HYPOTHETICAL: R420 two months earlier than nV40?

[Joe DeFuria]

So then (whistling innocently), what do you offer as proof?

 

[Tim Murray]

So then (whistling innocently), what do you offer as proof?

That David Kirk has been travelling and giving demonstrations of the NV40 to selected developers (and I think a few members of the press)? I doubt that such presentations would be possible if there was no NV40 silicon; I can't imagine NVIDIA telling developers how cool and fast NV40 will be after they did exactly that with NV30, which everyone remembers.

 

[Joe DeFuria]

So then (whistling innocently), what do you offer as proof?

That David Kirk has been travelling giving demonstrations of the NV40 to selected developers (and I think a few members of the press)? I doubt that such presentations would be possible if there was no NV40 silicon; I can't imagine NVIDIA telling developers how cool and fast NV40 will be after they did exactly that with NV30, which everyone remembers.

And how does it follow from this that it would be the "16 pipe" NV40 that he showed, and not the 6x2 thingie or whatever the "original" NV40 was?

 

[PaulS]

So then (whistling innocently), what do you offer as proof?

That David Kirk has been travelling giving demonstrations of the NV40 to selected developers (and I think a few members of the press)? I doubt that such presentations would be possible if there was no NV40 silicon; I can't imagine NVIDIA telling developers how cool and fast NV40 will be after they did exactly that with NV30, which everyone remembers.

And how does it follow from this that it would be the "16 pipe" NV40 that he showed, and not the 6x2 thingie or whatever the "original" NV40 was?

Because there have been presentations since the 16 pipe news broke.

Worth noting that the bench numbers Baron mentions could be nothing more than guesstimates, but similarly they could be based on actual figures they've gotten. Really no way of knowing at this point. Certainly not unusual for IHVs to let developers know of the performance ballparks.

 

[Reverend]

That David Kirk has been travelling and giving demonstrations of the NV40 to selected developers (and I think a few members of the press)?

He actually performs these tasks? Gosh. I woulda thought there's a solid DevRel dept at NVIDIA...

 

[digitalwanderer]

So then (whistling innocently), what do you offer as proof?

That David Kirk has been travelling and giving demonstrations of the NV40 to selected developers (and I think a few members of the press)? I doubt that such presentations would be possible if there was no NV40 silicon; I can't imagine NVIDIA telling developers how cool and fast NV40 will be after they did exactly that with NV30, which everyone remembers.

So are you SAYING that David Kirk has been going around demonstrating the nV40, and that it is the ACTUAL nV40 and not the earlier-and-now-abandoned-nV40 of a few weeks ago? :|

 

[PaulS]

So are you SAYING that David Kirk has been going around demonstrating the nV40, and that it is the ACTUAL nV40 and not the earlier-and-now-abandoned-nV40 of a few weeks ago? :|

Definitely been presentations to select groups, silicon or no silicon. Quite close to home, you could speculate 8)

 

[digitalwanderer]

So are you SAYING that David Kirk has been going around demonstrating the nV40, and that it is the ACTUAL nV40 and not the earlier-and-now-abandoned-nV40 of a few weeks ago? :|

Definitely been presentations to select groups, silicon or no silicon. Quite close to home, you could speculate 8)

I question the value of any presentation without silicon, 'specially from nVidia...and if they AIN'T got working silicon yet than I think July might just be a conservative estimate on when cards will be available.

 

[Joe DeFuria]

Because there have been presentations since the 16 pipe news broke.

That doesn't mean anything.

And "presentations" happen all the time without working silicon.

Worth noting that the bench numbers Baron mentions could be nothing more than guesstimates, but similarly they could be based on actual figures they've gotten. Really no way of knowing at this point. Certainly not unusual for IHVs to let developers know of the performance ballparks.

Agreed.

All I'm saying is that someone claimed pretty definitively that the "16 pipe NV40" is in working silicon. Nothing here would back that assertion up. (Again, this doesn't mean there isn't working silicon...can't tell. But I wanted to know why the assertion was being forwarded.)

 

[ben6]

I'm not going to go into this battle again (got burned by NV30 launch). I will say that The Inq's rumors of Cebit launch for either part was NEVER the plan as of January.

I will also say with some certainty that NVIDIA's next generation part and ATI's next generation part are at least highly looking like near simultaneous release to retail, maybe off by a week or two, but certainly not more than a month.

 

[2senile]

But what this is really shaping up to be is that everything else out there is going to be more of a DX8.5 part (Including the R300), and the NV30 is going to be the only true DX9 part..

Ben6, hey! You didn't post that so why worry.

 

[digitalwanderer]

But what this is really shaping up to be is that everything else out there is going to be more of a DX8.5 part (Including the R300), and the NV30 is going to be the only true DX9 part..

Ben6, hey! You didn't post that so why worry.

Now that's just plain mean...albeit it in a mean/funny-as-hell kind of way.

I will also say with some certainty that NVIDIA's next generation part and ATI's next generation part are at least highly looking like near simultaneous release to retail, maybe off by a week or two, but certainly not more than a month.

Based on information from a month ago maybe, but right now I'm still betting it's late July if they make July at all...and I kind of doubt it with nVidia's track record.

Why do you think they'll be close? The only way I see 'em being close is if ATi really holds back a while.

 

[ben6]

Huh? Where did that quote come from. I've never stated anything but good things on the entire R3xx series

 

[ben6]

My information didn't come from NV or ATI directly but from a third party source.

I don't have any inside information on NV or ATI except informal discussions , I will say that unless I totally misunderstood the timeframes spoken of at IDF by PR of one of the companies, the launches/hitting retail will be very close given what information I have today

 

[2senile]

Huh? Where did that quote come from. I've never stated anything but good things on the entire R3xx series

It's not yours.

There was so much disinformation about the nV30 I actually lost my cool & insulted the verasity of a well respected member of B3D. I apologised immediately but the reason I posted that quote was that it seems to me that too much credibility is being given to rumour & speculation. As at the time that statement I quoted was made.

Over six mnths ago I have seen statements from ATi employees saying the tran' count was in the HUNDREDS of millions but now we have people saying it is less than 200Mil'.

On the nV side, during the same time period, I saw a statement where the nV40 would be a more traditional design yet we have speculation about it being 16x0 etc.

With the way websites were used in the past I think I'd ignore everything until the cards are out there & tested.

 

[Arun]

For the last darn time since I seen at least 2 people making the mistake here...
16x1/32x0.
Something I'm considering is that NVIDIA might be more aggressive in their usage of the "combining pipelines is fun" technology... 8x2,4x4, ... - although that'd only be useful for very big programs; I guess it wouldn't be worth it on smaller ones, where the lack of internal parallelism wouldn't compensate the lower register usage penalties.
But then again, it's also extremely likely NVIDIA isn't even doing any of that...


Uttar

 

[THe_KELRaTH]

Maybe NV are trying to be different - 64x-1

Seems with the rumour of the NV40 now being cancelled ATI are cancelling the R420 as it was never meant to compete against the NV45.
The R500 would have been launched instead but NV have just caught wind of this and cancelled the NV45 as it's not designed as a competing product against the R500. The now announced NV50 would have been launched but ATI............... ...

Of course ATI should have just renamed the R420 to R-42 as that's bound to be a sure winner - after all it's the answer to thee question

 

[2senile]

I can't resist!

ATi & nV have both canceled their planned chip.

The ATi 9999 wil be available in April. My sources tell me.

The nV9999.7 will be available in June. My sources tell me.

 

[DoS]

Because there have been presentations since the 16 pipe news broke.

That doesn't mean anything.

And "presentations" happen all the time without working silicon.

All I'm saying is that someone claimed pretty definitively that the "16 pipe NV40" is in working silicon. Nothing here would back that assertion up. (Again, this doesn't mean there isn't working silicon...can't tell. But I wanted to know why the assertion was being forwarded.)

Going the "i ll believe it when i see it" route is fair enough. However, IMHO NVIDIA took the decision to can the old NV40 and proceed with NV45 earlier than most ppl in this forum think. The reason i am making this assumption is that we heard some time ago (by multiple sources) that the original NV40 had tapped-out by late 2003...it's almost 4 motnhs since then and as someone else pointed out in this thread the CEBIT launch was never the plan as of January. What i am saying is that all indications are that NVIDIA took the decision to proceed with NV45 earlier than most here think, and if that was late December last year then a May launch makes sense. I strongly expect a May launch for both products

my 2 eurocents

 

[PatrickL]

Hum, i have not really the technical knowledge but from my readings here since late 2002 and the whole nv 30-nv 35 story, you just can't decide to skip a step and keep same time frame for release.