A Semi-accurate Look into Fermi

[SKYMTL]

do Fermi have eyefinity or something similiar?
of not, dosnt matter how fast it might be or how advanced calculations it can do.

its a failure.

That's a hilarious statement.

Yes, Eyefinity is great for pushing the limits of current graphics cards by asking them to render a huge amount of pixels and it really ups the immersive experience when gaming. However, much like NVIDIA's 3D Vision it relies on driver support which means it may or may not work with all games. As such, you can't determine success or failure based on a product's support of a niche market.

The main issue we now have is that current game engines just can't push the current generation of DX11 GPUs far enough. On the plus side, DX11 doesn't seem to be turning out to be the amazingly efficient API many promised so the extra rendering power will come in handy in certain situations. If you thought DiRT took a huge hit when rendering the DX11 code path, wait 'till you see what happens in AvP. This also brings into doubt the real selling point of lower-end cards that support a next-gen API but that happens in every generation.

All this means is that consumers won't have to spend the amount of money they used to when searching for an optimal gaming experience. I remember spending over $600 for an 7800 GTX which could play most games at high resolutions with IQ settings near max. However, stepping down to even a 7800 GT at $450 meant sacrificing that optimal gaming experience. Now, you can buy a $299 HD 5850 or whatever NVIDIA comes out with to compete with it and you can play every single current game at the highest detail settings while still sitting pretty when upcoming games are released. What's not to like about that?

Basically, what I am saying is that having massively powerful high end cards makes their offshoot mid-range products all the more appealing from a price / performance perspective. So, I can't see why anyone would call the possibility ultra high-end Fermi performance a "failure"

This doesn't make Fermi-based cards a failure, it just means that both Fermi and ATI's current generation of cards may have a longevity that far surpasses the G80.

[flippin_waffles]

lol wow, the web is a fickle place. all those that despise Charlie D suddenly approve of him because his article supports their loyalties... in reality, his writing style hasn't changed one iota. the only thing that's changed is the calender and it's getting closer to when nvidia finally has working silicon, so that is what he is reporting. so it turns out it should have been nv those people should have been dispising all along. oh the irony!

given patterns like this, it seems there must be some real inherent social dangers to the internet.


/observation

[skugpezz]

mh, i don't agree, you could also say the 5000 series is a failure as it doesn't support physx

eyefinity does not destroy games. By making it seem like walls cannot break, and dust particles are not possible without physx. physx FTL

[skycrane]

40%?
Hm, we'll see.
Hope that means 40% in games then.

knowing Nvida, its a 40% increase over the gtx295 @ 500 that would mean you could get a fermi for the low price of $700...

PS, the egg has the 5870 in stock for $ 400

http://www.newegg.com/Product/Produc...82E16814150443

[NickF]

I'm not sure what your source is, but that is an irrelevant detail as it is wrong. Nvidia has already provided Fermi based cards to some of its close partners, and performance is outstanding to say the least. The MSRP will not be higher than that of the GTX280, and this time, in comparison with the competition, there will be a very good price/performance ratio. If ATI does not significantly lower prices of their 59xx and 58xx cards, they will come off as a waste of money next to Fermi and its derivatives.

If that were true, nVidia would not be waiting for a third spin on its silicon (A3) right now, and it would have been launched in November.The obvious fact that Fermi is still not here is reason enough to believe that clock yields were NOT what nVidia was looking for- rather they completely missed the mark. So, I rather doubt that performance is as as outstanding as you claim.
I think we are going to see circumstances rivaling those of last generation.
-nVidia releases huge, bloated chip.
-Ati, this time feet wet in the market, drops prices and still maintains profit margins higher than nVidia.
-nVidia responds, cuts prices, loses money
-ATi responds, releases rv890-esque card
-nVidia responds, makes slightly weaker chip than its flagship to compete directly with the new ati card, and ends up destroying sales of its own flagship.

Just speculation, of course. Responding to a thread that got closed.

However, if these 40% perfromance domination over 5870 rumors are true, no amount of spinning will get ATi out of hot water. They'd essentially have to move the 5970 directly against the Fermi board, and if nVidia manages to overcome the thermals and design a nice dual chip board, ATi won't have an answer. I've seen the rumors on this 40% argument a few times, but yet to find credible reason to believe its true. Anyone have a source?

[003]

If that were true, nVidia would not be waiting for a third spin on its silicon (A3) right now, and it would have been launched in November.The obvious fact that Fermi is still not here is reason enough to believe that clock yields were NOT what nVidia was looking for- rather they completely missed the mark. So, I rather doubt that performance is as as outstanding as you claim.
I think we are going to see circumstances rivaling those of last generation.
-nVidia releases huge, bloated chip.
-Ati, this time feet wet in the market, drops prices and still maintains profit margins higher than nVidia.
-nVidia responds, cuts prices, loses money
-ATi responds, releases rv890-esque card
-nVidia responds, makes slightly weaker chip than its flagship to compete directly with the new ati card, and ends up destroying sales of its own flagship.

Just speculation, of course. Responding to a thread that got closed.

However, if these 40% perfromance domination over 5870 rumors are true, no amount of spinning will get ATi out of hot water. They'd essentially have to move the 5970 directly against the Fermi board, and if nVidia manages to overcome the thermals and design a nice dual chip board, ATi won't have an answer. I've seen the rumors on this 40% argument a few times, but yet to find credible reason to believe its true. Anyone have a source?

The A1 and A2 revisions were operational but yields were not fantastic nor did clock speeds measure up to targets. This is where A3 steps in. I doubt clocks will reach 100% of the original targets, but they will be quite a bit better than A2, which is a very good reason for another respin. Things are going well now and everything is on track for a launch in mid Q1.

[NickF]

The A1 and A2 revisions were operational but yields were not fantastic nor did clock speeds measure up to targets. This is where A3 steps in. I doubt clocks will reach 100% of the original targets, but they will be quite a bit better than A2, which is a very good reason for another respin. Things are going well now and everything is on track for a launch in mid Q1.

My point exactly. There's no chips that any "close partners" could have access to that would be performing up to par. Even nVidia won't have them until A3 rolls out of TMSC. If any of nVidia's close partners were sampled an A2 silicon, it would have been a disappointment, and would not have delived "outstanding" performance.

[Helloworld_98]

Guys, I was just wondering, now that the Acer 24" 3D display is coming out, and 3D is becoming a major thing,

do you think the 5890 won't sell very well because of no 3D?

[NickF]

Guys, I was just wondering, now that the Acer 24" 3D display is coming out, and 3D is becoming a major thing,

do you think the 5890 won't sell very well because of no 3D?

I don't think 3d will make a dent in the market this year.
You wanna stay on topic next time?

[bfar]

Guys, I was just wondering, now that the Acer 24" 3D display is coming out, and 3D is becoming a major thing,

do you think the 5890 won't sell very well because of no 3D?

3D Vision, Eyefinity, Physx etc - all useless gimmicks

It's all about raw graphics power and good driver compatibiliy, and it will stay that way for some time to come.

gpu compute will sell gpu's to the home market, but not $600 ones.

[Helloworld_98]

I don't think 3d will make a dent in the market this year.
You wanna stay on topic next time?

it's fairly on topic considering if it was a big thing people would turn to Fermi instead of the 5 series because nvidia is the only graphics company with 3D drivers.

[D749]

mh, i don't agree, you could also say the 5000 series is a failure as it doesn't support physx

Well you can get ATI to play nice with a NV card for PhysX. You just can't use an ATI card for PhysX.

http://www.xtremesystems.org/Forums/...9&postcount=90

I'm looking forward to NV's new lineup. I want competition!

[003]

My point exactly. There's no chips that any "close partners" could have access to that would be performing up to par. Even nVidia won't have them until A3 rolls out of TMSC. If any of nVidia's close partners were sampled an A2 silicon, it would have been a disappointment, and would not have delived "outstanding" performance.

Actually, they have and currently are. I'm not going to say specifically who to avoid any issues over NDA but performance is very good. Better than the GTX295. I'm not sure if it was A2 or A3 that was and still is being sampled, but performance was in fact outstanding.

When I say partners I am not talking about AIBs such as eVGA but other companies who have close ties with Nvidia.

[Eastcoasthandle]

lol wow, the web is a fickle place. all those that despise Charlie D suddenly approve of him because his article supports their loyalties... in reality, his writing style hasn't changed one iota. the only thing that's changed is the calender and it's getting closer to when nvidia finally has working silicon, so that is what he is reporting. so it turns out it should have been nv those people should have been dispising all along. oh the irony!

given patterns like this, it seems there must be some real inherent social dangers to the internet.


/observation

And I thought I was the only one who noticed this. They seem to love Charlie now. After reading the article (not just what was quoted in the OP which omitted a lot) Charlie hasn't deviated at all. For example:

If you assume that A3 will fix all the problems, and that it will be the launch stepping, there are two main things that will affect the launch date from there. First are the risk wafers. If they are still valid, then that will probably shave a few weeks off the time until production silicon will start rolling off the lines. Lets call the savings about four weeks, or six weeks versus ten weeks if the risk wafers had to be scrapped.

The next variable is based on how sure Nvidia is that A3 worked out the way it wanted that to. If it is 100% sure that A3 is the one it will launch, Nvidia could have started running wafers at the same time it put the A3 hot lots in.

Since the problems that we hear about on A2 silicon are related to bin splits and clock speed problems, and general yield in the good versus bad chip sense, this might be a fairly large risk for Nvidia to take. Given that, six weeks out from Dec 1 if it put A3 in on day one and the risk wafers are still valid, ten weeks if not. If it waits, that is +6 and +10 weeks from Jan 1 or so.

Last we have the time it takes to make boards, put them in boxes, and ship them. A good guesstimate is two weeks for the initial boards, including air freighting them from Taiwan or mainland China to get them ready for the big party. This assumes the prep work has been done, and the PCBs are simply waiting for chips.

The first take home message is that Nvidia can likely show hot lot early A3 silicon at CES, but we doubt the company will launch it there. Tegra 2 is probably a better fit for that, but don't be shocked if it has A3 silicon floating here and there.

If all goes well, risk wafers valid, A3 production went in on Dec 1, and all the stars align, you are looking at Dec 1 +6 weeks +2 weeks, or about Feb 1 for real availability. More realistically, if it waits until the A3 hot lots get back, that would be Jan 1 plus eight weeks, or March 1 for a hard launch. Basically, the best case is Feb 1 for a real launch.

The negative view is that the risk wafers are invalid, and Nvidia waits for A3 to come back before it puts in production wafers. That would be the outer bound, Jan 1 +10 +2 weeks, or April 1, Q2 of 2010. If you have a really negative outlook and think there will be an A4 stepping, add about four to six weeks for every stepping past A3. That is really unlikely to happen - really, really unlikely.
...

[WeaponX]

umm... shouldn't it already be better than the GTX295? what kind of statement is that?

[Syn.]

This is a surprising decent article on Charlie part. Less hate than usual, I hope he can keep it up.

Maybe he realized that ranting and raving wins him no fans.

Maybe hes still the same and its just you two hating him less then before...

[Solus Corvus]

40% sounds nice. ATI should have some 5x90 cards out by then and hopefully we will see some price/perf wars.

Actually, they have and currently are. I'm not going to say specifically who to avoid any issues over NDA but performance is very good. Better than the GTX295. I'm not sure if it was A2 or A3 that was and still is being sampled, but performance was in fact outstanding.

When I say partners I am not talking about AIBs such as eVGA but other companies who have close ties with Nvidia.

Sure dude. Considering that you have never shown that you have any real connections in the industry and your past predictions were way off the mark - I'd suggest that it is just as likely that you read Charlie's article, liked what you saw, and ran with it. Especially since your brand new "insider info" only came out after charlie's article and doesn't contain any new information. You could resolve this by simply giving a bit of info that hasn't been publicly posted before, but we all know that isn't going to happen.

[Calmatory]

but I believe that adding in tessellation will be the PC-programming-lazy developers' way of making the game look better

Let me guess; you aren't a programmer, right? At least you've got no clue about the industry. But yeah, great claims there, pal. It's just the contrary.

[trinibwoy]

After reading the article (not just what was quoted in the OP which omitted a lot) Charlie hasn't deviated at all.

The content has stayed the same but this particular article is definitely written in a different tone and style. That is, generally free of the usual doom and gloom and mockery that permeates the usual stuff. It's either he's heard something good or he's starting to realize that his message gets lost when it's wrapped up in all that other drivel.

40% faster than Cypress is great and all but still below expectations. Right now Cypress is about 30-35% faster than the 285 so that puts Fermi targets at ~90% faster than the 285. That's disappointing given the specs and the architecture overhaul.

[Manicdan]

i think TSMC is just the one disappointing, both companies are trying hard to make the best of the silicon

[Eastcoasthandle]

The content has stayed the same but this particular article is definitely written in a different tone and style. That is, generally free of the usual doom and gloom and mockery that permeates the usual stuff. It's either he's heard something good or he's starting to realize that his message gets lost when it's wrapped up in all that other drivel...

I can't agree with that. His tone has been the same to me. His articles have always been entertaining. Perhaps you are reading more into it then you should? The content of his articles always suggested a 2010 release date, for example (as well as other tidbits). Even though some tried to call him out for a November 2009 release. So far, he's been fairly accurate regarding any news with that company. And, has offered his guesstimate on when he thinks their products would be available.

[trinibwoy]

Heh, when is the last time you've heard a positive slant on anything NV related from Mr Demerjian? No mention of big/slow/late/hot etc, etc either. Really strange.

"A3 will almost assuredly up the clocks quite a bit" seems quite generous to me given that he usually takes the other side of those bets.

"If you have a really negative outlook and think there will be an A4 stepping, add about four to six weeks for every stepping past A3. That is really unlikely to happen - really, really unlikely." is also smirk inducing given that he's usually the one with the really negative outlook.

Anyway, given that Charlie seems to be focusing less on his love affair with Nvidia, maybe we should follow his lead. Better writing is always welcome.

[Frag Maniac]

But February is traditionally a red month. <3


Maybe Nvidia and ATI can be Valentines and finally get along. LOL

[BababooeyHTJ]

He was correct so far when it comes to Fermi silicon state,he might be correct this time too.

I was thinking the same thing.

do Fermi have eyefinity or something similiar?
of not, dosnt matter how fast it might be or how advanced calculations it can do.

its a failure.

speed is of no essence, as current generations is fast enough.
features that are useful will be decding factor and ati got eyefinity right.

Current generations are not fast enough. Throw some texture mods at Oblivion or Fallout 3. Play a little Darkplaces. Throw some mods at Stalker, Crysis, Dragon Age, etc and tell me if you want more power.

Honetly eyefinity is about as usefull to the vast majority of their customers as that 3d vision crap.

[Cybercat]

lol wow, the web is a fickle place. all those that despise Charlie D suddenly approve of him because his article supports their loyalties... in reality, his writing style hasn't changed one iota. the only thing that's changed is the calender and it's getting closer to when nvidia finally has working silicon, so that is what he is reporting. so it turns out it should have been nv those people should have been dispising all along. oh the irony!

given patterns like this, it seems there must be some real inherent social dangers to the internet.


/observation

It is possible to write negative news about NVIDIA and not sound the way Charlie does. Most journalist do it all the time, and you don't see me or anyone else complaining about them. It has everything to do with how things are worded, and Charlie has always managed to sneak in a snide remark about NVIDIA's general competency as a company or their dishonesty where it had no relevance.

He editorializes more than he reports news, yet he claims to be a journalist.