TSMC 40nm issues largely fixed. TSMC roadmap. nVidia mainstream part in Q2-Q3'2010

[akugami]

Problems with TSMC’s 40nm Largely Behind, But Will Be Completely Solved by mid-2010 – Nvidia.

Xbitlabs article quoting nVidia as saying the 40nm issues are pretty much fixed at TSMC.

TSMC Pledges to Double 40nm Output by End of the Year.

Xbitlabs article saying that TSMC is now held back by the lack of 40nm equipment rather than yield issues. Also some roadmap info that says TSMC is skipping 32nm and going to 28nm then 22nm. 22nm process should be online in 2nd half of 2012.

Lower-Cost Fermi Graphics Processors Due in the Middle of the Year – Nvidia.

And for the finale, an Xbitlabs article speculating on the "GF 104" to be out in the middle of the year (Q2 to Q3). This is to be a lower cost mainstream part. Take that as you will.

 

[cbn]

Lower-Cost Fermi Graphics Processors Due in the Middle of the Year – Nvidia.

And for the finale, an Xbitlabs article speculating on the "GF 104" to be out in the middle of the year (Q2 to Q3). This is to be a lower cost mainstream part. Take that as you will.

These might end up being super nice parts (especially if Nvidia continues to apply what process tech they have already learned on the big Fermi).

 

[JSt0rm]

These might end up being super nice parts (especially if Nvidia continues to apply what process tech they have already learned on the big Fermi).

If the tech is good like g80 they could get some mileage out of it.

 

[GodisanAtheist]

Looking forward to those mid-end parts.

 

[nosfe]

Read the article more carefully, it talks about lower-cost Quadro and Tesla cards, nothing about lower-cost GeForce cards

 

[GodisanAtheist]

God damn it.

 

[alyarb]

relax. there will be reduced shader/reduced bus variants just like every other generation.

 

[Lonyo]

relax. there will be reduced shader/reduced bus variants just like every other generation.

Oh yeah, those cut down GTX280's are SWEET AS.

 

[Piotrsama]

Lower-Cost Fermi Graphics Processors Due in the Middle of the Year – Nvidia.

And for the finale, an Xbitlabs article speculating on the "GF 104" to be out in the middle of the year (Q2 to Q3). This is to be a lower cost mainstream part. Take that as you will.

That would be great if Nvidia makes a new mainstream chip.

 

[alyarb]

Oh yeah, those cut down GTX280's are SWEET AS.

you mean GTX 260? The core-216 was a fine card...

 

[GodisanAtheist]

relax. there will be reduced shader/reduced bus variants just like every other generation.

-I'm not doubting that, I'm just wondering WHEN THE HELL THEY'RE GOING TO SHOW UP.

My golden rule for upgrading GPU: Twice the performance for ~$150-$200 dollars. So long as nothing shows up in that category I get to keep kicking my feet. With rumors of GTX 4xx series performance indicating that ATI isn't going to be driving down its prices as far as we'd all like them too, I'm out in the cold until maybe the Fermi respin or ATI 6xxx series launches.

Oh well the ole 4850 is dutifully plugging away.

 

[postmortemIA]

<troll> February 2010 is almost over, we should start talking about Q1 2011 availability </troll>

 

[akugami]

Read the article more carefully, it talks about lower-cost Quadro and Tesla cards, nothing about lower-cost GeForce cards

I read it again, it doesn't state the parts are Quadro/Tesla parts. Granted it doesn't explicitly state the parts are general consumer Geforce parts either. My assumption would be they are Geforce cards because of this quote by Mr. White "By the time we begin launching Fermi, which is in volume production today, and which will really become mainstream for us in Q2 – Q3 from a volume standpoint, the yield issues will be past us." I can't link any comments about a mainstream card to Tesla or Quadro which are more for professional use and not mainstream which I equate with gaming.

The "modular design" of Fermi also seems to be aiding nVidia in getting lower end parts out faster it seems because one of my criticisms of the GT200 parts were that mainstream and mobile parts were slower in coming. It seems that while mobile parts are still not here or even announced, at least there are hints of lower end Fermi parts coming out soon.

 

[dzoner]

http://www.xbitlabs.com/news/video/...ors_Due_in_the_Middle_of_the_Year_Nvidia.html

The pertinent bit was:

" ...we will (be) taping out and announcing a whole stack of Fermi products that will be very specific for computing at the very high-end of the stack and others will be targeted at desktops and personal computing,&#8221; added Mr. White."

WILL BE taping out ...

After the present batch of usable chips from the current run of fubared Fermi A3 wafers, there ain't gonna be any more chips, high end or mainstream, for a looooong time.

By the way, the last day of that promised (financial) Q2 - Q3 availability is Oct 31 ... Halloween.

Trick or Treat?

 

[Hyperlite]

http://www.xbitlabs.com/news/video/...ors_Due_in_the_Middle_of_the_Year_Nvidia.html

The pertinent bit was:

" ...we will (be) taping out and announcing a whole stack of Fermi products that will be very specific for computing at the very high-end of the stack and others will be targeted at desktops and personal computing,” added Mr. White."

WILL BE taping out ...

After the present batch of usable chips from the current run of fubared Fermi A3 wafers, there ain't gonna be any more chips, high end or mainstream, for a looooong time.

By the way, the last day of that promised (financial) Q2 - Q3 availability is Oct 31 ... Halloween.

Trick or Treat?

ouch. That doesn't really make sense though, and i think he either misspoke or that was misquoted. Fermi was "taped out" years ago.

 

[Dark4ng3l]

ouch. That doesn't really make sense though, and i think he either misspoke or that was misquoted. Fermi was "taped out" years ago.

Yeah but they are not talking about Fermi here but about the derivatives. Completely different silicon.

 

[akugami]

http://www.xbitlabs.com/news/video/...ors_Due_in_the_Middle_of_the_Year_Nvidia.html

The pertinent bit was:

" ...we will (be) taping out and announcing a whole stack of Fermi products that will be very specific for computing at the very high-end of the stack and others will be targeted at desktops and personal computing,” added Mr. White."

WILL BE taping out ...

After the present batch of usable chips from the current run of fubared Fermi A3 wafers, there ain't gonna be any more chips, high end or mainstream, for a looooong time.

By the way, the last day of that promised (financial) Q2 - Q3 availability is Oct 31 ... Halloween.

Trick or Treat?

I think that the quote you point to is way too ambiguous and could cover pretty much every single product from high end to low end. My conclusion from the section you quoted is that you can't really take anything from it. I'm still of the opinion that Mr. White isn't talking about Quadro or Tesla parts and was talking about a more mainstream priced Fermi variant for gaming.

If you still feel the way you do, we'll just have to agree to disagree.

 

[GodisanAtheist]

I read it again, it doesn't state the parts are Quadro/Tesla parts. Granted it doesn't explicitly state the parts are general consumer Geforce parts either. My assumption would be they are Geforce cards because of this quote by Mr. White "By the time we begin launching Fermi, which is in volume production today, and which will really become mainstream for us in Q2 – Q3 from a volume standpoint, the yield issues will be past us." I can't link any comments about a mainstream card to Tesla or Quadro which are more for professional use and not mainstream which I equate with gaming.

The "modular design" of Fermi also seems to be aiding nVidia in getting lower end parts out faster it seems because one of my criticisms of the GT200 parts were that mainstream and mobile parts were slower in coming. It seems that while mobile parts are still not here or even announced, at least there are hints of lower end Fermi parts coming out soon.

“Fermi will show up as the GeForce and Tesla first and it will definitely show up first in the highest-performance configuration. Sometime in the mid-part of the year we will see a lower-cost version of that come out both for Quadro and Tesla,,” said David White, chief financial officer of Nvidia, at Goldman Sachs Technology Conference.

-I think that's the part where you start going "and mainstream geforce... AND MAINSTREAM GEFORCE?!"

 

[dzoner]

I think that the quote you point to is way too ambiguous and could cover pretty much every single product from high end to low end. My conclusion from the section you quoted is that you can't really take anything from it. I'm still of the opinion that Mr. White isn't talking about Quadro or Tesla parts and was talking about a more mainstream priced Fermi variant for gaming.

If you still feel the way you do, we'll just have to agree to disagree.

It wasn't 'covering every single product from high end to low end', it was 'for computing at the very high-end of the stack and others will be targeted at desktops and personal computing', note the specific reference to 'very high end of the stack', indicating the redesigned 512 shader part has NOT been taped out yet. Tapeouts of derivatives would almost certainly await risk silicon/validation from the base layer redesign/tapeout before proceeding.

It was Nvidia's chief financial officer at a Goldman Sachs Technology Conference and he said what he said for a reason. It's called covering your ass with the investors, whom the company officers absolutely, positively, do not want to see take them to court, apply discovery, and lay bare an attempted cover-up/mis-representation of pertinent facts the investors are legally entitled to know about.

The final redesign/tapeout will necessarily await Nvidia completing their homework on the particular problems of Fermi's A3 chip on TSMC's 'fixed' fabrication process, which did NOT fix the yield problem. Homework that AMD successfully completed a year ago with the 4770.

Keeping in mind Anand's article on how Nvidia is substantially behind AMD's expertise in troubleshooting new processes, it's easily possible Nvidia might just now be starting the final tapeout of the redesigned 512 shader Fermi chip.

 

[edplayer]

ouch. That doesn't really make sense though, and i think he either misspoke or that was misquoted. Fermi was "taped out" years ago.

Years ago?

I read that Fermi was taped out summer 2009

 

[dzoner]

ouch. That doesn't really make sense though, and i think he either misspoke or that was misquoted. Fermi was "taped out" years ago.

Why doesn't it make sense ... what set of facts or reasoned suppositions bring you to that conclusion?

By the specific technical definition of the word tapeout, Fermi's tapeout was completed at the end of June, 2009 ... so what were you referring to when you said Fermi was "taped out" years ago ... assuming those quotation marks had actual significance and meant you weren't referring to the technical definition of tapeout?

 

[Lonyo]

The new rumour is that graphics card prices are going to increase because the production still isn't great and increasing RAM prices.

Although I can't remember where I ready the story, some major players were talking about 10~15&#37; price increases because even with increased TSMC production they still can't keep up with demand.

 

[bryanW1995]

the good news just keeps on comin'!!!

 

[edplayer]

the new rumor is that I am not buying a videocard till the next glut

Whether that be Christmas 2010 or sometime in 2011

 

[v8envy]

Wow, I never thought I'd be using my 8800GT for 3 years... Which, coincidentally, is also the warranty length.

I expect my first viable upgrade window some time between Black Friday 2010 and Feb 2011 unless more "good news" arrives. Luckily I have more games than time, so my backlog of older games, eve and league of legends will keep me content until then. Or I might need to downgrade my monitor back to 1680x1050.