If fermi flops what happens to Nvidia

[OCGuy]

Guess you meant this as a joke.

As the lack of artwork on the box could also be saying what evga thinks of it.

I guess you have a point on the $$$'s sitting on the sidelines waiting to see if fermi was worth the wait.

Given the amount of time fermi has been delayed and the lack of nvidia hype it doesn't look good for fermi.

But I guess the diehard nvidia fanboys always will have fermi 2 to wait for. Or will just buy a slightly worse or better card because it's a nvidia!

I don't care either way if it flops or not as I'm not in the market for a video card anyways.

I only brought up the subject to see what others thought about the situation. It's best for all parties concerned to have multiple players in the market as it spurs innovation and helps keep prices in line.

Yes, the lack of artwork on the box shows what EVGA thinks of it. If there were 6 anime chicks on it, then you know the card is l33t!

More than just "diehard nV fanbois" are waiting for the card. Fermi is going to sell like hotcakes unless they price it out of the market or it comes to light that it somehow causes cancer.

Is there a chance for the former? Absolutely, but I think most of us are hoping they learned their lesson with the GT200 launch price fiasco that causes XFX to give up exclusivity.

 

[thilanliyan]

Either way, there is a ton of enthusiast $$$ on the sideline waiting for the 480. Just looking at the EVGA box art got my Debit Card trembling...

You don't even know the performance yet and your debit card is trembling? :\

 

[SlowSpyder]

Fermi isnt flopping, much to Charlie's dismay.

It will be the fastest chip, and they can only screw it up by pulling a GT200 and pricing too high at launch.

Either way, there is a ton of enthusiast $$$ on the sideline waiting for the 480. Just looking at the EVGA box art got my Debit Card trembling...

I think you're wrong... my guess is Fermi will be a bit faster than the 5870 and nearly impossible to find due to product shortage... at least for a while.

Isn't Fermi already considered somewhat of a flop? Where is it? Nvidia's compeitor has had their DX11 parts out for going on 6 months with no competition.

As far as the topic of this thread, Nvidia will be fine. They'll go full speed ahead on Fermi 2 and be right back in the race, probaby be the leader in performance in a year. Nvidia has a lot of different markets they have their hands in.

 

[Mr. Pedantic]

I love how so many people think that the high-end consumer GPU business is the be-all and end-all of NVidia.

NV has other markets, but gpu/gpgpu is their core business and their future relies on the technology they produce. ATI on the other hand is just the high end graphics division of a cpu/chipset/gpu maker. Should the ATI graphics division start to hemorrhage cash, how long will AMD keep them on life support? I'm pretty sure this is not lost on ATI management. For ATI, I think being successful is a do or die situation. For NV, it just means they have to dig in, and get back to work.

In the recent past, the graphics division of AMD was the only bit of them that was turning a profit. And I don't think the GPU division's successes with RV700 and RV800 are flukes; I think AMD's strategy is a good one, it's based on sound reasoning given their primary GPU market and backed by solid engineering and design principles, and ATI (if I may call the GPU division that) will go from strength to strength, at least in the short term. Obviously, with parallel design teams, the success of the HD6000 series is anyone's guess, but if AMD follow the principles they have so far, it is looking good.

AMD has already sold off GF, the thing that was dragging them down most; if anything, next to go would be the CPU division.

 

[cbn]

I love how so many people think that the high-end consumer GPU business is the be-all and end-all of NVidia.

Is the GPGPU business and other non-gaming areas the major reason why Nvidia wants to keep Physx proprietary?

I think I read somewhere that Nvidia is approaching technological lock-in for certain productivity software types such as Adobe CS5, but I am not certain about this because I don't use those types of programs.

I guess Nvidia can't make Physx/CUDA proprietary for productivity software but open source for games right? (Yep, that would make sense. This is probably a rhetorical question)

 

[OCGuy]

I think you're wrong... my guess is Fermi will be a bit faster than the 5870 and nearly impossible to find due to product shortage... at least for a while.

Isn't Fermi already considered somewhat of a flop? Where is it? Nvidia's compeitor has had their DX11 parts out for going on 6 months with no competition.

As far as the topic of this thread, Nvidia will be fine. They'll go full speed ahead on Fermi 2 and be right back in the race, probaby be the leader in performance in a year. Nvidia has a lot of different markets they have their hands in.

I guess we'll see at launch. My guess is it will be hard to find due to limited supply and high demand.

They'll be the single GPU performance leader much sooner than a year.

Will it be worth it? Who knows. That comes down to the MSRP and if the e-tailers take advantage of limited supply and raise prices on their own.

My instinct tells me to sit on my hands until Fall.

 

[TuxDave]

General musing:
I'm not sure how many other nanoscience savvy members we have on board but I always thought the Fermi name was almost ironic. Coming from the term Fermi energy, which grossly overly simplified, deals with the energies of promoting electrons from their valance band, past the band gap and up to the conduction bands. Even more simply, the energy needed to turn a semiconductor from an insulator to a conductor. The Fermi energy, in many cases, actually resides at or just below the top of the valance band. Viewing Fermi as the nonconducting (not currently doing any work...ok that's a bit of stretch, as gates DO need to be off too, but stay with me here on this) state of the architecture and refresh is given the extra oomph required to push things up to the conduction band and get interesting. Now that I've cemented myself as the nerdiest of the nerds I'll just shut up

Most energy band diagrams (semiconductor physics) I have seen shows the Fermi level betwen the valence and conduction band. Basically marking the halfway point of the energy gap (the energy required to move an electron/hole from the valence to the conduction band). Not really sure what this symbolizes except that if you get to Fermi, you're only halfway there?

 

[Apocalypse23]

I don't think much will happen if Fermi flops, regular users still buy any "fast" video card if the price is right. Nvidia cards have generally been good to me in the past, I have switched to ATI due to performance and cost issues lately. I'm sure Nvidia has a stable fan following, and I honestly don't think it will flop, it will perform around the 5870 standards which is quite good honestly for a Dx11 card. Yes it's very late but the average Joe's won't notice the discrepancy. Even if the cards are slower than the Radeon's, fans and everyday users will continue to buy them due to their past experiences (positive/negative depending). You can think of it as folks buying new cars or leasing new cars every few years, they don't look at the performance, they just buy it since it's new and expect it to be a better and a more efficient product than the predecessor.

Nvidia would be in the blues territory if their new cards performed significantly poorly to the Ati counterparts, I'm talking 30-50 fps slower territory, which is unlikely.

 

[NoQuarter]

I guess we'll see at launch. My guess is it will be hard to find due to limited supply and high demand.

They'll be the single GPU performance leader much sooner than a year.

Will it be worth it? Who knows. That comes down to the MSRP and if the e-tailers take advantage of limited supply and raise prices on their own.

My instinct tells me to sit on my hands until Fall.

Wait does sit on your hands until Fall mean you will have avoided upgrading for a year just to get a Fermi?

 

[Keysplayr]

Wait does sit on your hands until Fall mean you will have avoided upgrading for a year just to get a Fermi?

It could mean that. It could also mean he can wait until fall to see what pans out. Or it could mean that he will buy an Nvidia card no matter what.

I'm pretty sure I know which one you might think it is.

 

[SlowSpyder]

I guess we'll see at launch. My guess is it will be hard to find due to limited supply and high demand.

They'll be the single GPU performance leader much sooner than a year.

Will it be worth it? Who knows. That comes down to the MSRP and if the e-tailers take advantage of limited supply and raise prices on their own.

My instinct tells me to sit on my hands until Fall.

I'm not saying it'll take a year from now for them to be the performance leader, I meant it as a year from now when we look at the available cards Nvidia will probably have the faster part.

But when Fermi launches, if the rumors of 5000-10000 parts are true, even with moderate demand they'll likely sell out. Once it launches and the initial launch parts are gone, I bet it'll be pretty hard to find one for a while... they'll trickle in slowly. If that happens, can we say that Nvidia has the faster GPU when it's nearly impossible to find?

I know you strongly prefer Nvidia, so I wouldn't count on you getting a Radeon setup no matter how Fermi performs or costs. I imagine dual GTX280's is still pretty capable anyway (I thought that's what you had anyway?)

 

[Dribble]

Success of fermi isn't a binary thing - it's not either a complete failure or the greatest success ever, there's a whole pile of possibilities in-between. Chances are that's what we'll get.

As for it effecting nvidia's future I don't think nvidia are that worried - they were never going to make most of the mony selling high end geforce cards even if it was perfect. The money is in quadro, telsa, and the oem markets. Quadro/tesla have small numbers but huge markups, nvidia has dominance there now, and the gpu compute angle which is fermi's strength should guarantee them dominance in the future. Oem's have low markup's but huge quantity and they care more about new numbers on the cards then what's actually in them (i.e. they seem quite happy with rebadged 8800's so when fermi arrives I don't see them caring if it's not quite as fast as it could have been).

I suspect nvidia were much more worried about intel locking them out of the chipset market as a lot of nvidia's revenue came from that. To get around that they have the already released optimus which seems to work very well, although whether it will make up for all those lost chipset sales is debatable.

As others have said AMD is the one with real worries, fundamentally they don't seem to be able to make money. Ati has moments of success, but it can't keep the company in the black alone. They still have huge amounts of debt, and most worryingly are obviously not able to invest in the new up-coming markets (mobile/netbook and gpu compute) like intel and nvidia can do. Mid to long term future is still pretty bleak

 

[Kenmitch]

I suspect nvidia were much more worried about intel locking them out of the chipset market as a lot of nvidia's revenue came from that. To get around that they have the already released optimus which seems to work very well, although whether it will make up for all those lost chipset sales is debatable.

Thats kinda why I was thinking it might be time for nvidia to lay in bed with intel. Once Apple moves to the i5 and i7's nvidia will even lose more of it's reveniew. The match kinda seems like it would be a win/win for both. Intel gets greenlight on IGP nvidia gets some kind of CPU capabilities in return. The 2 combined would dominate the market.

 

[Mr. Pedantic]

General musing:
I'm not sure how many other nanoscience savvy members we have on board but I always thought the Fermi name was almost ironic. Coming from the term Fermi energy, which grossly overly simplified, deals with the energies of promoting electrons from their valance band, past the band gap and up to the conduction bands. Even more simply, the energy needed to turn a semiconductor from an insulator to a conductor. The Fermi energy, in many cases, actually resides at or just below the top of the valance band. Viewing Fermi as the nonconducting (not currently doing any work...ok that's a bit of stretch, as gates DO need to be off too, but stay with me here on this) state of the architecture and refresh is given the extra oomph required to push things up to the conduction band and get interesting. Now that I've cemented myself as the nerdiest of the nerds I'll just shut up

My thoughts too. I doubt Fermi itself will do very well at all, but I think the architecture has potential, and will lead NVidia into a very interesting and exciting future.

 

[RussianSensation]

Fermi isnt flopping, much to Charlie's dismay.

It will be the fastest chip, and they can only screw it up by pulling a GT200 and pricing too high at launch.

Either way, there is a ton of enthusiast $$$ on the sideline waiting for the 480. Just looking at the EVGA box art got my Debit Card trembling...

If 480 does not outperform 5870 by at least 20%, I guess we will know who the NV fanboys are on this forum [who have been patiently waiting for 6 months for performance that is superior to 5870, not just 5-10% faster].

 

[OCGuy]

If 480 does not outperform 5870 by at least 20%, I guess we will know who the NV fanboys are on this forum [who have been patiently waiting for 6 months for performance that is superior to 5870, not just 5-10% faster].

So if it is 19%, then anyone who has it is a fanboy, and if it is 21%, then it is acceptable. And this is all without knowing pricing.


Just want to be sure my notes are correct here.

 

[OCGuy]

I'm not saying it'll take a year from now for them to be the performance leader, I meant it as a year from now when we look at the available cards Nvidia will probably have the faster part.

But when Fermi launches, if the rumors of 5000-10000 parts are true, even with moderate demand they'll likely sell out. Once it launches and the initial launch parts are gone, I bet it'll be pretty hard to find one for a while... they'll trickle in slowly. If that happens, can we say that Nvidia has the faster GPU when it's nearly impossible to find?

I know you strongly prefer Nvidia, so I wouldn't count on you getting a Radeon setup no matter how Fermi performs or costs. I imagine dual GTX280's is still pretty capable anyway (I thought that's what you had anyway?)

Well according to your theory, the 5970 does not exist then, because it is impossible to find.

Yes I am still using the 2X280s. 5870 does not offer anything above that except for DX11, which just is not that important to me due to the current game selection. It isnt going to magically become important to me when 480 is released either.

I was an early adopter last round, so I think I am going to sit back and wait for the refresh at the end of the year. Unless the pointless-upgrade bug strikes

 

[SlowSpyder]

Well according to your theory, the 5970 does not exist then, because it is impossible to find.

Yes I am still using the 2X280s. 5870 does not offer anything above that except for DX11, which just is not that important to me due to the current game selection. It isnt going to magically become important to me when 480 is released either.

I was an early adopter last round, so I think I am going to sit back and wait for the refresh at the end of the year. Unless the pointless-upgrade bug strikes

Hard to find/not at Newegg is different than impossible to find. Going to Google shopping I found a few places that have them in stock. If you think the 5970 is hard to find, I'm willing to bet Fermi will make it look plentiful in comparrison. Just my guess... we'll see soon enough.

 

[Munky]

If Fermi flops, Nvidia will crank up the marketing effort, and the fanboys will still rave about it while holding out for "how Fermi should have been done."

 

[RussianSensation]

So if it is 19%, then anyone who has it is a fanboy, and if it is 21%, then it is acceptable. And this is all without knowing pricing.

Just want to be sure my notes are correct here.

Let me guess you are an accountant? Or a micro-manager?

Think about it, every 12-15 months, videocard performance increases 50-70%+, often even more (just look at historical releases of videocards at gpureview.com).

If NV is 6 months late to the table, (for those who waited) it can't be just marginally quicker (i.e., <10%). Most importantly consider the efficiency of the architecture if it does not outperform 5870 by a significant margin: 500 mm^2 die on 384-bit memory bus (expensive) vs. 334 mm^2 die for 5870 on 256-bit memory bus (cheaper).

When 5850 launched, its MSRP was supposed to be $259. Thanks to NV, most 5850s currently hover at $300+; that's 6 months after launch. Back when there was little competition in 2008, NV released GTX280 for $649 (in June of 2008: http://www.anandtech.com/video/showdoc.aspx?i=3334). ATI responded with a card which offered 80% of the performance for less than half the price:
http://www.anandtech.com/video/showdoc.aspx?i=3341

Further, in April of 2009, ATI launched 4890 series for just $250-260. By June of 2009, you were able to purchase a 4890 ,which offered performance within 10% of GTX280, for $180-$200. In exactly 12 months from GT200's launch we were able to purchase 90% of the performance of a $600 graphics card for 3x less. That's progress.

Fast forward to today. If NV simply releases GTX480 at $399 with similar performance to ATI, that's not progress. The card has to be faster at the same price or cheaper at a level of performance. Based on the complexity of the design, it most likely will not be cheaper. Therefore, it is expected to be faster.

 

[NoQuarter]

Think about it, every 12-15 months, videocard performance increases 50-70&#37;+, often even more (just look at historical releases of videocards at gpureview.com).

That's why it wasn't worth the wait even if it's 20%. At 20% over 6 months it's not even on par with the schedule you set out, which means you didn't get some big performance advantage by waiting. If waiting 2 months got you 20% it woulda been great, but now there was no reason not to be using a 5870 this whole time.

 

[sxr7171]

The same that happened to Intel after P4 (Williamette, Northwood etc.). They have talented engineers, they'll figure a way out.

It will suck to be us, because unlike AMD which at that time used the opportunity to gain marketshare the new ATI/AMD will price gouge us for another year.

 

[scooterlibby]

I guess we'll see at launch. My guess is it will be hard to find due to limited supply and high demand.

They'll be the single GPU performance leader much sooner than a year.

Will it be worth it? Who knows. That comes down to the MSRP and if the e-tailers take advantage of limited supply and raise prices on their own.

My instinct tells me to sit on my hands until Fall.

Sadly, I agree. I am looking for a single chip solution that beats my old school (192sp) 260's in SLI. Doesn't have to be by much, either and Fermi looks like it might do that (5870 runs slightly slower than two 260's). If Fermi comes out at MSRP, beats the 5870, and is available I'll hop on it. Or if the 5970 goes back down to MSRP I'd get that (even though it is dual chip, but I like it on one PCB). Those scenarios do not seem likely so waiting until Fall looks like a real possibility. I thought I'd have a new card by Winter 2009! So foolish.

 

[scooterlibby]

In regards to the OP - I think several people have hit it right on the money that a flagship flop is not a fatal blow to Nvidia. They make their money in many different areas and flagship is a small piece of that pie.

 

[evolucion8]

Serious Sam and Unreal Tournament? How many years ago was that? And if it were such a "hit" as you say, WHY didn't ATI keep pushing it??? How many years have they let go by without trying to push for that standard? Years man.

It was a hit, period. ATi's relationship at that time was mediocre at best, and it had only improved slighly. It didn't flopped like nVidia's N-patches implementation which was rather patthetic.

Thats kinda why I was thinking it might be time for nvidia to lay in bed with intel. Once Apple moves to the i5 and i7's nvidia will even lose more of it's reveniew. The match kinda seems like it would be a win/win for both. Intel gets greenlight on IGP nvidia gets some kind of CPU capabilities in return. The 2 combined would dominate the market.

And now that Apple is buying stocks of the HD 5750, it will only shrink their marketshare further.

Sadly, I agree. I am looking for a single chip solution that beats my old school (192sp) 260's in SLI. Doesn't have to be by much, either and Fermi looks like it might do that (5870 runs slightly slower than two 260's). If Fermi comes out at MSRP, beats the 5870, and is available I'll hop on it. Or if the 5970 goes back down to MSRP I'd get that (even though it is dual chip, but I like it on one PCB). Those scenarios do not seem likely so waiting until Fall looks like a real possibility. I thought I'd have a new card by Winter 2009! So foolish.

The GTX 260 and HD 4870 performance is great by themselves, so the only real upgrade would be to the HD 5870, and if you are running the former cards in multi-GPU configurations, there's no upgrade currently besides of the HD 5970, nice performance showdown of the previous generation of cards, specially with the crappy console ports that keep coming.