• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Fermi actually in January?

akugami

Diamond Member
Feb 14, 2005
4,633
69
91
If this is true, kiss any early release advantages for ATI's Radeon 5xxx series goodbye. The Radeon 5xxx series best hope for a head start was always a very late Fermi launch due to the extremely limited number of GPU cores that TSMC is able to manufacture.

**Edited some wording.
 

x3sphere

Senior member
Jul 22, 2009
722
24
81
www.exophase.com
Fud was reporting a late November launch before, even when those reports regarding the yield issues cropped up. I wouldn't put much faith in what they say.
 

nitromullet

Diamond Member
Jan 7, 2004
9,031
36
91
Would be nice if it's true. ...even a paper launch. My step-up expires in Feb, and It would be nice to get my name on the list even if I couldn't get hold of a card right away.
 

error8

Diamond Member
Nov 28, 2007
3,204
0
76
If this is true, kiss any early release advantages for ATI's Radeon 5xxx series goodbye. The Radeon 5xxx series best hope for a head start was always a very late Fermi launch due to the extremely limited number of GPU cores that TSMC is able to manufacture.
What are you talking about? All the 58XX series ATi made, were sold like hot donuts. ATi already had the head start and there is still December on the way, when we know that GPUs sell like crazy and since competition is still lacking, guess who will sell most of the gpus!?

It would be great if this rumor comes out true, since I like to see price dropping so I can finally afford a DX11 card. :)
 

akugami

Diamond Member
Feb 14, 2005
4,633
69
91
What are you talking about? All the 58XX series ATi made, were sold like hot donuts. ATi already had the head start and there is still December on the way, when we know that GPUs sell like crazy and since competition is still lacking, guess who will sell most of the gpus!?

It would be great if this rumor comes out true, since I like to see price dropping so I can finally afford a DX11 card. :)
It's not how well they sold, it's how much in this case. ATI could have sold a lot more video cards but is being limited by TSMC's manufacturing issues. Thus the lead is not as great as one would think.

While ATI still has a lead in sales (cause nVidia hasn't sold a single Fermi yet) the fact that nVidia has the more popular name brand and how well nVidia cards have sold means that when Fermi does get released it is going to sell at a fast pace. ATI's early release advantage was always to satisfy the pent up enthusiast demand before Fermi could get released.

This is probably ATI's best hope at regaining some of the market share they lost during the blunders of the previous years. At least their best chance in the foreseeable future. If it evaporates because Fermi is not going to be as late as rumored, then ATI needs to hit a home run on their next architectural update.

We know that the 5xxx series is due for a tweak as is usually the case after a major overhaul or change. The 6xxx series needs to even out the performance playing field and not just from a bang for the buck standpoint. nVidia got caught off guard by the 5xxx series but don't expect them not to be ready to fight back.
 

dug777

Lifer
Oct 13, 2004
24,778
3
0
It's not how well they sold, it's how much in this case. ATI could have sold a lot more video cards but is being limited by TSMC's manufacturing issues. Thus the lead is not as great as one would think.

While ATI still has a lead in sales (cause nVidia hasn't sold a single Fermi yet) the fact that nVidia has the more popular name brand and how well nVidia cards have sold means that when Fermi does get released it is going to sell at a fast pace. ATI's early release advantage was always to satisfy the pent up enthusiast demand before Fermi could get released.

This is probably ATI's best hope at regaining some of the market share they lost during the blunders of the previous years. At least their best chance in the foreseeable future. If it evaporates because Fermi is not going to be as late as rumored, then ATI needs to hit a home run on their next architectural update.

We know that the 5xxx series is due for a tweak as is usually the case after a major overhaul or change. The 6xxx series needs to even out the performance playing field and not just from a bang for the buck standpoint. nVidia got caught off guard by the 5xxx series but don't expect them not to be ready to fight back.
I'm confused. As far as I was aware we don't know the prices or performance or availability volumes of Fermi at this stage, do we?

If we don't then you're rather jumping the gun in nailing shut the coffin on the 5XXX cards based on the latest in a series of rumours surrounding Fermi's launch dates, are you not?

For all we know, Fermi will have a horrible price/performance ratio (and IDC has said some interesting stuff in that regard on here)...
 
Last edited:

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
Fud was reporting a late November launch before, even when those reports regarding the yield issues cropped up. I wouldn't put much faith in what they say.
He's just reporting a "best case scenario, it will be here no earlier than this date" situation.

When he was reporting it would be here in Nov that was based on the possibility/expectation that A2 silicon stepping would turn out to be production worthy and TSMC's yields would not limit the gross margin prospects of a chip that size.

If A3 stepping works out to be production worthy AND yields improve at TSMC enough to make the economics of selling Fermi become viable THEN we have every reason to expect a January release.

It is the difference between having a justified opinion versus merely having an opinion. I could say "Fermi won't be here till June" but you might be inclined to ask me "based on what?"...it is the basis of my opinion that would determine whether my opinion is a justified one or merely an unjustified one.

Fuad is just airing his justified opinion, which you would expect to change over time to reflect the fact that the facts on the ground are changing over time. Nvidia assumed A2 stepping was good, that is why they went to the expense of taping it out and running it thru the fab. Until Nvidia knew A2 was not ready for production there is no way any one could have or would have known A2 was not production worthy.

The concern I have is with the folks who claim they knew since last summer that Fermi would not be here until Q1 2010...as those are just unjustified opinions masquerading as prognosticators by chance.

Had A0 silicon been production worthy and had TSMC not had yield issues crop back up then the timeline was setup for an October launch of Fermi.
 

Janooo

Golden Member
Aug 22, 2005
1,067
13
81
He's just reporting a "best case scenario, it will be here no earlier than this date" situation.

When he was reporting it would be here in Nov that was based on the possibility/expectation that A2 silicon stepping would turn out to be production worthy and TSMC's yields would not limit the gross margin prospects of a chip that size.

If A3 stepping works out to be production worthy AND yields improve at TSMC enough to make the economics of selling Fermi become viable THEN we have every reason to expect a January release.

It is the difference between having a justified opinion versus merely having an opinion. I could say "Fermi won't be here till June" but you might be inclined to ask me "based on what?"...it is the basis of my opinion that would determine whether my opinion is a justified one or merely an unjustified one.

Fuad is just airing his justified opinion, which you would expect to change over time to reflect the fact that the facts on the ground are changing over time. Nvidia assumed A2 stepping was good, that is why they went to the expense of taping it out and running it thru the fab. Until Nvidia knew A2 was not ready for production there is no way any one could have or would have known A2 was not production worthy.

The concern I have is with the folks who claim they knew since last summer that Fermi would not be here until Q1 2010...as those are just unjustified opinions masquerading as prognosticators by chance.

Had A0 silicon been production worthy and had TSMC not had yield issues crop back up then the timeline was setup for an October launch of Fermi.
When did 'A0' taped out? Maybe you meant A1?

It was clear at the end of August that Q1 was the best case scenario based on A2. Now another delay if A3 is required.
 

Wreckage

Banned
Jul 1, 2005
5,529
0
0
It makes me wonder if A2 could have been released but with TSMCs production problems they decided to go with A3 for better yields/clocks/whatever.

Sigh, either way...we will have to wait until next year to see just exactly what we are getting.
 

Dribble

Golden Member
Aug 9, 2005
1,923
456
136
I suspect it was a mixture of both. A2 wasn't clocking high as high as nvidia would like, and even if it had been TSMC couldn't produce it economically, hence it made sense to keep tweaking and produce an A3.
 

Seero

Golden Member
Nov 4, 2009
1,456
0
0
I'm confused. As far as I was aware we don't know the prices or performance or availability volumes of Fermi at this stage, do we?

If we don't then you're rather jumping the gun in nailing shut the coffin on the 5XXX cards based on the latest in a series of rumours surrounding Fermi's launch dates, are you not?

For all we know, Fermi will have a horrible price/performance ratio (and IDC has said some interesting stuff in that regard on here)...
No one know how well Fermi will perform. Rumors said it runs faster, yet not solid evidences shows this except the fact that it has more transistors.

You can jump the gun on buying a 5xxx now, if you can somehow get one. What if Fermi is better at competitive prices? Won't it mean there will be a price cut upon 5xxx its release? It is said that TSMC will fix its problem at December, meaning there will be a supply boost at the beginning of next year. Won't that change the pricing of 5xxx?

Only people with green glasses think 5xxx is bad. It is a very good release no matter how Fermi performs. If it is much faster, then it will be much more expensive. In terms of availability, 5xxx will be much better than Fermi simply because they are all ready in production. I believe the performance of Fermi will win, but the performance/dollar will lose.

Suppose the GT100 performs between 5870 and 5970, but with a $800-1k price tag, will people buy it?
 

Janooo

Golden Member
Aug 22, 2005
1,067
13
81
What about the A1 results precluded shipping product prior to Q1 2010?
Bad A1 -> A2 required, so:

- how long from A1 results to A2 tape out?
- how long from A2 tape out to getting A2 back
- how long from A2 evaluation to product launch?
 

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
Bad A1 -> A2 required, so:

- how long from A1 results to A2 tape out?
- how long from A2 tape out to getting A2 back
- how long from A2 evaluation to product launch?
And your conclusion is that the above sequence of events can't possibly be done in less than 5 months?
 

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
Yeah and it sounds like they knew ATI would have problems as well.
http://www.fudzilla.com/content/view/16547/1/
To be sure by having the same common foundry and by both attempting to get product out on the same leading-edge node Nvidia knew exactly how TSMC's yield and capacity situation would effect/limit AMD's ability to do much with Evergreen in Q4.

Its one of the disadvantages of being with the same foundry as your competition...it eliminates one more technology gap by which your competition could potentially differentiate their products relative to yours but at the same time it also means you won't have the possibility of doing the same either.
 

akugami

Diamond Member
Feb 14, 2005
4,633
69
91
I'm confused. As far as I was aware we don't know the prices or performance or availability volumes of Fermi at this stage, do we?

If we don't then you're rather jumping the gun in nailing shut the coffin on the 5XXX cards based on the latest in a series of rumours surrounding Fermi's launch dates, are you not?

For all we know, Fermi will have a horrible price/performance ratio (and IDC has said some interesting stuff in that regard on here)...
While volumes may be small at first they will get here and one has to assume that TSMC is working frantically to solve the yield issues as well as having those extra machines online as mentioned in a HardOCP article.

Cause it's not always about the performance. We've seen in the past where people stay with a known brand simply because they are used to using said brand or products from the same company. Heck, why were nVidia's cards such great sellers over the last two quarters when the bottom end is merely "recycled" video cards and the top end did not provide as much bang for the buck compared to ATI's lineup? That is what I'm basing it on.

The GT200 series were very good pure performers in benchmarks but when you look at it from a value standpoint, the ATI cards were equal if not better in many cases. And while sales were much better for ATI since the Radeon 4xxx series came out, they didn't show a substantial boost in sales.

Say what you will but nVidia has good developer relations and a top notch product as well as better marketing. I don't see that changing with Fermi. nVidia may not bring the best value but it is the number one video card company and have more consistently brought winners to the table from a performance standpoint. Their reward has been great brand loyalty even though ATI may be a better purchase at times.

When dealing with marketing and sales, it's not always about the better product or even the product that is of the best value. For example I thought Tickle Me Elmo was a stupid toy but we all remember a few years back how well it sold and for what prices.
 

T2k

Golden Member
Feb 24, 2004
1,664
4
0
Fud was reporting a late November launch before, even when those reports regarding the yield issues cropped up. I wouldn't put much faith in what they say.
Yeah, Fuzilla is an openly pro-NV rumor site with horrible track record, there's not much credibility. I stopped reading long time ago.
At least Charlie's anti-NV/pro-ATI FUD is usually right on the spot.
 

Janooo

Golden Member
Aug 22, 2005
1,067
13
81
And your conclusion is that the above sequence of events can't possibly be done in less than 5 months?
Well, it's actually 4 months and try to assign dates to that and you'll see it is not realistic.
Do you agree?
 

ASK THE COMMUNITY