Nvidia Kepler Yields Lower Than Expected –CEO. Fermi 2.0?

[Imouto]

So there's no link to what you read?

OFC

not

 

[omeds]

So there's no link to what you read?

I dont know where I read it tbh, I frequent a lot of forums and read a lot of articles and at the time I didnt think much of it.. You need to relax your rabid fanboy fangs and realise I am just reporting what I read, along with stating I dont know if theres any truth to it.. relax buddy.

 

[NIGELG]

I dont know where I read it tbh, I frequent a lot of forums and read a lot of articles and at the time I didnt think much of it.. You need to relax your rabid fanboy fangs and realise I am just reporting what I read, along with stating I dont know if theres any truth to it.. relax buddy.

Is that you, Demo?....lol.

 

[maddie]

I dont know where I read it tbh, I frequent a lot of forums and read a lot of articles and at the time I didnt think much of it.. You need to relax your rabid fanboy fangs and realise I am just reporting what I read, along with stating I dont know if theres any truth to it.. relax buddy.

Now that is funny.

Make some wild predictions, then claim you don't know or remember anything.

Any unique beliefs or thoughts why some here have a problem with your statements?

 

[omeds]

I didnt make any predictions. I said I read that yeilds were a few % below expected and that I dont know if theres any truth to that.

 

[railven]

Okay, so I really don't care about half the crap you guys argue about - since I'm just a consumer who plays games.

What I care about is - because so many opinions are stated as "facts" and I personally lost track of which rumor is the most recent, I just want to know - will mid-range Kepler be out sometime in March or is that also pushed back to April? By Mid-Range, I mean $200-350 price bracket, ie GTX 660.

I had to swap out a failing HDD in the GF's PC and saw her little GTX 460 SC trucking along. We're both in desperate need of upgrades haha. I might just cave and buy Sandy Bridge so both our systems get a long over due upgrade.

Thanks in advance.

EDIT: I'd like to say I'm also sadden by the lack of price cuts. GTX 570/HD 6970 prices don't seem to be budging.

 

[Magic Carpet]

When can I actually buy this gem?

 

[omeds]

Probably April, not sure about pricing, Nv may price it fairly high (in true Nv style) if it competes well with AMD 7 series.

 

[Imouto]

I didnt make any predictions. I said I read that yeilds were a few % below expected and that I dont know if theres any truth to that.

If you don't get what's wrong with that statement lemme say this:

I read somewhere yields are terrible and Nvidia is going bankrupt but I don't know if there's any truth about this.

 

[maddie]

Okay, so I really don't care about half the crap you guys argue about - since I'm just a consumer who plays games.

What I care about is - because so many opinions are stated as "facts" and I personally lost track of which rumor is the most recent, I just want to know - will mid-range Kepler be out sometime in March or is that also pushed back to April? By Mid-Range, I mean $200-350 price bracket, ie GTX 660.

I had to swap out a failing HDD in the GF's PC and saw her little GTX 460 SC trucking along. We're both in desperate need of upgrades haha. I might just cave and buy Sandy Bridge so both our systems get a long over due upgrade.

Thanks in advance.

EDIT: I'd like to say I'm also sadden by the lack of price cuts. GTX 570/HD 6970 prices don't seem to be budging.

When can I actually buy this gem?

Sadly NOBODY HERE KNOWS.

Any estimate you read about is at present, wishful thinking.

'omeds' being correct will be purely accidental with the available public knowledge.

 

[Keysplayr]

Okay, so I really don't care about half the crap you guys argue about - since I'm just a consumer who plays games.

What I care about is - because so many opinions are stated as "facts" and I personally lost track of which rumor is the most recent, I just want to know - will mid-range Kepler be out sometime in March or is that also pushed back to April? By Mid-Range, I mean $200-350 price bracket, ie GTX 660.

I couldn't agree more. Too many people are stating opinions and repeating rumors as if they are facts. Please use clarifiers such as "if" "perhaps" "I think" "maybe" "it's possible" and such to show you are just expressing your opinions and not facts. If you have facts, please provide sources for backup and we can all decide after.

You can't just come out and say Kepler will be arriving in mid-June for example. If you can't back it up, don't say it, or else use a clarifier to show it's only your opinion or feeling or prediction. etc. etc.

Thanks in advance. Getting weary of wading through all the FUD MUD.

 

[notty22]

http://www.pcper.com/reviews/Editorial/NVIDIA-Reports-Q4-and-FY2012-Earnings-Tegra-3-28nm-Kepler

Q1 will certainly not be a great start to the year, but once yields improve and more wafer starts become available for 28 nm then the company will start to see a good rise in margins and a lowering of costs. NVIDIA is obviously quite bullish on Kepler and what it will represent to the market. From all the hints running around, NVIDIA will be taking a slightly different road with the design and actual introduction of this part. NVIDIA did share that yields for Kepler on 28 nm is much better than the first Fermi parts coming off of 40 nm.
We can certainly look forward to another very interesting year in desktop, notebook and mobile graphics. Competition will be tough, but some very interesting products will be delivered. NVIDIA looks to remain healthy with a wide variety of products they are offering to the respective markets.

As soon as AMD fills the big hole still left in their gpu lineup, then Nvidia will be late in that area to.

 

[Keysplayr]

Sadly NOBODY HERE KNOWS.

Any estimate you read about is at present, wishful thinking.

'omeds' being correct will be purely accidental with the available public knowledge.

Heh, yeah, kind of like Charlie D. I agree with you.

 

[omeds]

If you don't get what's wrong with that statement lemme say this:

I read somewhere yields are terrible and Nvidia is going bankrupt but I don't know if there's any truth about this.

Well thats a little different by you adding the "going bankrupt" part.. If you said I read yeilds are terrible I could accept that, because they are not good apparently, all I said was I read (a rumor) they were a little less than expected. Sorry if you read too much into that..

Sadly NOBODY HERE KNOWS.

Any estimate you read about is at present, wishful thinking.

'omeds' being correct will be purely accidental with the available public knowledge.

April is what the rumor mill says, the guy was asking so thats the best answer I can give him.. "probably April". Its better than saying no idea, maybe Jan 2028.

 

[railven]

Sadly NOBODY HERE KNOWS.

Any estimate you read about is at present, wishful thinking.

'omeds' being correct will be purely accidental with the available public knowledge.

That really sucks. I'd personally prefer to get new generation (since again my buying cycles are now limited to every other year) but if the GTX 570/580 drops price before I get a solid idea of when to expect Kepler (GTX 660 again, or 670) I might just have to pull the trigger.

Oh well, back to waiting.

 

[Grooveriding]

So there's no link to what you read?

There is, I have the link right here.

http://en.wikipedia.org/wiki/Into_the_Land_of_the_Unicorns

 

[SirPauly]

What was let out at the conference call is now being cushioned (Damage Control).

What damage control? Curious if you can expand on this and see what you offer.

 

[omeds]

There is, I have the link right here.

http://en.wikipedia.org/wiki/Into_the_Land_of_the_Unicorns

Ok that was funny.



Stop it, this thread is serious buzinesss apparently

 

[SirPauly]

Hey, you said it brother! Nvidia doesn't have a damn thing!

That you know of.

 

[SirPauly]

Minor off topic rant for a second, and something that bugged me while using 580s. Nvidia still doesn't allow you to select SSAA within the driver, you have to use an external application. Before some NV guy attacks me, its pretty stupid that it is NOT IN the driver - and you have to use an external app like nvidia inspector. Before anyone starts getting all apologetic, this is something that should really be in the CP and i've posted on the nvidia forums in the past stating such. I don't understand the omission.

In my ideal world AMD would improve their profile program slightly within CCC, while NV would include SSAA and FXAA injection in their driver. (and not require a separate d/l to enable it)

It's in the driver but one has to download a tool from nVidia's site or use a third party tool.

 

[Grooveriding]

Would be nice if nvidia could also get their act together on SLI profiles and fix their drivers, they have been piss poor for months. Alan Wake - no SLI profile, game is unplayable for me on a single card. The Darkness II - no SLI profile as well.

Latest 295.51 nvidia beta drivers have a memory leak and also make BF3 performance go down significantly. So no SLI scaling in Kingdoms of Amalur unless you want to use broken memory leaking drivers and have poor perforance in BF3

No WHQL driver from nvidia since November 2011. The TDR bugs that people are experiencing with nvidia cards are still not fixed and that has been going on for almost six months.

Nvidia where r u stellarzzz driverzz suproortz ? I've found it's better to see if anyone has found solutions on other forums or in PC gaming forums here, because if you ask about these broke ass nvidia drivers in VC&G - the nvidia fanboys start frothing at the mouth and going ape****.

 

[Idontcare]

Yes. AMD was very conservative with clocks and power draw for SI. A re-spin combined with a more mature process has a lot of potential to say the least. People are getting excellent overclocks on an immature, poor yielding process, consider what will happen when things get sorted out at TSMC and AMD has had a chance to get to know the process better.

SI also scales very well with clock speed, something in the past was a weakness for AMD.

~210W, they have a good 90W left.

There are lots of ways to measure power usage, but in OCCT Anand has it at 94W less than the 580.

This is what Guru3D posted loaded while gaming. They have room to grow.

Excellent :thumbsup:

So this would certainly make for more evidence that TSMC's parametric yields are fine (enough) and that if anything is holding AMD (and thus Nvidia) back it would be the functional yields (again, as expected for a new node which itself is not "broken", merely needs reduction in defect-density).

That also shows how likely it is that AMD's 7970 release was strategic sand-bagging to wait until Kepler showed up before AMD shows their full hand. They can easily release a higher performing single-GPU SKU that goes above 7970 on the basis of a clockspeed bump alone.

All things considered, outside of warranty restrictions there seems to be very little reason not to overclock the 7970 on its default voltage. Even a conservative overclock of 1050MHz would add 13% to the core clocks (and as a result performance in virtually all GPU-limited scenarios), which is a big enough leap in performance to justify spending the time setting up and testing the overclock. By not raising the core voltage there’s effectively no power/noise tradeoff and this seems to be achievable by virtually every 7970, making this a freebie overclock the likes of which we’re more accustomed to seeing on high-end CPUs than we are flagship GPUs.

Anandtech Article

 

[Idontcare]

Nvidia where r u stellarzzz driverzz suproortz ? I've found it's better to see if anyone has found solutions on other forums or in PC gaming forums here, because if you ask about these broke ass nvidia drivers in VC&G - the nvidia fanboys start frothing at the mouth and going ape****.

It takes two to tango, for every "frothing a the mouth" nvidia fanboy here we also have someone who busies themselves posting inflammatory antagonistic hyperbole to ensure they make their own prophecies come true...

If you take a crap in your toilet, don't get mad at your toilet when your toilet smells like your own crap. :hmm:

 

[railven]

Excellent :thumbsup:

So this would certainly make for more evidence that TSMC's parametric yields are fine (enough) and that if anything is holding AMD (and thus Nvidia) back it would be the functional yields (again, as expected for a new node which itself is not "broken", merely needs reduction in defect-density).

That also shows how likely it is that AMD's 7970 release was strategic sand-bagging to wait until Kepler showed up before AMD shows their full hand. They can easily release a higher performing single-GPU SKU that goes above 7970 on the basis of a clockspeed bump alone.

I've personally never been a fan of OCing GPUs mostly because I can't wait and buy at launch which often comes with a not so good cooler.

But when even Ryan Smith was basically telling us to OC the damn things, well then, I give his opinion a lot of weight. Then the benches hit the scene and I was just floored.

Even if they improve it, and refresh is just the same card with the monster OC stock - I'd be happy buying now. The power draws aren't too much (when setting a reasonable OC), so I'm hoping a good cooler will cover my bases.

 

[Grooveriding]

It takes two to tango, for every "frothing a the mouth" nvidia fanboy here we also have someone who busies themselves posting inflammatory antagonistic hyperbole to ensure they make their own prophecies come true...

If you take a crap in your toilet, don't get mad at your toilet when your toilet smells like your own crap. :hmm:

Fortunately I only share my toilet with my wife and we both remember to flush. Lots of endorsed floaters around here.