that is, Comdex 2003 ) so we can have an interesting battle - okay?

Absolutely, it's going to be an exciting race. Again.

I'm not going to buy either of these cards though... I'm not. If I say this to myself enough it might come true...

BigFish7
06-23-03, 08:47 AM
to me it seems totaly real...
i was hoping for 550MHz
550-600 sounds fine to me on 0.13 (if everything goes well)
i find the memory speed a little to high but once again i'll trust uttar on that one...
we aren't talking about out of this world clock for the memory, we are talking about really High clock.
i'm kinda 100% sure it would be
550MHz - core
700MHz - memory
256bit
8X2 architecture

lets hope for good product from both camps ;)

BigFish7
06-23-03, 08:48 AM
Originally posted by Hanners
Absolutely, it's going to be an exciting race. Again.

I'm not going to buy either of these cards though... I'm not. If I say this to myself enough it might come true...

ahh don't worry just don't eat a couple of months and you should be fine with teh $ :D

borntosoul
06-24-03, 12:56 AM
i read someplace that both ati and nvidia ordered a heap of gdd3.

Ninja Prime
06-24-03, 01:24 AM
i read someplace that both ati and nvidia ordered a heap of gdd3.


I don't think so... it was just this month that Micron said that they had just barely shipped test samples to ATI and Nvidia.

borntosoul
06-24-03, 01:38 AM
they may have placed orders for a centain date, i didnt mean to imply that they will be getting them soon

Geforce4ti4200
06-24-03, 03:01 AM
well didnt the geforce2u have 500MHz ddr ram like 9 months ago? so 700MHz sounds reasonable but I believe itll cost like $700 for nv40 cards if they use that fast ram :eek:

Uttar
06-24-03, 04:45 AM
Err, IIRC, the GeForce2U was released a LOT more than 9 months ago, and it used 500Mhz *effective* DDR, that means 250Mhz DDR Raw. Here, we're talking about 700Mhz DDR Raw or 1400Mhz Effective.

I got no info related to the R420 refresh or the NV45, but I'd still bet big time they'll both use GDDR3.


Uttar

DSC
06-24-03, 04:55 AM
Geforce2 Ultra was announced in August 2000. Thats way more than 9 months. ;)

Even the GF4s are 1 year old now.