thilanliyan
Lifer
- Jun 21, 2005
- 11,912
- 2,130
- 126
I'm sure they already know what to expect from Fermi. .
But if the ship date to the AIBs was missed (from the link in the OP), how could they know what gaming performance is?
I'm sure they already know what to expect from Fermi. .
But if the ship date to the AIBs was missed (from the link in the OP), how could they know what gaming performance is?
But if the ship date to the AIBs was missed (from the link in the OP), how could they know what gaming performance is?
I'm sure they've had access to beta or even alpha hardware. Not to mention those rumors are highly unsubstantiated.
True...I forgot that even hardware can have alpha and beta versions.
Does anyone know if when they go for a new revision of the silicon, is it just for bug fixes, or can Nvidia be getting a new revision with the idea of looking for higher clock speeds?
My logic impressed, a hypnotic effect, a latent patent you can call it a gift.
My take on current Fermi A3 stepping is that since TSMC can't really yield cost-effective Fermi's at this time anyways there is time for Nvidia to continue to improve the design for gross margins (clockspeed, etc) purposes while TSMC gets their 40nm up to speed.
It's amazing how ATI having difficulty producing enough chips is equated to Nvidia being unable to release a chip.
Fermi is looking like Nvidia's 2900. And the 2900 was Ati's 5700. So Fermi is nvidia's 5700 !
My logic impressed, a hypnotic effect, a latent patent you can call it a gift.
The longer nVidia takes to release Fermi, the more it favors ATI.
Yes.
Just like with CPU's there are steppings to fix critical "fatal flaw" type bugs (TLB in B2 stepping Phenom) and then there are steppings which fix minor non-showstopper bugs/errata but are primarily designed to maximize gross margins...be it higher clockspeeds so higher ASPs or to eliminate speedpaths or power-consumption so parametric yield for existing SKU's can be increased.
My take on current Fermi A3 stepping is that since TSMC can't really yield cost-effective Fermi's at this time anyways there is time for Nvidia to continue to improve the design for gross margins (clockspeed, etc) purposes while TSMC gets their 40nm up to speed.
That's my thought as well. There was some info months ago that Fermi had taped out. I'm sure NVIDIA could have rushed a card to market, but they probably knew there would be major supply issues.
Clearly this is what ATI is going through. So why not improve the design and bring real product to market. If Fudzilla is to be believed...the GTX 260s are still selling well. I would imagine the other cards are probably also selling plenty.
http://www.fudzilla.com/content/view/16275/1/
So a lack of supply from the competition + current product selling well + no major DX11 titles = room to breath.
It seems NVIDIA was able to actually benefit from the 40nm debacle.
Say what you will but the later it takes Fermi to come out, the worse it will be for nVidia. As of right now, things aren't bad for nVidia, they may be late but supply issues by TSMC are ensuring the negative effects are minimal. This is very good news for nVidia because unlike when ATI missed their release dates (announced or not) on previous cards, nVidia was there to capitalize and take up a lot of sales that would otherwise have gone to ATI. ATI hasn't had that luxury due to TSMC's manufacturing issues.
The bad news for nVidia is that the longer it takes them to actually release the product and ramp up production, even if in a limited fashion, the more ATI stands to benefit. Now, these are only rumors right now but if nVidia does take until late March or even April you can bet that ATI will have tweaked refresh products rolling out soonish to better compete with nVidia's cards. I think nVidia could lose some of the projected performance advantages in that case.
With the fact that all news indicating TSMC should have all of their marbles together by the end of November means that from early to mid December onward we should see a much higher amount of Radeon 5 series cards on the market and we could see some sales start to swing ATI's way. Assuming that there is much higher availability by the middle of December and that nVidia can't release a card till late February at the earliest then ATI may have close to an effective two month head start. Assuming as the latest rumors indicate that a worse case scenario has nVidia not releasing a video card until March means that ATI would have a whole quarter of decent availability.
As well as nVidia has been performing, you know they have to be getting antsy when looking at the possibility of three months of decent to maybe high availability of Radeon 5 series cards while they still haven't released their next gen product. The way I look at it is that if by the end of December we haven't heard much from nVidia then it looks like they are having some serious issues with Fermi. nVidia can't sit idly by and watch ATI start pushing the Radeon 5 series cards without at least some "leaked" benchmarks or gathering OEM's and giving them an "inside look" that will usually be leaked to counter any positive press that ATI can generate.
If nVidia comes out too late, it could run into a respin of the 5870 that may close some of the projected performance gap. ATI isn't going to sit idly and not work on a respin until nVidia is out with Fermi after all. While nVidia isn't in any trouble and is still doing quite well with only the GT200 series, it needs Fermi to be released ASAP. The longer nVidia takes to release Fermi, the more it favors ATI.
That's my thought as well. There was some info months ago that Fermi had taped out. I'm sure NVIDIA could have rushed a card to market, but they probably knew there would be major supply issues.
Clearly this is what ATI is going through. So why not improve the design and bring real product to market. If Fudzilla is to be believed...the GTX 260s are still selling well. I would imagine the other cards are probably also selling plenty.
http://www.fudzilla.com/content/view/16275/1/
So a lack of supply from the competition + current product selling well + no major DX11 titles = room to breath.
It seems NVIDIA was able to actually benefit from the 40nm debacle.
Your points are beyond pathetic.
Is it the facts that bothered you? Your reply was not clear enough and lacking substance.
Unless nVidia is being productive in this idle time, improving and tweaking fermi while waiting for TSMC to catch up, making Fermi even better than expected, which could spell disaster for ATI, especially if NV will then be able to compete on price
Don't worry about the sales guys. Unless you're a stockholder, really you shouldn't have a care in the world about sales.
Given a $100 250 or $140 260 vs $170 5770 it's a pretty clear choice re: which card to get.
Why is that? ATi is immediately selling every card that comes off the production line. Seems to be a good position to be in. All they need now is for TSMC to ramp up production. If the Fermi delays hold true, ATi will have plenty of time to get their 5800 series out en masse.
And if it's proving this difficult for TSMC to get a 2 million transistor GPU out in large numbers on 40nm, can you imagine the issues they'll have with Nvidia's 3 million transistor design?
ATi's troubles will be Nvidia's troubles X2.
Is it the facts that bothered you? Your reply was not clear enough and lacking substance.