The fall and rise of Fermi and nVidia.

[1h4x4s3x]

Try reading carefully. I never said I saw the end result.

I'm just trying to gather information that you obviously have seen or read from a knowledgeable source. So, why not share it with us?
Or how do you come to the conclusion that the image quality will be on par with CPU rendered ones?

How did my logic fail?

I repeat.
Nvidia isn't going to compete with consumer cards against their workstation cards.

Did I mention that it's still all good and perfect? [..] Don't imply something other than that.

I didn't imply anything. All I did was pointing out your faulty logic.

"Keep it mind there's limitation set on GTX 285: only 3 tracks on timeline are CUDA supported. Fermit will, I bet, be supported fully."

I explained to you why GTX 285 is limited and why likely the same will happen to the GTX 480/470.

Fermi is not a name of a consumer level GPU. Rather, it is a code name for
"The Next Generation CUDA Architecture."

Fermi is the code name for the GF100. It's the architecture used in all the GF 400 cards.

Again, you fail at implying. You see, I never said CS5 will fully support consumer level cards. I said 'CS5 will support Fermi fully.'

That's what you said.

Before you accuse me of stuffs I never said, think about your own limitations and reading comprehension problems. You succeeding on those problems? "I don't think this is going to happen."

I'm not quite sure why you're so aggressive in your tone but you may want to calm down a bit.

 

[Qbah]

Looking here we can see that as far as capabilities goes:

Quadro FX 3800 / 4800 ~ GTX260 with less memory bandwidth
Quadro FX 5800 ~ GTX280

So as far as what the GPU can do, it's a purely software/driver limitation that the GTX260/GTX275/GTX280 cards aren't supported. I'm surprised that the GTX280 is not supported as it's basically identical to the GTX285. Looks like the touted CUDA-support in Adobe's CS5 is not available for the average Joe... yet. I wouldn't hold my breath for those cards either (not to mention all the people running GTS250 and lower). Though I'm pretty sure they will at least add the GTX480 at some time. Probably with more artificial limitations, but oh well... They need to make people buy Quadro FX cards somehow

 

[WaitingForNehalem]

The 5800 Ultra was competitive with the 9700 Pro in a lot of games actually. Much later on when DX9 games started coming out, the gap started to grow though. Here's Anand's review of it when it launched: http://www.anandtech.com/show/1062/6

After reading that review, you realize how similar Fermi is to the FX 5800 series.

The most stunning part of the GeForce FX 5800 Ultra is its utterly insane cooling system and the first thought that popped into our minds when we saw it was: noise. With the GeForce FX, NVIDIA borrowed some of their mobile technology to help minimize the amount of noise their cooling system would produce.

​

So there you have it, NVIDIA's response to ATI's Radeon 9700 Pro - but does anyone else feel unfulfilled by the GeForce FX? A card that is several months late, that is able to outperform the Radeon 9700 Pro by 10% at best but in most cases manages to fall behind by a factor much greater than that. Granted the problems that plagued the launch of the FX weren't all up to NVIDIA's control, after all the decision to go 0.13-micron was made 1 - 2 years ago based on data that was available at the time. ATI took a gamble on producing a 0.15-micron part and NVIDIA did the same on their 0.13-micron NV30, and it looks like ATI guessed right.

 

[nitromullet]

After reading that review, you realize how similar Fermi is to the FX 5800 series.

So there you have it, NVIDIA's response to ATI's Radeon 9700 Pro - but does anyone else feel unfulfilled by the GeForce FX? A card that is several months late, that is able to outperform the Radeon 9700 Pro by 10% at best but in most cases manages to fall behind by a factor much greater than that. Granted the problems that plagued the launch of the FX weren't all up to NVIDIA's control, after all the decision to go 0.13-micron was made 1 - 2 years ago based on data that was available at the time. ATI took a gamble on producing a 0.15-micron part and NVIDIA did the same on their 0.13-micron NV30, and it looks like ATI guessed right.

Why not look at the entire statement instead of just selectively bolding the part that supports your argument?

A card that is several months late, that is able to outperform the Radeon 9700 Pro by 10% at best but in most cases manages to fall behind by a factor much greater than that.

This is where 5800 Ultra and GTX 480 differ... The GTX 480 beats the 5870 by 10-15% almost across the board, and when it doesn't it only loses by a little bit. Take min frames into consideration, and chances are the GTX 480 offers a smoother gaming experience overall.

Generally speaking, GTX 480 >= 5870, while 5800 Ultra <= 9700 Pro. Do you see the difference?

It's all right here:

I'd liken the GTX 480 more to the X1800XT than to the 5800 Ultra or 2900XT.

 

[WaitingForNehalem]

I was doing it out of humor, I know the GTX 480 is the faster card.

 

[nitromullet]

I was doing it out of humor, I know the GTX 480 is the faster card.

lol, next time try using "/end sarcasm"

 

[SlowSpyder]

OP, when most people here post in regards to being somewhat disappointed with Fermi, they're doing so while looking at it from a gaming stand point. That is what the GTX4x0 cards are meant to do well first and foremost afterall. When compared to the competition many can't help but feel a bit underwhelmed by the performance when you factor in the heat/power/noise/how long it took to get to market compared to AMD.

On the other hand, I think it's generally accepted that Nvidia is well ahead of AMD in regards to GPGPC. If you were to build a system (or systems) for similar work based on CPU's to get the equvilent performance of a single Fermi based system, well then Fermi looks like a shining star. The amount of money it'd take in CPU's, the amount of power, the amount of heat created to get an equal amount of work done would probably make Fermi look very, very effieienct and cost effective. I don't think anyone denies that. But the GTX2x0 were also very good at a lot of GPGPC tasks, but did it while being competitive with heat/power/noise and launching very near when AMD launched their first Radeon 48xx cards (Nvidia was actually out a week or so earlier).

So while Fermi is going to give next generation GPGPC performance and still be much more efficient than CPU's for many tasks, it leaves a bit to be desired in some ways from the gaming perspective. It seems that AMD's current generation and Nvidia's prior generation of parts were much more well rounded than Fermi.

 

[WaitingForNehalem]

lol, next time try using "/end sarcasm"

Well I wasn't being sarcastic. I just thought it was funny how similar the situation is with the exception of the GTX 480 not being slower. That is why I did not bold that part.

 

[EarthwormJim]

The 5800 Ultra was competitive with the 9700 Pro in a lot of games actually. Much later on when DX9 games started coming out, the gap started to grow though. Here's Anand's review of it when it launched: http://www.anandtech.com/show/1062/6

It wasn't just DX9 games that separated the two. Initially when the 9700pro and even by the time the 5800ultra came out, Ati still had immature drivers.

Once Ati got their act together as far as drivers go, the 5800ultra wasn't even close to the 9700pro. Nvidia historically has had better drivers even on launch day than Ati did.

The 5800ultra was really a 4 pixel pipeline device trying to compete with an 8 pixel pipeline 9700pro.


That really was a great card, the 9700pro. It held it's value too. I bought it second hand and ended up selling it for more than I payed for it.

 

[ronnn]

Fermi certainly isn't dead. Likely the 470/80 are paper place holders, but working variations with new numbers should be on the way.

 

[OCGuy]

OP, when most people here post in regards to being somewhat disappointed with Fermi, they're doing so while looking at it from a gaming stand point. That is what the GTX4x0 cards are meant to do well first and foremost afterall. When compared to the competition many can't help but feel a bit underwhelmed by the performance when you factor in the heat/power/noise/how long it took to get to market compared to AMD.

On the other hand, I think it's generally accepted that Nvidia is well ahead of AMD in regards to GPGPC. If you were to build a system (or systems) for similar work based on CPU's to get the equvilent performance of a single Fermi based system, well then Fermi looks like a shining star. The amount of money it'd take in CPU's, the amount of power, the amount of heat created to get an equal amount of work done would probably make Fermi look very, very effieienct and cost effective. I don't think anyone denies that. But the GTX2x0 were also very good at a lot of GPGPC tasks, but did it while being competitive with heat/power/noise and launching very near when AMD launched their first Radeon 48xx cards (Nvidia was actually out a week or so earlier).

So while Fermi is going to give next generation GPGPC performance and still be much more efficient than CPU's for many tasks, it leaves a bit to be desired in some ways from the gaming perspective. It seems that AMD's current generation and Nvidia's prior generation of parts were much more well rounded than Fermi.

Hmmm....

A well-reasoned, thought out post.

 

[mmnno]

Though Adobe CS5 is not officially out as of now, I saw a brief benchmark that was done on the beta version. With its native CUDA support, encoding a H.264 clip that took 45 minutes without CUDA support, took only 4 minutes. Unlike before, this was done without visual penalties and limitations. Needless to say, it is a dramatic difference. Even better, the test was done on GTX 285. Considering how Fermi is desinged to perform much better than GTX 285 in terms of CUDA performance, I can only wonder what Fermi can do.

Doesn't really matter what it can do, unless it can go from 4 minutes to negative 41 minutes. If the 285 is already so low, how mindblowing would it be if the 480 cut that in half? You saved a whole 2 minutes? If that kind of time is worth it, you may be in the market for a Quadro anyway.

 

[Cattykit]

Doesn't really matter what it can do, unless it can go from 4 minutes to negative 41 minutes. If the 285 is already so low, how mindblowing would it be if the 480 cut that in half? You saved a whole 2 minutes? If that kind of time is worth it, you may be in the market for a Quadro anyway.

2 min. of improvement may seem not much of a dea in this casel. However, don't forget they add up. Those 2 min. may extend into 20 minutes, 2 hours, and so on.

Micro seconds of delay in timeline is a huge annoyance. In cases where 285 does not have enough horse power to elimitate such delay, 480 should be able to pull it off.

 

[Genx87]

Doesn't really matter what it can do, unless it can go from 4 minutes to negative 41 minutes. If the 285 is already so low, how mindblowing would it be if the 480 cut that in half? You saved a whole 2 minutes? If that kind of time is worth it, you may be in the market for a Quadro anyway.

It is a percentage game. Cut the work in half and you double output. Doubling output means increased revenue.

 

[RussianSensation]

September 22, 2009: HD5xxx series launch
NV: ?

December 18, 2009: 800,000 HD series cards shipped.
http://www.xbitlabs.com/news/video/...of_DirectX_11_Graphics_Chips_Shipped_ATI.html
NV: ATI's lead is insignificant.

January 7, 2010: ATI ships over 2 million HD series cards.
http://www.xbitlabs.com/news/video/..._Millionth_DirectX_11_Graphics_Processor.html
NV: ?

April 15, 2010: ATI ships 6 million DX11 graphics chips
http://www.xbitlabs.com/news/video/...Graphics_Chips_Shipped_by_ATI_CEO_of_AMD.html
NV: ?

So Fermi is not exactly a FLOP from a performance perspective (but noise and power are a problem). If you want the fastest single GPU, GTX480 is it.

What Fermi is though is a FLOP from a business perspective (late to the market means it lost a lot of potential sales, too expensive to manufacture due to die chip being too large which reduces yields and gross margins, will likely continue to be low selling, esp. since there will be no offerings < $200 for a while, and bloated architecture that is not a good base to start from for GTX5xx series.)

 

[zod96]

The only people that will buy a fermi card are die hard nvidia fans. People that could careless about ati or nvidia but want the best bang for their buck I'm betting would pick up a 5870/5850 over any fermi card right now...

 

[NoQuarter]

Doesn't really matter what it can do, unless it can go from 4 minutes to negative 41 minutes. If the 285 is already so low, how mindblowing would it be if the 480 cut that in half? You saved a whole 2 minutes? If that kind of time is worth it, you may be in the market for a Quadro anyway.

It is a percentage game. Cut the work in half and you double output. Doubling output means increased revenue.

I think that's why he said you may be in the market for a Quadro anyway. From a consumer standpoint 4min vs 2min isn't a big deal, from a business standpoint it's huge.. but you'd also be buying the Quadro version and not the consumer 480 version.

 

[Cattykit]

I think that's why he said you may be in the market for a Quadro anyway. From a consumer standpoint 4min vs 2min isn't a big deal, from a business standpoint it's huge.. but you'd also be buying the Quadro version and not the consumer 480 version.

True, but don't forget there are those who fall somewhere in between consumer and professional categories.

 

[EarthwormJim]

The only people that will buy a fermi card are die hard nvidia fans. People that could careless about ati or nvidia but want the best bang for their buck I'm betting would pick up a 5870/5850 over any fermi card right now...

If you factor in overclocking, which some people do, I'd say Fermi is a better value. You'd really have to water cool Fermi though or else you'll suffer through unbearable fan noise.

 

[Stoneburner]

Fermi is not the 5800

Fermi is Prescott.

 

[BD231]

Actually Fermi is Presquatt :sneaky:

 

[hpmoon]

You realize that 3 tracks are rather useless for people who do it professionally?

You couldn't have been more obvious in revealing that you are not a professional, and you know nothing about professional video editing.

With a GTX 285, three OF the video tracks in a sequence will have CUDA hardware acceleration. Your guesstimation of what professional video editors actually work with is probably based on "ooh-aahing" control rooms with banks of monitors, imagining that any production involves switching between numerous "feeds." Nonsense. Adobe Premiere Pro is for independent filmmakers much more than Hollywood A-listers -- you do realize that, right? We do NOT average much more than two or three video tracks per project.

 

[Madcatatlas]

You couldn't have been more obvious in revealing that you are not a professional, and you know nothing about professional video editing.

With a GTX 285, three OF the video tracks in a sequence will have CUDA hardware acceleration. Your guesstimation of what professional video editors actually work with is probably based on "ooh-aahing" control rooms with banks of monitors, imagining that any production involves switching between numerous "feeds." Nonsense. Adobe Premiere Pro is for independent filmmakers much more than Hollywood A-listers -- you do realize that, right? We do NOT average much more than two or three video tracks per project.

Welcome to the forums! I hope youll enlighten us with your knowledge in the times ahead!

 

[jimmyparky]

I agree, who cares of delay.

 

[SlowSpyder]

I agree, who cares of delay.

The 6+ million people who were looking to buy between September 23rd 2009 and April 2010.