Nvidia works on second generation Fermi

[PingviN]

I would think pushing back Fermi 2's release is something they would never consider.

"Hey, lets push back our refined, higher performing product which is quite possibly cheaper to make in favour of keeping our much delayed product on the market longer, while letting out competitors increase their lead".

No thanks.

On the other hand, announcing the next gen will without a doubt put a dent in GF100's release. I'm sure Nvidia would love to keep a lid on this, why would they want to leak this kind of information when it will only damage sales on their own product?

 

[pmv]

So true. The competitors also know they're the only game in town currently so your lunch will be costing more than usual today.

Fortunately I'm not that hungry at the moment.

Really wish I'd bought sandwiches though.

 

[TheRickRoller]

On the other hand, announcing the next gen will without a doubt put a dent in GF100's release. I'm sure Nvidia would love to keep a lid on this, why would they want to leak this kind of information when it will only damage sales on their own product?

Really? I knew about Haswell in either '08 or '09 (release in '12), the next-next-next gen after penryn. Didn't stop me from getting a penny, though.

 

[MrK6]

Really? I knew about Haswell in either '08 or '09 (release in '12), the next-next-next gen after penryn. Didn't stop me from getting a penny, though.

Fermi 2 isn't four years away though.

 

[Keysplayr]

So. If somebody asked you (MrK6) or you (PingVin) today whether they should buy a 5870 or wait a month for Fermi, you'd say....?
Probably say to buy what you want right now. Is this correct? Why wait for a month for Fermi when you can get what you want right now. Yes?
Which is why I'm having a hard time imagining why you guys might think that a Fermi2 Launch 1/2 to 3/4 of a year after Fermi 1 would make people wait for it and not buy Fermi 1.
Ya see what I did there?
Seriously?
But I'm more than willing to admit I'm incorrect. For all I know, you very well could tell people to wait the month for Fermi. Or wait the 6 to 9 months for Fermi 2. Would you do that?

 

[MrK6]

So. If somebody asked you (MrK6) or you (PingVin) today whether they should buy a 5870 or wait a month for Fermi, you'd say....?
Probably say to buy what you want right now. Is this correct? Why wait for a month for Fermi when you can get what you want right now. Yes?
Which is why I'm having a hard time imagining why you guys might think that a Fermi2 Launch 1/2 to 3/4 of a year after Fermi 1 would make people wait for it and not buy Fermi 1.
Ya see what I did there?
Seriously?

Um, you reiterated the situation in your own words, what exactly did you do beyond that? I don't understand why that's a difficult concept. We aren't talking low-end/entry/budget here, we're talking flagship. People that buy flagship products generally have a different ideology than those upgrading out of necessity in the low-end/mid-range markets. Let me explain.

To answer your question, if someone posted "Want to upgrade to a pwnage video card, 5870 or Fermi?" in this forum right now, I'd ask them the situation first. First, if they were 1920x1200 or below, I'd say go grab 5850, because there is nothing in the foreseeable future that is going to be able to even touch that price/performance ratio. If they simply wanted the best single GPU money can buy, well, I'd have to split hairs. Right now we know all about the 5870 - Great performance, great overclockability, cool running with low TDP of 190W load/25W idle, quiet, some great features like Eyefinity, etc., overall a great package. From there, I would extrapolate based on what we know about Fermi - faster than the 5870 (I'd say 20-25% is a safe estimate), much higher power draw (current estimates I've seen point to the 250W TDP range), unknown aesthetic/environmental characteristics (sound, idle/load efficiency), new features but unknown how they work (NVIDIA Surround, etc.), higher priced (who knows how much?). The user then has to guess, based on those criteria, is it worth it or not to wait, also considering the potential price drops on the 5xxx series should they be necessary upon Fermi's release.

Now how does this relate to my Fermi 2 suggestions? Well, say you're someone like me and already have a 5870 or other high-end card setup. The 5870 runs every game I've played on it since September at 2560x1600 and maxed settings, I don't really need an upgrade. Now if Fermi was going to be the last video card to be released for awhile (haven't heard anything about a 5xxx refresh, let's say that doesn't come out until August anyway and isn't anything spectacular), it might be worth it to someone like me to buy Fermi should it be an acceptable part, just to tinker around with a faster part. However, if Fermi "done right" is being released 6 months later, well, I might just wait to try that one. It's a situation where the upgrade isn't needed, but just done out of interest/boredom.

 

[Keysplayr]

I didn't ask that. We are talking time here. All other things being equal. But you knew that.
There is no right or wrong answer here.

 

[MrK6]

I didn't ask that. We are talking time here. All other things being equal. But you knew that.
There is no right or wrong answer here.

So then beyond being confrontational, what was the purpose of your post?

 

[RussianSensation]

I am betting all they are going to do is a Fermi refresh with faster clock speeds similar to 8800GT to 9800GT or maybe add some extra shaders and call it a new generation... Even to get a modest jump as was the case from 6800U to 7800GT is going to take a die shrink on their part.

If anyone has any shot at releasing a new generation this year, it's ATI.

 

[Seero]

Those who determined to buy ATI's 5xxx series, no one is stopping you, go ahead and buy it. Fermi is going to be much more expensive. All factors showed that. Long R&D, bad yield, and expensive parts means high price. If price/performance is what you want, go for ATI. Fermi is not going to be cheap, so your ATI cards will retain its value.

It is clear that they have their re-engineered GT200 core up and running. Yes, Nvidia can simply copy the design and create a list of product range from low to high end based on different parts, or they can actually alter the design to branch it out. I7 and I5 are not from the same die, and they are used for different markets. What is wrong with that?

Fermi was known to have crazy ECC RAM, which is good for computing. Yet there were also voices saying that this is an overkill. The number of resistors may affect yield, can it be reduce to increase yield, thus lead to a cheaper chip? Is the die size too big? So does the new design worked? Does it allow games to run at high FPS? PhysX running okay? CUDA running okay? Can power consumption be decrease? Can efficiency be increase? So many questions and we don't have the answer, yet the answers are sitting in Nvidia's Lab. So they have the answer, what are they waiting for?

Now, using Intel's Tick-Tock theory, Fermi is clearly the Tick as the design is so new. I bet engineers have worked day and night just to patch the chip, trying to make it run at a stable clock speed without frying things. The 2nd generation is probably a plan to create the Tock version of Fermi.

 

[Keysplayr]

So then beyond being confrontational, what was the purpose of your post?

In other words, the way I phrased that post, you can't answer it. It wasn't confrontational, or wasn't "meant" to be confrontational. My questions were quite clear. If you can't answer them, then by all means, don't. Nobody says you "have" to.

 

[shangshang]

Keys,
ask every major marketing dept of every major company that has products to sell, from consumer electronic to medical electronic to car company, and most will tell you that their customers (big buyers, small buyers, and in between) do tend to wait and sit on their purchasing decision when they see new products on the horizon. And MrK6 is right, the bigger the purchase, the more likely that customers will "wait and see".

If Fermi2 is 6 months out from Fermi1... and with most "last generation high end" users possessing either a GTX295, 285, HD4890,... which can satisfied most current games and with no real games on the horizone demanding Fermi1 power,... you can bet there will be many high-end buyers saying "i'll keep my current card and let's wait and see about Fermi2 when some more demanding games will be out" attitude.

 

[Lonyo]

Keys,
ask every major marketing dept of every major company that has products to sell, from consumer electronic to medical electronic to car company, and most will tell you that their customers (big buyers, small buyers, and in between) do tend to wait and sit on their purchasing decision when they see new products on the horizon. And MrK6 is right, the bigger the purchase, the more likely that customers will "wait and see".

If Fermi2 is 6 months out from Fermi1... and with most "last generation high end" users possessing either a GTX295, 285, HD4890,... which can satisfied most current games and with no real games on the horizone demanding Fermi1 power,... you can bet there will be many high-end buyers saying "i'll keep my current card and let's wait and see about Fermi2 when some more demanding games will be out" attitude.

AMD seem to have no problem announcing that they will be releasing new higher performing products this year, which will be around 9 or maybe a few more months after the release of the HD5870.
Any enthusiast who also follows any of this stuff already knows that 6~12 months for refreshes or new cards is fairly standard anyway, and if you are waiting another 9 months for Fermi Mk2, then hey, you will get to wait some more for ATIs next products after their refresh, because by then they will be less than 6 months away.

As I have already said, you can play the waiting game for as long as you want, there will always be new products coming, and people KNOW there will always be new products coming, because that's how it always happens.

I can tell you right now that after this "new" Fermi product, within another 6 to 9 months there will be yet another product from someone that will be even better than the "new" Fermi that is 6~9 months away from now.

And 6~9 months after that product? There will be something even better!

We already know that ATI have something up their sleeves for later this year, now we know Nvidia do as well.
If you want to wait, then wait, if not, don't.

 

[Keysplayr]

Keys,
ask every major marketing dept of every major company that has products to sell, from consumer electronic to medical electronic to car company, and most will tell you that their customers (big buyers, small buyers, and in between) do tend to wait and sit on their purchasing decision when they see new products on the horizon. And MrK6 is right, the bigger the purchase, the more likely that customers will "wait and see".

If Fermi2 is 6 months out from Fermi1... and with most "last generation high end" users possessing either a GTX295, 285, HD4890,... which can satisfied most current games and with no real games on the horizone demanding Fermi1 power,... you can bet there will be many high-end buyers saying "i'll keep my current card and let's wait and see about Fermi2 when some more demanding games will be out" attitude.

While this is true in some cases, it certainly does not reflect all cases.
What I wanted to know was, how can they think that Fermi2 launching Q3/Q4 would stop anyone from buying Fermi1? If they want it, they'll buy it. And, that's when I asked about the 5870 or wait for Fermi example. You see, if he says he'd tell them to buy now, that goes against their comments earlier in this thread. If he says, wait six to nine months, then that would negate any further recommendation he can make to buy a 5870 now.

So, I wanted to know where he stood. Both of them actually. Which is it?
And, neither of them will tell me.

 

[v8envy]

While this is true in some cases, it certainly does not reflect all cases.

Personal anecdote. I'm *still* on an 8800GT. I'm desperate to scratch the upgrade itch, but not willing to abandon Linux or settle for an overpriced, underfeatured, EOLed G200. At this point it looks certain that the 1st gen Fermi is going to be a repeat of the FX series. I'm gambling on NV having a Fermi done right sooner rather than later -- just like ATI fixing the X1800 faux pas with the X1900 as soon as they could.

So, these news have convinced me to hang out till Fermi2, running hardware from 2007. Go me.

 

[MrK6]

In other words, the way I phrased that post, you can't answer it. It wasn't confrontational, or wasn't "meant" to be confrontational. My questions were quite clear. If you can't answer them, then by all means, don't. Nobody says you "have" to.

I answered your questions. If those weren't the answers you were looking for, then you phrased your questions poorly.

While this is true in some cases, it certainly does not reflect all cases.
What I wanted to know was, how can they think that Fermi2 launching Q3/Q4 would stop anyone from buying Fermi1? If they want it, they'll buy it. And, that's when I asked about the 5870 or wait for Fermi example. You see, if he says he'd tell them to buy now, that goes against their comments earlier in this thread. If he says, wait six to nine months, then that would negate any further recommendation he can make to buy a 5870 now.

So, I wanted to know where he stood. Both of them actually. Which is it?
And, neither of them will tell me.

I told you quite clearly my recommendations. Shangshang understood them as well, why can't you? What are you trying to do here besides coyly dance around an issue? If you have a question, please ask it directly, evidently because my answering multiple questions in a post confuses you.

 

[ronnn]

This is getting silly. Nvidia pr have been broadcasting the greatness of fermi for several months. Most of us believe the reason for this was to get people to wait before buying an ati product. If knowing fermi will be great and available by april, may or june can affect sales of the 5870 now, why would knowing fermi 2 is just around the corner not affect sales of fermi? Calling out individuals as hypocrites really doesn't change this.

 

[Phil1977]

95% of Nvidias customers don't read Anandtech or any hardware sites for that matter. That's why this news info will have no impact on fermis success...

People buy a computer and parts on a need basis. What good is Fermi rev. 2 if you need a PC / part / upgrade right now or you want to play this particular game and are happy to buy whatever is the best...

Fermis success depends on pricing, performance and how it compares to ATIs offerings...

 

[Martimus]

So. If somebody asked you (MrK6) or you (PingVin) today whether they should buy a 5870 or wait a month for Fermi, you'd say....?
Probably say to buy what you want right now. Is this correct? Why wait for a month for Fermi when you can get what you want right now. Yes?
Which is why I'm having a hard time imagining why you guys might think that a Fermi2 Launch 1/2 to 3/4 of a year after Fermi 1 would make people wait for it and not buy Fermi 1.
Ya see what I did there?
Seriously?
But I'm more than willing to admit I'm incorrect. For all I know, you very well could tell people to wait the month for Fermi. Or wait the 6 to 9 months for Fermi 2. Would you do that?

Does this mean that Fermis is being released next month? If so, that is awesome. If not, that kind of sucks. I don't want to spend $500+ on a 5870, and the Fermi release may drop the price down to something I would find more palatable.

 

[Obsoleet]

Yet another reason not to buy until November!

Nah, the November after is better.

 

[dguy6789]

95% of Nvidias customers don't read Anandtech or any hardware sites for that matter. That's why this news info will have no impact on fermis success...

People buy a computer and parts on a need basis. What good is Fermi rev. 2 if you need a PC / part / upgrade right now or you want to play this particular game and are happy to buy whatever is the best...

Fermis success depends on pricing, performance and how it compares to ATIs offerings...

95% of Nvidia's customers will never buy a high end GPU. Most people that buy the highest of the high end parts are computer enthusiasts who frequent forums like this one. The rest are probably rich people who just buy the most expensive Alienware model they see on their site.

That said, I don't think this news will impact Nvidia negatively like some people think. The sales of the highest end parts aren't even important to Nvidia or ATI, they are extremely tiny niche markets(Some people are surprised that AMD's market share didn't rise a ton because of the 5800, but I'm not because I know high end sales do not put a dent in overall marketshare). The high end stuff is done almost entirely for marketing purposes. A customer sees a review showing the fastest Nvidia card beating the fastest ATI one and all they remember is Nvidia good ATI bad when they go to a retail store and look at $50-$100 graphics cards. The mid and low end GPU sales(and integrated solutions) are what these companies care about.

 

[tviceman]

Nah, the November after is better.

If I were you, i'd just wait until DNF is finished and bundled with the best available GPU to buy.

 

[Seero]

I failed to see what is there to debate. Traditionally, Nvidia releases a new line of chips every year. The last one was delayed by about 3 months, but this may not impact the next one which is suppose to be on released at the end of the year.

To Nvidia hater, they complain everything. Shall they delay the next chip, they will complain. Shall they not delay the next chip, they will complain. To prove this, no one here even know what the next generation is going to be, but are already complaining about the "time of release". Come on guys, there is nothing about the next generation yet other than there will be one and its release time may not be impacted by the delay of the current one.

Excuse me but there are faster video cards coming out every month or so anyways. 5870 -> 5970 -> 5870 OC -> 5870 extreme -> 5970 on cracks -> 5890 deep freeze. GTX 280 -> 295 -> 280 SC -> 280 SSC -> 285 -> 285 SC -> 285 SSC -> 8800 rebrand, I mean come on. Who had brought them all here? Say 5870 is your cup of tea, but 5850 is cheaper and 5890 is around the corner, may be you will wait, or maybe you won't, but doesn't matter what you do, some newer cards will come out. So ATI sit their 5890 for 12 months because those who just brought 5870 won't be buying it? Get real. The current Fermi probably cost too much to build, and the next gen is probably a cut down of the original Fermi + design fixes, or maybe a simple shrink die, who knows?

 

[secured2k]

I checked out the Fudzilla Article and cannot seem to determine that this refresh means a better card. As I look at the release dates for the original branded GeForce 8 series, I see about 5 months before lower end and cheaper refined G8x core products came out. 6 months later came an Ultra card. G92 products didn't come out until after a full year after the original 8800 series release.

It's my guess that the product refresh will be the same - new lower end cards from chips that couldn't be used with all the cores or parts enabled. As manufacturing processes improve, the cream of the crop chips will get their Ultra-like card.

11-2006: 8800 GTS (G80)
11-2006: 8800 GTX (G80)

04-2007: 8500 GT (G86)
04-2007: 8600 GS (G84)
04-2007: 8600 GT (G84)
04-2007: 8600 GTS (G84)

05-2007: 8800 Ultra (G80)

06-2007: 8400 GS (G86)

07-2007: 8300 GS (G86)

12-2007: 8800 GT (G92-200)
12-2007: 8800 GTS 512 (G92-400)
12-2007: 8400 GS (G98)

01-2008: 8800 GS (G92-150)

 

[taltamir]

Yet another reason not to buy until November!

so... you justify waiting 10 months by the fact that... *drumroll*; 10 months from now there will be better hardware out?
look up moore's law. There has been and there will be "something better" 10 months from now for any product, ever... assuming nvidia even sticks to the time table and manages to get it 10 months away and not months later...

Every piece of hardware out there... heck, even your CAR, will have a better version available a year from now.