NV30: Part of the process done months ago?

[Uttar]

Hello everyone,

Yes, this is speculation, but this is a theory i've seen no where else, so i'd still like to say it.

You see, Jen Hsun Huang said back in april i think that the NV30 had taped out. And then he said in a recent conference call that it didn't tape out.
But he had NO advantage to lie back in april. So what happened?

Most likely, the NV30 changed. nVidia seen TSMC wasn't ready and that they won't be ready for august. So they delayed it ON PURPOSE to have a more mature process and, at the same time, a more mature part.

However, in april, they didn't suspect that. Else they wouldn't have said they taped out.

At the same time, some nV engineers ( they got several teams ) were already working on the NV35 ( once something is done, it's time to do something else, after all! ) , and they also were fixing the NV30 bugs that the teams still working on the NV30 found - the NV35 will most likely be like the NV25 where 50% of the transistors are left unchanged from the previous generation.

Then, after some time ( i won't try to speculate, it could be 10 days, it could be 60 days. I don't know! ) , they stopped it all, seeing TSMC wasn't ready. And they renamed what was to be called the NV35 and said it will be the NV30.

However, not much work was done - bugfixes, minor features increases and maybe one or two great features. So maybe 10% of the transistors changed.

My point(s)? If what i said is true ( or partly true ), the following is highly likely:

1. The NV30 was originally planned to be about the same thing as the R300 is today, just a few more transistors to increase performance a little and higher clock.
2. They dropped it all and decided to perfection it while waiting for TSMC.
3. Part of the bug testing process is already done.

And what does the third part say us? Some bugs already being fixed, it is highly likely the NV30 will only take the minimum 90 days and will be ready in end november.


Uttar

 

[Saist]

Um? What are you smoking?

April was E3. Trust me, I was there. Nvidia made no such announcements at having Nv30 even close to finished then. 2 of the reps probably remember me harrasing them about "the next Geforce" Freaky white guy, ugly yellow backpack. They made a couple other tech announcments, a new Nforce and some other stuff. Nv30 wasn't in there.

I'm not going to comment on the rest of your theory except to say it has a base that doesn't exist.

 

[jbond04]

Interesting theory, Uttar...

Do you work for nVIDIA? In another thread, you seemed to know quite a bit about how paid time off worked for nVIDIA...so perhaps you have some inside information that we don't. But this is probably just wishful thinking on my part.

If you could provide a link to where Jen Hsun Huang said that the NV30 was taped out in April, this post might get a lot more interesting. I don't ever remember hearing that, which kind of makes your whole theory suspect. You do make a couple of good points:

1. Anand said in his review of the GeForce 4 that nVIDIA could have released the NV25 core in December, but they didn't want to cut into the profits of the Geforce 3. This leads to a question...what did the engineers do for 9 months then? (From January 2002 to August 2002) That is an awful lot of time for nVIDIA to "waste", which might lend credence to your theory that a partially designed NV35 is actually the NV30 (or that the NV30 stole some of the NV35's features).

2. nVIDIA would not seem like a company to sit around idle while waiting for TSMC...

However, these are big, huge "what-if's" that I am making, so it's probably just my nVIDIA fanboy roots taking hold of me here. *Sniff* If I were a betting man (which I am), I would say that all of this is just wishful thinking on the part of nVIDIA fanboys, like myself, who are having a hard time coping with all of the ATi fans proclaiming to high heaven how nice their new R300 is (which it is).

Just some thoughts.

 

[AdvancedRobotics]

Originally posted by: jbond04
Interesting theory, Uttar...

Do you work for nVIDIA? In another thread, you seemed to know quite a bit about how paid time off worked for nVIDIA...so perhaps you have some inside information that we don't. But this is probably just wishful thinking on my part.

I was thinking he may work for nVidia when I read his post:

"I think you do not know exactly how vacations work at nVidia:

Paid Time Off: If you join NVIDIA, you'll accrue 15 "PTO" days during your first year of employment. That's 3 weeks with which you can do whatever suits your needs. For each additional year that you're with the Company, you will accrue an additional day, up to a maximum annual allotment of 20 days.

And, for those employees who want to save up for that long-awaited summer vacation or some other major event, NVIDIA allows employees to accrue up to 35 days of PTO. In addition, for those employees who stop accruing because they hit the maximum accrual, NVIDIA also allows employees to "sell" some of their PTO.


Uttar"

Funny how you'd know all of that.

"Some bugs already being fixed, it is highly likely the NV30 will only take the minimum 90 days and will be ready in end november."

I'm sure nVidia will give you it sooner for being an employee.

"You see, Jen Hsun Huang said back in april i think that the NV30 had taped out. And then he said in a recent conference call that it didn't tape out.
But he had NO advantage to lie back in april."

Can you show me some sort of link to where this was said, or was it said in some nVidia employee's only place?

"So what happened?"

I'm sure you'd know.

I think the release of the nv30 was more planning then anything else. There is no reason for the 9700 to come out this soon, no games will take advantage of it's graphics capabilites. Yes, UT2003 will work better, but save yourself some money and get the Ti4200 and then get the nv30 closer to the release of Doom3 when it will be more necessary.

 

[Orbius]

I think the release of the nv30 was more planning then anything else. There is no reason for the 9700 to come out this soon, no games will take advantage of it's graphics capabilites. Yes, UT2003 will work better, but save yourself some money and get the Ti4200 and then get the nv30 closer to the release of Doom3 when it will be more necessary.

Theres about a thousand reasons, i.e. running every PC game w/4xFSAA and full Ansio. The card rocks it totally leaves the 4600 in the dust, I feel sorry for the people who bought into Nvidia's hype over the Geforce 4 line a card that at its architectural base shares more with the Geforce 2 than with the Radeon 9700.
Thank goodness for ATi if it wasn't for them we'd be paying $400 for what amount to Geforce 2 updates until the end of time.

 

[Uttar]

Originally posted by: Asuka

Originally posted by: jbond04
Interesting theory, Uttar...

Do you work for nVIDIA? In another thread, you seemed to know quite a bit about how paid time off worked for nVIDIA...so perhaps you have some inside information that we don't. But this is probably just wishful thinking on my part.

Sorry, the vacation info is public. It's at http://www.nvidia.com/view.asp?PAGE=benefits

And as to give you a link to show you that the NV30 has taped out, well, i'm quite annoyed here - now that i did a google search, i might be wrong.

All the original having taped-out info came from:
http://www.3dgpu.com/yabb_se/index.php?board=2;action=display;threadid=525;start=0

But then, now, Wired archived that article at http://www.wired.com/wired/archive/10.07/Nvidia.html
And i don't find anything about tape-out in that...

So either that poster at 3dGPU was a liar, or Wired changed the article to supress the "it taped out" part because some other nVidia executive was angry at Huang for broking a NDA.

I still got that magazine somewhere in the house, should be able to find it tommorow.


Oh, and i found a very small thing to support the fact that nVidia changed the NV30 since april: In that Wired article, it says the NV30 has nearly 120 million transistors. But in a recent investor meeting ( which was +- explained shortly at the Motley Fool ) , the NV30 had 128 million transistors! That's nearly 10% more transistors, as i suspected. Now, that can't be all bug-fixing, can it? And if it was 128 million back then, Wired would have said nearly 130 million transistors.


Uttar

 

[Uttar]

Sorry, i forgot to retrieve a quote marker... Everything which isn't a jbond04 quote is said by me.

 

[Wolfsraider]

So either that poster at 3dGPU was a liar

no he may have just been reading someone elses post some where and added it to his theories as is done all the time without proof though it tends to lead to fud

 

[merlocka]

Here is an idea... complex GPU's are actually hard to make and nVidia's schedule slipped.

Does anyone really think that if nVidia had the design done in April they would not send out to fab until now? Got Weed?

 

[sxr7171]

Originally posted by: Orbius

I think the release of the nv30 was more planning then anything else. There is no reason for the 9700 to come out this soon, no games will take advantage of it's graphics capabilites. Yes, UT2003 will work better, but save yourself some money and get the Ti4200 and then get the nv30 closer to the release of Doom3 when it will be more necessary.

Theres about a thousand reasons, i.e. running every PC game w/4xFSAA and full Ansio. The card rocks it totally leaves the 4600 in the dust, I feel sorry for the people who bought into Nvidia's hype over the Geforce 4 line a card that at its architectural base shares more with the Geforce 2 than with the Radeon 9700.
Thank goodness for ATi if it wasn't for them we'd be paying $400 for what amount to Geforce 2 updates until the end of time.

I feel that it was M$ and DirectX 9 compatibility that made it necessary to drastically change the architecture of video cards this generation.