Nvidia kills GTX285, GTX275, GTX260, abandons the mid and high end market

Page 6 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Creig

Diamond Member
Oct 9, 1999
5,171
13
81
Originally posted by: Kingbee13
He's saying that if intel and AMD put good GPU's on die with the CPU, there would no longer be a discrete market for GPU's, at least thats my interpretation.

For the very low end that are currently on-board graphics, perhaps. But not for the mid-range or enthusiast level cards.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Originally posted by: Compddd
So Charlie was right. He's been right or partially right a lot it seems lately.

What remains to be determined is whether Charlie is right for all the right reasons or if he is right for all the wrong reasons.

There is a difference, and to us laypeople that difference matters in terms of the credibility (or lack thereof) we place on Charlie's future yet-to-be-written articles.

Charlie got bumpgate right for all the right reasons. I think it is very telling that even Anand refers to "it" as bumpgate ;)

But it would appear that we can safely conclude now that Charlie got the Fermi yield story right for all the wrong reasons...relying on a mistranslated Japanese blog that was referring to the month of the year numerically for which Charlie mistook that number as being the number of A1 stepping samples Nvidia received from TSMC. But in this case nevertheless the yield from those four wafers of that hot lot were "low" for all the right reasons as we discussed in that particular AT thread at the time.

So the question we have before us in regards to GT200b production limits is did Charlie get it right for all the right reasons, or is he right for all the wrong reasons? We ask ourselves this question as the answer matters when it comes to whether we view future Charlie articles with any more or less credibility.

Is there a trend towards higher quality/reliability in semi-accurate articles or do we maintain the status quo as a community and continue to regard his articles more as elaborate blog-type stories which are drafted merely with the hope that something in them is true?
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
didn't he hit the nail on the head with 4xxx also? or was that fuad? all those inq alums kind of run together for me...

it does seem that charlie has gotten some better sources over the past 18 mos. the hard part is getting through the crap to find the nuggets of truth in most cases.
 

SlowSpyder

Lifer
Jan 12, 2005
17,305
1,001
126
So does this mean Nvidia would rather sell nothing above the GTS250 vs. sell at a loss to keep their name fresh with consumers in those price ranges? Given that they are getting rid of these parts, they must really be pouring their resources into getting Fermi out on time, as not having it out on time would probably cause huge problems with no GTX cards on the market any longer. I think they better get it out before AMD has their x2 part out, or they'll be hurting...
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Originally posted by: bryanW1995
didn't he hit the nail on the head with 4xxx also? or was that fuad? all those inq alums kind of run together for me...

I don't recall, wasn't paying much attention on 4xxx when it was breaking news but my opinion is that all the INQ jr's are growing up and becoming more reliable, from BSN to FUDzilla to Semi-accurate.

Originally posted by: bryanW1995
it does seem that charlie has gotten some better sources over the past 18 mos.

I agree, in particular some of the numerical specifics in this article - from 530mm^2 diesize to 9000 risk wafers to 60% yield threshold for full-wafer payout - identically matched my own then-confidential infos regarding the Fermi situation at NV and TSMC.

So at a minimum some of his sources are running in the same circles as some of mine, could even be the same sources for all I know, not that that means either of our sources are actually in the know and are telling us infos that are accurate.

But the numerical specifics that will become verifiable - die size - will be proof that the tip of the iceberg exists and from there we can allow ourselves to believe the iceberg itself existed. At least I am willing to allow that much.

Originally posted by: bryanW1995
the hard part is getting through the crap to find the nuggets of truth in most cases.

That's where it helps to know people who themselves might know whether there is anything rotten in Denmark when the crap is slung onto the wall.

Theo, Fudo, and Charlie aren't the only people who have friends and past co-workers spread all across the industry, but at the same time not everyone who does feels like blogging about it every other day either.

So seek out those whom you do have confidence in, the kind of people who might not be at liberty to provide you "proof" that the crap on the wall is just that but at least they can give you all the assurances you should need to believe it to be so. (guys like Anand and Kyle for example)
 

BladeVenom

Lifer
Jun 2, 2005
13,540
16
0
Charlie wasn't too far off. Just read what Anand said.

NVIDIA told me two things. One, that they have shared with some OEMs that they will no longer be making GT200b based products. That?s the GTX 260 all the way up to the GTX 285. The EOL (end of life) notices went out recently and they request that the OEMs submit their allocation requests asap otherwise they risk not getting any cards.
 

SlowSpyder

Lifer
Jan 12, 2005
17,305
1,001
126
Originally posted by: BladeVenom
Charlie wasn't too far off. Just read what Anand said.

NVIDIA told me two things. One, that they have shared with some OEMs that they will no longer be making GT200b based products. That?s the GTX 260 all the way up to the GTX 285. The EOL (end of life) notices went out recently and they request that the OEMs submit their allocation requests asap otherwise they risk not getting any cards.

I wonder how this will affect their partners. Obviously there are still boards out there, and looks like their partners can get one last order of some of these EOL parts. But if Fermi is late, or Nvidia's partners run out of parts earlier than Fermi arrives in volume, I can't see those companies being too happy trying to live off of sales of GTX250's and other lower end parts while XFX, Asus, etc. are happily selling high end parts.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Originally posted by: SlowSpyder
Originally posted by: BladeVenom
Charlie wasn't too far off. Just read what Anand said.

NVIDIA told me two things. One, that they have shared with some OEMs that they will no longer be making GT200b based products. That?s the GTX 260 all the way up to the GTX 285. The EOL (end of life) notices went out recently and they request that the OEMs submit their allocation requests asap otherwise they risk not getting any cards.

I wonder how this will affect their partners. Obviously there are still boards out there, and looks like their partners can get one last order of some of these EOL parts. But if Fermi is late, or Nvidia's partners run out of parts earlier than Fermi arrives in volume, I can't see those companies being too happy trying to live off of sales of GTX250's and other lower end parts while XFX, Asus, etc. are happily selling high end parts.

I imagine it can't be all that different from the ramifications in say the auto-industry when a large company like GM shutters a product lineup like Saturn.

The hurt is felt throughout the entire supplier eco-system for those who were overly reliant on that particular product lineup versus those competing products which stand to benefit from the reduction in competition. (suppliers to Asus, XFX, as you duly noted)

The thing is that with the semiconductor industry being so inherently volatile from a year-on-year revenue volume perspective for the past 30yrs I just can't imagine anyone actually having allowed themselves to be in a position where NV's decisions can cause significant material impact to their bottom line, including NV.

Everybody knows to count on nobody and nothing, you keep your fingers in as many pies at the same time as possible because one-trick ponies in this industry are pre-destined to fail within one four-year business cycle. Plenty of them do anyways (be one-trick ponies), to their own peril, but we are supposed to view that as capitalism at its finest.
 

SlowSpyder

Lifer
Jan 12, 2005
17,305
1,001
126
Originally posted by: Idontcare
Originally posted by: SlowSpyder
Originally posted by: BladeVenom
Charlie wasn't too far off. Just read what Anand said.

NVIDIA told me two things. One, that they have shared with some OEMs that they will no longer be making GT200b based products. That?s the GTX 260 all the way up to the GTX 285. The EOL (end of life) notices went out recently and they request that the OEMs submit their allocation requests asap otherwise they risk not getting any cards.

I wonder how this will affect their partners. Obviously there are still boards out there, and looks like their partners can get one last order of some of these EOL parts. But if Fermi is late, or Nvidia's partners run out of parts earlier than Fermi arrives in volume, I can't see those companies being too happy trying to live off of sales of GTX250's and other lower end parts while XFX, Asus, etc. are happily selling high end parts.

I imagine it can't be all that different from the ramifications in say the auto-industry when a large company like GM shutters a product lineup like Saturn.

The hurt is felt throughout the entire supplier eco-system for those who were overly reliant on that particular product lineup versus those competing products which stand to benefit from the reduction in competition. (suppliers to Asus, XFX, as you duly noted)

The thing is that with the semiconductor industry being so inherently volatile from a year-on-year revenue volume perspective for the past 30yrs I just can't imagine anyone actually having allowed themselves to be in a position where NV's decisions can cause significant material impact to their bottom line, including NV.

Everybody knows to count on nobody and nothing, you keep your fingers in as many pies at the same time as possible because one-trick ponies in this industry are pre-destined to fail within one four-year business cycle. Plenty of them do anyways (be one-trick ponies), to their own peril, but we are supposed to view that as capitalism at its finest.

I agree with what you said, and I think that actually agrees with where I was pointing to with my original post.

We love auto industry comparrisons here, don't we. :) If a company was to specialize in just Saturn after market, they're about to be hurting if they don't take on new parts providers. EVGA and BFG, without having the numbers, I'm willing to bet make the vast majority of their revenue and profits from sales of Nvidia based video cards. I know BFG also sells power supplies, EVGA has motherboards. But I'm betting grphics cards are their bread and butter. If Fermi is late or underwhelming, much like the company that specializes in Saturn parts, they'll have to take on a different supplier (AMD, maybe Intel) or change the way they do business... I don't see either of these companies that like to advertise theirsevles as 'highend' being content with GTS250 sales only. ;)
 

v8envy

Platinum Member
Sep 7, 2002
2,720
0
0
EOLing midrange and up parts 5 weeks before Black Friday could have NV handing ATI the best Christmas gift of all. The entire market and mindshare for 2009 and early 2010. We'll have confirmation of this if Fermi-based cards aren't in select reviewers hands in the next few weeks. With controlled leaks from multiple sources.

Awesome that they shared this with <b>some</b> OEMs. Which implies <b>some</b> OEMs are going to be in for quite a nasty surprise.
 

Genx87

Lifer
Apr 8, 2002
41,095
513
126
Originally posted by: v8envy
EOLing midrange and up parts 5 weeks before Black Friday could have NV handing ATI the best Christmas gift of all. The entire market and mindshare for 2009 and early 2010. We'll have confirmation of this if Fermi-based cards aren't in select reviewers hands in the next few weeks. With controlled leaks from multiple sources.

Awesome that they shared this with <b>some</b> OEMs. Which implies <b>some</b> OEMs are going to be in for quite a nasty surprise.

Nvidia is asking for last run allocation from the OEMs. Expect supply of G200 parts through Feb per the article. This tells me Nvidia is expecting Fermi to be able to replace those parts before Feb.
 

v8envy

Platinum Member
Sep 7, 2002
2,720
0
0
The exact quote is:

<quote>The second was that despite the EOL notices, end users should be able to purchase GeForce GTX 260, 275 and 285 cards all the way up through February of next year.</quote>

I can still purchase Radeon 9550 and FX5200 series cards today, so the expectation of having a few overpriced cards rotting on the shelves early next year doesn't imply a whole lot lot. It really doesn't sound like there will be enough chips for partners to continue production for very long. Submit allocation requests ASAP or risk not having any chips sounds fairly dire to me.

One of two things will happen. Fermi will appear earlier than expected (in time for Black Friday retail, which I don't see as possible -- we'd have seen a reference board by now), or NV is leaving OEMs high and dry at the midrange and up segment for this holiday season. I don't see a third alternative.
 

Genx87

Lifer
Apr 8, 2002
41,095
513
126
Originally posted by: v8envy
The exact quote is:

<quote>The second was that despite the EOL notices, end users should be able to purchase GeForce GTX 260, 275 and 285 cards all the way up through February of next year.</quote>

I can still purchase Radeon 9550 and FX5200 series cards today, so the expectation of having a few overpriced cards rotting on the shelves early next year doesn't imply a whole lot lot. It really doesn't sound like there will be enough chips for partners to continue production for very long. Submit allocation requests ASAP or risk not having any chips sounds fairly dire to me.

One of two things will happen. Fermi will appear earlier than expected (in time for Black Friday retail, which I don't see as possible -- we'd have seen a reference board by now), or NV is leaving OEMs high and dry at the midrange and up segment for this holiday season. I don't see a third alternative.

They are taking orders now. It usually takes a good 12-15 weeks from the initial production start before the GPU is sitting on a retail shelf. The article doesnt really say when the final runs will start. I suspect early November which would line up well with the Feb quote.

 

MegaWorks

Diamond Member
Jan 26, 2004
3,819
1
0
I'm going to put this in one sentence instead of writing two or three paragraphs :p.

So basically nVidia=Owned?

Question is AMD going to keep producing the 48xx series when it's finally confirm that the GT200b is not in production anymore?
 

dreddfunk

Senior member
Jun 30, 2005
358
0
0
We'll probably just have to wait and see what the availability/costs of these parts are over the upcoming months, and what the precise release frame for FERMI is.

The problem with the EOL, as I see it, isn't so much that it abandons the high end; it's that it creates a mid-range gap between FERMI and G92b parts once FERMI debuts.

That is, of course, unless NV pulls a top-to-bottom launch with FERMI. Or surprise launches FERMI before the end of the year and has mid-range derivatives out by February '10.

Unless one of those two things happens, you have G92b < [blank] < FERMI.
 

v8envy

Platinum Member
Sep 7, 2002
2,720
0
0
Originally posted by: Genx87

They are taking orders now. It usually takes a good 12-15 weeks from the initial production start before the GPU is sitting on a retail shelf. The article doesnt really say when the final runs will start. I suspect early November which would line up well with the Feb quote.

That long of a assembly line to retail shelf pipeline does seem to imply Fermi better be available to partners by late November or early December if we're to have any hope of retail Q1 2010 availability.

I'll definitely keep my eyes peeled for the announcement of chips shipping to partners.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Originally posted by: SlowSpyder
I agree with what you said, and I think that actually agrees with where I was pointing to with my original post.

Yeah, my post was really meant to be a +1 post to yours :thumbsup:, not an "on the other hand" type post.

No surprise really as we both seem to be "common denominator" reductionists ;).

Originally posted by: SlowSpyder
We love auto industry comparrisons here, don't we. :) If a company was to specialize in just Saturn after market, they're about to be hurting if they don't take on new parts providers. EVGA and BFG, without having the numbers, I'm willing to bet make the vast majority of their revenue and profits from sales of Nvidia based video cards. I know BFG also sells power supplies, EVGA has motherboards. But I'm betting grphics cards are their bread and butter. If Fermi is late or underwhelming, much like the company that specializes in Saturn parts, they'll have to take on a different supplier (AMD, maybe Intel) or change the way they do business... I don't see either of these companies that like to advertise theirsevles as 'highend' being content with GTS250 sales only. ;)

As an armchair economist I am digging the opportunity this is going to provide for us to scrutinize with hindsight the actual ramifications of one company having reigning authority in the high-end (presumably high ASP means high gross margins for AMD) versus one company abdicating the high ASP market segment for what looks to be nearly a full business quarter (albeit overlapping two fiscal quarters, mid-Q4 thru mid-Q1).

So come Q2 earning reports next year it will be nice to go back and scrutinize the Q4/09 and Q1/10 data for both NV and AMD's graphics divisions and see what effect, if any, the headstart on HD5870 gave AMD in gaining marketshare and gross margins (and can we be so hopeful as to suggest net profits maybe?) versus the deterioration in NV's margins and ASP's that are to be incurred by basically dialing down the volumes of GT200b based shipments to near-zero.

Could become a cool case study in the market economics relevance of the top-end GPU market.
 

bryanW1995

Lifer
May 22, 2007
11,144
32
91
Originally posted by: MegaWorks
I'm going to put this in one sentence instead of writing two or three paragraphs :p.

So basically nVidia=Owned?

Question is AMD going to keep producing the 48xx series when it's finally confirm that the GT200b is not in production anymore?

No, they appear to be eol'ing the 4xxx also. soon we will be very short of gtx 260/75/85 as well as 4xxx, with the only "midrange" cards left as 5770 at lower mid and 5850 at upper mid. the past few months have been a great price/perf time, but that appears to be ending now.

@idc; maybe nvidia should change the 8800gts 512, uh, I mean, 9800gtx, no, wait, 9800gtx+, oops, I mean, gts 250. whatever it is, they should change it to the gtXXX 259, they'll get plenty of sales with that kind of marketing genius!
 

v8envy

Platinum Member
Sep 7, 2002
2,720
0
0
Originally posted by: Idontcare

Could become a cool case study in the market economics relevance of the top-end GPU market.

It would, but as all purely academic exercises it wouldn't be very informative. What is the value of lost mindshare and prestige from such a move? How much extra $ will be spent on marketing to convince customers Nvidia is still a "leader from a strategic position in the rear"? What is the long term impact of leaving high profile board partners high and dry without a high ASP part for a quarter?

The last is particularly tricky. EVGA and BFG have spent a lot of money on their image as premier, high performance gaming hardware vendors. They also have investors who need to be convinced that a quarter of competing only in the nearly zero margin value segment is good for the company.

Anyway, the ancillary $ losses are clearly much smaller than losses from trying to fight a price war. But the effects might not be limited to a single quarter.

The timing of splitting the financial impact across two fiscal quarters was brilliant though.
 

BladeVenom

Lifer
Jun 2, 2005
13,540
16
0
Originally posted by: Idontcare

As an armchair economist I am digging the opportunity this is going to provide for us to scrutinize with hindsight the actual ramifications of one company having reigning authority in the high-end (presumably high ASP means high gross margins for AMD) versus one company abdicating the high ASP market segment for what looks to be nearly a full business quarter (albeit overlapping two fiscal quarters, mid-Q4 thru mid-Q1).

So come Q2 earning reports next year it will be nice to go back and scrutinize the Q4/09 and Q1/10 data for both NV and AMD's graphics divisions and see what effect, if any, the headstart on HD5870 gave AMD in gaining marketshare and gross margins (and can we be so hopeful as to suggest net profits maybe?) versus the deterioration in NV's margins and ASP's that are to be incurred by basically dialing down the volumes of GT200b based shipments to near-zero.

Could become a cool case study in the market economics relevance of the top-end GPU market.

Have you looked at what it cost ATI when they were late with their HD 2000 series?
 

v8envy

Platinum Member
Sep 7, 2002
2,720
0
0
Originally posted by: Vertibird
What about a die shrink of GT200b?

Wouldn't it still sell well in the midrange as a DX10 part?

You would think that. But I'm sure people far better informed than us did the analysis and concluced a DX10-only part would be nigh unsellable as a premium midrange product even in Q1 of 2010, never mind over long enough to pay for the effort. Also remember that we've got a 256 bit bus to the RAM which adds significant cost over and above the GPU.

I'd certainly consider a green team *DX11* competitor to the 5770 in the $110-130 range (which is where it will be Q1 of 2010). A DX10 one for $170? Not so much. And in the under $100 space you've already got a plethora of undifferentiated products, for which the G92 re-badges are good enough.
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
I'm having a really hard time understanding this- they are flushing the distribution channels prior to the launch of new parts, and this somehow equates out to them abandoning the graphics market.....?

Current estimates have it part way through Q1 '10 before current supplies are exhausted, and everyone is expecting widespread availability of their next gen part by then.

Maybe I hit my head and didn't realize it, but reducing inventory levels before being required to clearnace them is normally considered smart business. The only thing this is really saying to me is that nV is planning on more of a top to bottom launch approach then they have been using for the last several years, nothing more and certainly nothinig remotely approaching what a lot of people are trying to read into it.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
Originally posted by: Vertibird
What about a die shrink of GT200b?

Wouldn't it still sell well in the midrange as a DX10 part?

That was attempted, it was plan A for this fall, but it failed for the same reasons RV740 came and then dissappeared for a while before coming back again. Once the window of opportunity was missed, the effort to commercialize the 40nm shrink of GT200b was scrapped entirely.

If this wasn't already discussed elsewhere then what I meant to say was - yeah, good question, me wonders why they didn't do that?