Before I rebut your point, could you please do me a favor and point out what I meant by this post of mine, and it's context?
NVIDIA GF100 & Friends speculation
- Thread starter Arty
- Start date
Before I rebut your point, could you please do me a favor and point out what I meant by this post of mine, and it's context?
I see no difference to AMD. r600 --18 months --> rv770 -- 15 months --> rv870.
And nVidia said that the problem with GF100 was a design problem because of the new architecture. So maybe we will see the next Chip not 21 months after GF100...
Last edited by a moderator:
Bouncing Zabaglione Bros.
Legend
So because we've decided that R870 is evolutionary and Fermi is a bigger change, that somehow absolves Nvidia from the need to execute?
Fact is that having a trick new architecture is no use if you can't turn it into a product that you can ship, or one that can barely compete with the half year old competition.
If the real GF100 is going to be B1, then that's a year left open to AMD - yayy for new architectures! I'm sure AMD will be crying all the way to the bank while Nvidia cheers their way to the poorhouse!
Which doesn't mean they can't have problems. Same with NVIDIA. Their execution from the GeForce 6 series up to G92 was almost flawless. Still they had problems with GT200 and now the delays with Fermi.
There won't be a shrink before next year...I'm talking about Fermi 1 with its target clocks or beyond, against what you believe will be released this year (NI)
Like R600 vs R580 ?
That's what we all expect, but it may not pan out quite as well. If these Fermi numbers are true, then R600 repeats itself on NVIDIA's side, even if slightly better, assuming that the GTX 480 is faster than the HD 5870.
What does that have to do with what I said ? I barely even mentioned RV770. I only used it as a performance reference for RV870 in the text you quoted me...
Deja vu? I recall similiar proclamations at GT200/RV770 launch and it hasn't materialized to date.
That's because they apparently/probably have longer design cycles now than AMD, because they are making chips that are >50% larger in die size/trannies.
That's because there is a good chance that B1 will compete with NI longer than it competes with Cypress.
I don't know, rv 670 vs rv770 perhaps
RV770 was a new architecture.
And why do you assume that they'll always be following this strategy ? Fermi is a stepping stone for at least two generations of their products. Do you think that they will follow the same "big chip" strategy, if it doesn't work out for them ? G80 turned out to be very good, actually one of their best, if not the best. It was only with GT200 that they had "some" problems. If Fermi, in its current state isn't profitable, you can bet that they'll change their philosophy.
rpg.314 said:
Again you're assuming that NI will be out this year AND on time
rpg.314 said:I don't know, rv 670 vs rv770 perhaps
That doesn't represent a new architecture.
rpg.314 said:RV770 was a new architecture.
Since when ?
Things are different this time, if only somewhat. We'll see the financials soon enough.
I'll be quite happy if they do change their strategy to follow the small die/sweet spot one. Let's wait and watch.
Ask Jawed. I am sure he'll be pleased to give you all the gory details. :smile:Since when ?
If you look at all the changes it was a major rework/overhaul, if not a brand new/mint fresh one. For starters, you could begin by asking yourselves where did the 40% increase in raw flops/mm2 on the same process come from?
If that's in their best interest, they will.
Overhaul...yeah sure. It was trying to fix what was broken with R600/RV670, but it's the same basic architecture as R600/RV670.
Meh...
Bouncing Zabaglione Bros.
Legend
You think the people at the top of Nvidia have been happy with their financial performance over the last couple of years? Or their execution?
Another one of those causation / correlation canards...
For a chip that has logic cut-and-pasted multiple times over, there's little be gained by reducing die size.
A chip that completely revamped shader cores. Is first to have multiple (distributed?) triangles per clock front end, a R/W L2 cache, and ECC logic.
Yet somehow, it's the cut and paste operation that probably increased the design cycle...
Yeah, that must be it.
Hmm...maybe I didn't express myself very well
What I meant was that I believe it to be an overhaul as you described and not a new architecture.
What does that have to do with your earlier comment?
Bouncing Zabaglione Bros.
Legend
What did your earlier comment have to do with my previous comment? Or are you going to argue that "getting products out" is somehow not better than "not getting products out"?
What in the world are you talking about? You made a comment about AMD raking in cash while Nvidia goes to the poorhouse and I said people predicted the same thing at GT200/RV770 launch and it didn't materialize. What does Nvidia's execution or the happiness of their executives have to do with that?
You made a comment about AMD raking in cash while Nvidia goes to the poorhouse and I said people predicted the same thing at GT200/RV770 launch and it didn't materialize. What does Nvidia's execution or the happiness of their executives have to do with that?Bouncing Zabaglione Bros.
Legend
What in the world are you talking about?
You're shrugging off Nvidia's poor execution with "it didn't kill the company last time". I'm pointing out that the people who know, the people who matter at the top of Nvidia probably don't agree with you that it's not important, or that's it's something to be shrugged off. And things are worse in the market space this time around. Are you really going to say you think that Nvidia don't think their execution is important or contributes to their profits?
How about you say what you're talking about instead of just throwing non-sequiturs about?
DavidGraham
Veteran
Similar threads
- Deleted member 2197
- Architecture and Products
