NV31 Ultra's may be clocked slower than previewed?

[McElvis]

Although the NV31's were previewed at 350/350, it appears that the actual cards clock speeds may vary between board makers.

For example, Digit Life is reporting that the ASUS NV31 Ultra will be clocked at only 325/275.

But the new solutions are still on paper, and testers simply hang around looking for what to test. Old products of 2002 which everyone is tired of? Or maybe something that is expected this Spring? But everything that could be said is said; we just hope that production cards will be at least 90% similar to what we got in the tests. I mean NV31 (GeForce FX 5600 Ultra) which is going to have clock speeds different from the reference card (for example, ASUSTeK mentioned they would be 325/275 (550) MHz against 350/350 (700) MHz of the reference solution). It makes me think that the new-comers will be slower than expected. It also concerns ATI when higher-rated cards (9600) are slower than those marked 9500. But ATI has already offered DX9 accelerators for the mainstream market (RADEON 9500/PRO) though they are still in short supply.

Edit : link http://www.digit-life.com/articles2/gffx/asus-v9900u.html

Is this going to become the norm? Reference cards clocked higher than retail cards? I really hope not [/b]

 

[Arun]

It's actually slightly more complex than that, AFAIK.

The NV31 reference Regular is at 325/275.
The NV31 reference Ultra is at 350/350.

But nVidia probably allows their partners to use NV31 Regular chips and name them as "Ultra" is they include some other stuff, such as dual DVI maybe.


Uttar

 

[MuFu]

NV31 must be such a hard chip to market. OEMs would have been told the inital projections - 450Mhz Ultra, 350MHz non-Ultra in Q4'02 and incorporated such products into their roadmap. The NV31 "range" is horribly compressed because of the poor scaleability.

I bet we'll see NV36 debut with NV35 next month/early June. The gap between NV31 and NV35 is just too large (remember ATi having a very detrimental 8500=>9700 Pro hole for a while?).

Then they should have NV34, NV35 and NV36 as their product lineup - nice and tidy, heh. 8)

MuFu.

 

[Dave Baumann]

You'll note that the NV31's previews so far are only A01 stepping - board vendors I've talked to say they won't have NV31's until about May time, indicating they are going for another spin. I don't think there will be any chips coming to replace NV31 anytime soon. However, another spin could result in higher core speeds.

 

[MuFu]

You'll note that the NV31's previews so far are only A01 stepping - board vendors I've talked to say they won't have NV31's until about May time, indicating they are going for another spin. I don't think there will be any chips coming to replace NV31 anytime soon. However, another spin could result in higher core speeds.

Perhaps. A01 eh? Means no major changes since November - I find that very hard to believe. Not quite sure how nV markup their ASIC revisions though - NV30 samples seem to be A02, although it is common knowledge that there have been more than 5 major revisions over the course of R&D.

There have been NV31 samples running at 450MHz in the lab; they are only a tiny fraction of the overall yield though.

MuFu.

 

[McElvis]

Thanks for the info Dave.

By the way, are we going to see an Beyond3D NV31 preview anytime soon? Or are you going to wait until you can get retail cards?

 

[Dave Baumann]

Not quite sure how nV markup their ASIC revisions though - NV30 samples seem to be A02, although it is common knowledge that there have been more than 5 major revisions over the course of R&D.

IMO, the 'A02' we see is the second rev of the non low-k process version. Previous ones probably were low-k ones but were all DOA.

By the way, are we going to see an Beyond3D NV31 preview anytime soon? Or are you going to wait until you can get retail cards?

Next week.

Its taken me a bit longer than expected because I'm getting results that are, in some cases, significantly lower than the Ti4200 that I'm comparing against and I want to make sure they are correct. Few of the NV31 results tally with other sites since I'm using 'correct' (Application) Trilinear in the normal rendering (all B3D reviews should have Trilinear as the default IQ these days, as all my reviews - bar NV30 by accident - have over the past year) and the only Ti4200 board I have is a rather high spec one from XFX that uses a Ti4600/4800 board design that is pretty fast (even when clocked to Ti4200 speeds).

 

[galperi1]

Man, if I were one of Nvidia's board partners I would be FURIOUS.

Not having SIGNIFICANT amount of new chips to sell since when??? the Geforce 4? That is not a good way to do business. As it stands, they won't even have the mainstream cards out till May??

 

[LeStoffer]

You'll note that the NV31's previews so far are only A01 stepping - board vendors I've talked to say they won't have NV31's until about May time, indicating they are going for another spin.

... which is why nVidia went for the 0.15 process for the NV34 and made the NV34 Ultra to cover at least some ground until the NV31 on the 0.13 process picks up in volume.

I still wonder whether ATI have the same problems with the 0.13 process (the RV350).

 

[McElvis]

Dave - looking forward to it, thanks

 

[MuFu]

Not quite sure how nV markup their ASIC revisions though - NV30 samples seem to be A02, although it is common knowledge that there have been more than 5 major revisions over the course of R&D.

IMO, the 'A02' we see is the second rev of the non low-k process version. Previous ones probably were low-k ones but were all DOA.

Ah, yes. Good call. Still can't believe that NV31 is only on it's first revision though - A01 yields were only about 15-20% at the beginning of the year and A02 was due in Feb. =\

MuFu.

 

[Arun]

Thanks for the info, Dave!
I'd be surprised if nVidia increased clock speed. Their yields probably aren't so good right now, and one respin won't suddently make them godly. They'd prefer to have higher yields to reduce cost rather than higher clocks and have insufficent avaibility and low margins.

I'd guess the NV36 would be launched in late August, with retail avaibility in early October. That would be pretty much the same dates as the NV18/NV28 last year.


Uttar

 

[Dave Baumann]

Competetively speaking, performance is what they need to aim for. NV31's die is larger then RV350 and it needs faster RAM, so unless NV are willing to take hefty hits on the margins for NV31's then I don't think the prices are going to be lower than RV350 based boards.

 

[MuFu]

I'd guess the NV36 would be launched in late August, with retail avaibility in early October.

That would suprise me. From what I know of NV36 it only involves fairly minor architecture changes so an R&D run of almost a full year seems a little excessive.

I think they'll take the hits on NV31 and try and get NV36 out quickly - like May/June.

MuFu.

 

[martrox]

Can nVidia do anything right? First the NV30, now the NV31. Neither product is readily available, yet are so uncompetitive that they need replacing. For all they (nVidia) have done right, the last year and a half really says nothing good for them........

 

[Tahir2]

I predict the NV35 will change that. If it doesn't then I will be really worried about NVIDIA and their products. At the moment its just short term.. they got caught with their pants down.

 

[MuFu]

Yeah - hopefully they'll be unveiling it at E3 as planned.

MuFu.

 

[Arun]

Yeah - hopefully they'll be unveiling it at E3 as planned.

MuFu.

Yep, let's hope so... Would have quoted you on that, but I already posted that info on GPU:RW only based on E3's expo page


Uttar

 

[BRiT]

Does anyone else see the humor in connecting the following three statements:

they got caught with their pants down.

Yeah - hopefully they'll be unveiling it at E3 as planned.

Yep, let's hope so...

Maybe it's this 1 hour time-change that makes it funny...

 

[martrox]

Does anyone else see the humor in connecting the following three statements:

they got caught with their pants down.

Yeah - hopefully they'll be unveiling <b>it</b> at E3 as planned.

Yep, let's hope so...

Maybe it's this 1 hour time-change that makes it funny...

Anyone ever read the Carl Hiaason book titled....... Sick Puppy?