Fermi GTX480 is broken and unfixable

[spigzone]

Funny excerpt from Charlie's article that might explain the origin of his 'pipeline' into Nvidia.

Charlie - "Nvidia on the other hand did not do their homework at all. In its usual 'bull in a china shop' way, SemiAccurate was told several times that the officially blessed Nvidia solution to the problem was engineering by screaming at people. Needless to say, while cathartic, it does not change chip design or the laws of physics. It doesn't make you friends either."

 

[vengence]

BTW, Nvidia has already come out with a response to this fine piece of journalism. Take it for what it's worth (both Charlie and nvidia, for that matter…):
http://twitter.com/Igor_Stanek/status/9243031134
http://twitter.com/Igor_Stanek/status/9242904593

From a PR man. However with conference call tonight I'd lean a little more credibility than I usally do to PR men.

He isn't always off base. I think he tends to take some truth and then sensationalizes it to death. His previous G92 arguments were proof of that. He claimed they were all defective. That was horseshit as only some of them were. When the article came out about Fermi production yields he was close, but exaggerated the percentages slightly. Actually he nearly cut them in half or something like that as I recall.

The problem with Charlie is it is hard to tell where the truth ends and the sensationalism begins. This is why I dislike his work, though I do understand what he is doing and why.

I'm still waiting for my G92s to die, just like I said I would when his article came out. It's a shame cause then maybe I'd get lucky and they would send me a pair of 275s .

I must admit that I'm very surprised at the level of denial that still exists. If the SA article is wrong, be sure to point it out with your (presumably new and informed) facts, not character assassination.

Of course, one could just ask Nvidia for their side of the situation. Kyle, any chance of asking, just for the record/fun?

Probably won't get much out of Kyle. NDAs really are serious buisness.

http://www.hardforum.com/showthread.php?t=1491199Graphics AIB shakeout coming?

I doubt it. This was rumored by Charlie a LONG time ago. Besides, EVGA is doing really well with motherboards like the EVGA classified at the moment. I don't think the picture is all that sour for them.

 

[spigzone]

Problem with this is time - if they decided in January to revive GT212 it would still take at least 6 months to release anything. Its IMO more likely nVidia would do complete respin of Fermi.

As you said, AT LEAST six months. Probably looking at Q4 2010/Q1 2011 - realistic.

By which time AMD willl have completed their promised 2H 2010 complete 5000 series refreshes and be readying Northern Islands at 28nm ... and it's Llano Fusion cpu at 32nm.

Don't see much light at the end of the tunnel for Nvidia there.

EVGA and BFG are no doubt thinking the same.

 

[Rebel44]

As you said, AT LEAST six months. Probably looking at Q4 2010/Q1 2011 - realistic.

By which time AMD willl have completed their promised 2H 2010 complete 5000 series refreshes and be readying Northern Islands at 28nm ... and it's Llano Fusion cpu at 32nm.

Don't see much light at the end of the tunnel for Nvidia there.

EVGA and BFG are no doubt thinking the same.

Maybe they see light at the end of the tunnel, but it can just be approaching train

 

[vengence]

This is from teh CFO notes released ahead of the conference call today. You can find these under investor relations at nvidia's site or here is a direct link


Under the tesla/compute section

The computing segment of this business was down slightly quatert over quarter as customers anticipated the availability of our newest GPU computing architecure, called Fermi, which is currently in volume production and will commence shipping later this quarter.

Under the GPU section

While the availability of 40nm products remained in tight supply throughout the quarter, 40nm process yield continued to improve. As a result, 40nm revenues were up 178 percent over the prior quarter.

 

[trinibwoy]

He also mentioned tight supply of Fermi during Q2 since the limited supply will be spread across Geforce, Quadro and Tesla. He didn't sound particularly excited about Geforce - seemed to focus more on growth in Tesla and Quadro lines. But of course that doesn't mean anything until we see the goods.

 

[Rebel44]

While the availability of 40nm products remained in tight supply throughout the quarter, 40nm process yield continued to improve. As a result, 40nm revenues were up 178 percent over the prior quarter.

Thats nice, but how much were their 40nm revenues last Q? I am sure it wasnt much because they only had those low end OEM cards (and not even those were available for whole Q) + those were very cheap low end GPUs with minimal profit for NV.

Also release of just highend GPUs isnt enough - most sales (and revenue) are in 100$-200$ territory.......

 

[Zachstar]

Conference calls are about one central theme... Money... You can spice it up all you like but generally the people listening in have about 1000 other things to do with stock and other things that day and could really care less if Fermi can do Fold@Home like a beast or it has DX11 all they want to know is will it sell....

Fact is we just wont really know until march. I dont think Nvidia could effectively hide such a failure if it exists and the PR guys are just laughing at Charlies face. And frankly I would like to see him badly discredited with a full fermi launch because when competition goes down prices skyrocket.

 

[spigzone]

I doubt it. This was rumored by Charlie a LONG time ago. Besides, EVGA is doing really well with motherboards like the EVGA classified at the moment. I don't think the picture is all that sour for them.

They're not going to continue doing well with their motherboards if there aren't any competitive Fermi boards incoming ... or maybe no Fermi boards available at all after the initial 5-10 thousand run ... their motherboard business is going to fall off a cliff.

 

[Mr. Pedantic]

http://en.wikipedia.org/wiki/Lent#Associated_customs

 

[Valset]

As you said, AT LEAST six months. Probably looking at Q4 2010/Q1 2011 - realistic.

By which time AMD willl have completed their promised 2H 2010 complete 5000 series refreshes and be readying Northern Islands at 28nm ... and it's Llano Fusion cpu at 32nm.

Don't see much light at the end of the tunnel for Nvidia there.

EVGA and BFG are no doubt thinking the same.

at 28nm the GT200 looks better though. to say there is no light at the end of the tunnel is overstating it a bit. and both BFG and EVGA have other business then graphics cards)

 

[Dan_D]

I'm still waiting for my G92s to die, just like I said I would when his article came out. It's a shame cause then maybe I'd get lucky and they would send me a pair of 275s .

Mine have yet to die either.

Probably won't get much out of Kyle. NDAs really are serious buisness.

Indeed they are.

I doubt it. This was rumored by Charlie a LONG time ago. Besides, EVGA is doing really well with motherboards like the EVGA classified at the moment. I don't think the picture is all that sour for them.

Well EVGA just re-brands reference designs for their video cards. If they aren't selling, they simply order less of them. Their motherboards on the other hand require R&D. Therefore that business unit has to cover its R&D costs at a minimum or they lose money.

 

[DeadSkull]

Yawn. Even if Fermi is a total POS nVidia can absorb the hit from the pissed of gamers and still make tons of money in the work server market which will pay boatloads for these cards. ATI is the one who cannot afford to make any mistake here so nVidia is not finished, not anytime soon.

We could end up in some kind of guerrilla warfare deal where all cards stock values are approximately the same power and it's just a case of what can be overvolted and overclocked out of the card, that's basically what the 5970 is already, it ships with 725/1000 clock speeds but with voltage bumps can hit the same kind of speeds as an overclocked 5870 1000/1300 and above, it was more or less marketed as an overclocking card.

Actually that might be what blew up my Antec 850W last weekend...ahem...yeah ¬_¬

Which one, Signature or TPQ?

 

[Dan_D]

They're not going to continue doing well with their motherboards if there aren't any competitive Fermi boards incoming ... or maybe no Fermi boards available at all after the initial 5-10 thousand run ... their motherboard business is going to fall off a cliff.

What makes you think that? You can use AMD cards on any EVGA motherboard. The Intel chipsets support Crossfire and CrossfireX. The need for motherboards will always remain regardless of what is happening in the GPU world.

 

[heatlesssun]

Accurate or no, you KNOW this guy won't say anything good about nVidia. Let's just wait and see. If he's right he's right, but gee he's been saying EVERY nVidia cards is horrible for years now. MY 2 year old GTX 280s are still kicking most games in the ass two years latter. So much for his 280 predictions.

 

[Zachstar]

at 28nm the GT200 looks better though. to say there is no light at the end of the tunnel is overstating it a bit. and both BFG and EVGA have other business then graphics cards)

28nm is supposed to let them scale down to main and entry levels. Having to do a full launch at an even smaller node could lead to an even worse situation for them...

I really am starting to think tho that Charlie wants clicks and in my opinion he is likely already in the bosses office with them wondering just what the heck does he think he is doing...

Now do I expect things to be flowers? Heck no. I expect another 4770 fiasco with no cards to buy for months as people with thousands of dollars set up preorder out the wazoo.

But the more I read the article and see the response the more it seems this is just crap.

 

[JayteeBates]

Seem's to me NV is coming up to the same problem intel had with the P4 in a slightly different way. While intel came up against a thermal ceiling NV is coming up to a complexity limit for MFG methods in a mass market product. That can go good or bad for them - we all know the shift from faster faster faster at intel turned out fantastic for intel and consumers. Hopefully NV will have a bit of a redesign and come out with something that smashes the current options.

 

[tannat]

To be harsh one could argue that it's TSCM fault if this is true. They seem to have screwed up the 40 nm node quite much. According to more sources than Charlie Ati made an excellent job with quality control anticipating, tracking, and preventing the TSCM shortcomings while Nvidia might have gambled or just did not have the skill to adapt.

 

[Valset]

Accurate or no, you KNOW this guy won't say anything good about nVidia. Let's just wait and see. If he's right he's right, but gee he's been saying EVERY nVidia cards is horrible for years now. MY 2 year old GTX 280s are still kicking most games in the ass two years latter. So much for his 280 predictions.

well yes he did, when he absolutely had to that is, he like Nvidia's new power saving mobile solution.

 

[Valset]

Mine have yet to die either.

obvously you haven't waited long enough, you know, for the right amount of heat cycles on a desktop

that was a serously bad spin, yeah all the chips are "defective" in the sense that they would have a high failure rate in a notebook but come on, the desktops have just not shown the issue, at least in a time frame that anyone cares about

 

[spigzone]

at 28nm the GT200 looks better though. to say there is no light at the end of the tunnel is overstating it a bit. and both BFG and EVGA have other business then graphics cards)

Nvidia couldn't make 40nm 260/280 refreshes manufactuarable. Plus they would have to re-engineer the chip to add DX11.

But realistically, JHH is going to continue to roll the Fermi dice (Fermi II), hoping for a miracle (maybe screaming really, really REALLY loud ....).

But whether Fermi II is a base layer respin (plus a metal layer respin or two on top of that) at 40nm (8 to 12 mo+) or a stab at trying to move it directly to 28nm (14 mo+++), or even a totally reworked design at 28nm, they are still going to be well behind AMD when it comes out, who will, within two years, have advanced Bulldozer/Northern Island Fusion chips on the way PLUS whatever they have planned after NI in the pipeline for their GPU boards.

 

[BronxBartoni]

Great news to hear on the day my nvidia branded case got delivered............ Im a fanboy i know i know

 

[sirmonkey1985]

/facepalm

Right or wrong even you amd fan boys know Charlie always paints nvidia in the worst possible light. Does anyone really think this guy doesn't have a {H} on for nvidia? I mean are you serious? How can anyone really give him credit even when he happens to be right knowing how one sided he is against nvidia. Come on, its retarded.

i agree hes an amd/ati fanboy.. but even with that said the information has been pretty accurate at this point.. and it does fuel the fire.. i mean honestly if this information wasnt accurate dont you think nvidia would of come out with something to shoot down what he has been saying? not once has nvidia ever released a single bit of information that has proven Charlie wrong.. im not trying to stand up for the guy and his views but as far as the important information and numbers have gone hes been pretty accurate about them.. until nvidia says something or does something all we are left with is hear say info..

i wouldnt hold my breath for 28nm from TSMC either though.. took them this long to get 40nm processing working correctly.. 28nm is going to be a mess which is probably why AMD/ATI has been making a big push to move to globalfoundry's for their future generation chips..

 

[vengence]

Well EVGA just re-brands reference designs for their video cards. If they aren't selling, they simply order less of them. Their motherboards on the other hand require R&D. Therefore that business unit has to cover its R&D costs at a minimum or they lose money.

Most are. They do have some non-reference designs, ala the 285 classified and their 2GB cards.

I don't think they have any danger of not selling enough motherboards. Their reputation is quickly turning them into both having the best boards and the best support.

 

[sweetreddyd]

They're not going to continue doing well with their motherboards if there aren't any competitive Fermi boards incoming ... or maybe no Fermi boards available at all after the initial 5-10 thousand run ... their motherboard business is going to fall off a cliff.

Why would they do worse on motherboards if fermi doesn't come out? EVGA makes some of the finest motherboards out there.

I run tri-fire on a first gen Evga X58, and my friend with 4890 crossfire just bought a classified. This is the derpiest post I've read all day.

 

[heatlesssun]

well yes he did, when he absolutely had to that is, he like Nvidia's new power saving mobile solution.

Good point, even Charlie can only go so far, and he might have had a kind word for Tegra IIRC. But make no bones about it, this guy is biased, but then I guess he's not trying to hide it which is better than some IT journalists.

But buy a GPU based on his recommendation or not, never.

 

[Zachstar]

I dont even know what he is trying to do.

Nvidia fans do not like him because he is totally biased against Nvidia to the point of outright lies in my opinion.

ATI fans don't like him because they are tired of the echo chamber and they are tired of the high prices that lack of competition has caused.

All he is doing is fanning the fires of fanboism and that is beyond silly in my opinion.

 

[spigzone]

Why would they do worse on motherboards if fermi doesn't come out? EVGA makes some of the finest motherboards out there.

I run tri-fire on a first gen Evga X58, and my friend with 4890 crossfire just bought a classified. This is the derpiest post I've read all day.

Some of the most expensive too.

Xbit has an interesting article on a EVGA motherboard ... top notch quality - ATROCIOUS docummentation ... http://www.xbitlabs.com/articles/mainboards/display/evga-p55-ftw.html

If Fermi is a washout, who in their right minds is NOW going to invest in an EVGA SLI board with it's NF200 performance penalty if they're going to buy AMD cards?

As soon as Nvidia releases their few thousand GT470/480 cards and the review site numbers come in, there's going to be a tidal wave of holdouts and fence sitters headed AMD's way.

It's just silly to contend Fermi failing isn't going to severly impact EVGA's future SLI board sales.

It doesn't matter how awesome (and expensive) EVGA boards are it they have no Nvidia video cards to build.

Not to mention the 275/280/285 cards were EOLed a while back.

 

[mafa]

Getting more exciting, I think the PC gaming would be better off without nvidia craps.

 

[vengence]

Some of the most expensive too.

Xbit has an interesting article on a EVGA motherboard ... top notch quality - ATROCIOUS docummentation ... http://www.xbitlabs.com/articles/mainboards/display/evga-p55-ftw.html

If Fermi is a washout, who in their right minds is NOW going to invest in an EVGA SLI board with it's NF200 performance penalty and last generation DX10 Nvidia GPU boards when they can crossfire more powerful AND cheaper DX11 boards on a non SLI capable/nop performance penalty board.

As soon as Fermi releases their few thousand cards and the numbers come in, there is going to be a tidal wave of holdouts and fence sitters headed AMD's way.

It's just silly to contend Fermi failing isn't going to severly impact EVGA's future SLI board sales.

It doesn't matter how awesome (and expensive) EVGA boards are, ALL of their upper range cards are SLI and have the NF200 chip, and if someone is going to buy two 5000 series cards, why would they buy a motherboard with a built in performance penalty?

No they don't all have the NF200. Only the original classified did. They released a version quickly afterwards that didn't.

 

[heatlesssun]

I dont even know what he is trying to do.

Nvidia fans do not like him because he is totally biased against Nvidia to the point of outright lies in my opinion.

ATI fans don't like him because they are tired of the echo chamber and they are tired of the high prices that lack of competition has caused.

All he is doing is fanning the fires of fanboism and that is beyond silly in my opinion.

He's a sensationalist to get hits. That's why I REFUSE to read his stuff anymore because you are EXACTLY right. The best way to deal with the Charlie's of the world is to ignore them.

 

[Zachstar]

Getting more exciting, I think the PC gaming would be better off without nvidia craps.

That is one of the most silly things I have EVER heard.

Do you have any idea how markets work? Are you enjoying paying 300USD for an upper mainstream card? How about 400USD to make it more enjoyable when ATI is the sole serious Graphics card maker on the market.

Now no need to worry. Even as an ATI fan I enjoyed using their 55nm parts. And 55nm is getting cheaper and cheaper to make. Worst comes to worst Nvidia can skip 40nm alltogether and focus on flooding the market with more 55nm stuff.

 

[osirus35]

im pretty sure that those evga boards run CF perfectly fine even though EVGA does not advertise it.

 

[spigzone]

He's a sensationalist to get hits. That's why I REFUSE to read his stuff anymore because you are EXACTLY right. The best way to deal with the Charlie's of the world is to ignore them.

Because the degree to which he's been factually right or accurate has no significance to discerning minds such as yours?

Commendably rational response.

Glen Beck fan?

 

[spigzone]

No they don't all have the NF200. Only the original classified did. They released a version quickly afterwards that didn't.

If it's an i3, i5 or i7 board and does SLI, it has an NF200 chip.

And if it's not an i3, i5 or i7 board, what's the point?

 

[Valset]

Because the degree to which he's been factually right or accurate has no significance to discerning minds such as yours?

Commendably rational response.

Rush Limbaugh fan?

how about he has an objection to his tactics and choices to voice them by not bringing revenue to his site? I don't agree but that still a valid point and stance.

 

[Valset]

Getting more exciting, I think the PC gaming would be better off without nvidia craps.

I'm not really happy with them either but this was NOT good news for anyone. the market needs competition, ATI is not milking us like they might (kudos to them for that as well) but if fermi is half as bad as whats being painted they are going to have NO reason to drop theirs. I don't care to buy a 400 dollar graphics card when I might only be paying 325 instead. that doesn't look like its happening anytime soon. unless you own a lot of AMD stock this is not a good thing

 

[spigzone]

Now no need to worry. Even as an ATI fan I enjoyed using their 55nm parts. And 55nm is getting cheaper and cheaper to make. Worst comes to worst Nvidia can skip 40nm alltogether and focus on flooding the market with more 55nm stuff.

That is one of the most silly things I have EVER heard. (how could I resist?)

Doesn't work that way. Once yields are up, it's inherently substantially cheaper to make 40nm chips than 55nm chips with similar performance.

 

[spigzone]

how about he has an objection to his tactics and choices to voice them by not bringing revenue to his site? I don't agree but that still a valid point and stance.

Personally, on a scale of 1 to 10, factual/well-reasoned/accurate rates a 9.999 and 'tactics and choices to voice them'/discernable bias rates a .001.

How bout you?

 

[Zachstar]

That is one of the most silly things I have EVER heard. (how could I resist?)

Doesn't work that way. Once yields are up, it's inherently substantially cheaper to make 40nm chips than 55nm chips with similar performance.

Its not cheaper if 40nm yields are failing left and right with your hardware. And 55nm stuff is still selling.

While its not the best situation its better than the silly fantasy of ATI taking over and somehow prices dont skyrocket. "I think the PC gaming would be better off without nvidia craps. " Comments like this I wonder how they appear on this forum.