The official GT300/Fermi Thread

[tajoh111]

http://www.dailytech.com/article.aspx?newsid=16401

ORNL to Use NVIDIA Fermi to Build Next Gen Super Computer

NVIDIA announced its new Fermi architecture at its GPU Technology Conference recently. The new architecture was designed from the ground up to enable a new level of supercomputing using GPUs rather than CPUs. At the conference, Oak Ridge National Laboratory (ORNL) associate lab director for Computing and Computational Sciences, Jeff Nichols, announced that ORNL would be building a next generation supercomputer using the Fermi architecture.

The new supercomputer is expected to be ten times faster than today's fastest supercomputer. Nichols said that Fermi would enable substantial scientific breakthroughs that would have been impossible without the technology.

Looks like NV has customers already for Fermi.

[FischOderAal]

Looks like NV has customers already for Fermi.

Which should be due to the fact that GPGPU on ATI pretty much sucks. Well, I think I'm exaggerating here but NVIDIA clearly has a lead there.

[Farinorco]

Which should be due to the fact that GPGPU on ATI pretty much sucks. Well, I think I'm exaggerating here but NVIDIA clearly has a lead there.

At least talking about supporting technologies and so (the new C/C++ compiled code support, for example). Regarding computing power, I don't have it so clear.

The only example that I have been able to find comparing GPGPU performance in equality of conditions (that includes both running the same code, so it has to be a DirectCompute or OpenCL piece), the HD5870 is pulverizing the GTX285.

And the funniest part is that said example, is an NVIDIA demo of DirectCompute (run on both cards in AnandTech).

To be exact, this one:



But yes, I'm of the opinion that NVIDIA is one generation ahead when talking about GPGPUs technologies. We will see in terms of GPGPU performance when we have something else than this little demo to compare. I think it's a too specific program to make any conclusion from this.

[hurleybird]

But yes, I'm of the opinion that NVIDIA is one generation ahead when talking about GPGPUs technologies. We will see in terms of GPGPU performance when we have something else than this little demo to compare. I think it's a too specific program to make any conclusion from this.

I would say half generation instead of full generation, seeing as RV870 looks to be a better GPGPU than GT200, while GT300 looks to be a better GPGPU than RV870.

[To(V)bo Co(V)bo]

At least talking about supporting technologies and so (the new C/C++ compiled code support, for example). Regarding computing power, I don't have it so clear.

The only example that I have been able to find comparing GPGPU performance in equality of conditions (that includes both running the same code, so it has to be a DirectCompute or OpenCL piece), the HD5870 is pulverizing the GTX285.

And the funniest part is that said example, is an NVIDIA demo of DirectCompute (run on both cards in AnandTech).

To be exact, this one:



But yes, I'm of the opinion that NVIDIA is one generation ahead when talking about GPGPUs technologies. We will see in terms of GPGPU performance when we have something else than this little demo to compare. I think it's a too specific program to make any conclusion from this.

here too is a comparison to crunching on guru.

http://www.guru3d.com/article/radeon...review-test/25

[Farinorco]

Oh man, I hate when things turn to a discussion over "why ATi is right and NVIDIA not? You're ATi... You're NVIDIA... you only say that because of hatred... you only say that because of love"... it makes keep the thread so difficult...

I'd like to imagine for a moment that those names belong to a pair of companies that make products for us to buy, like Philips, LG, Woxter, :banana::banana::banana:or, Sony... not to some charismatic heroes of an epic world, or something.

here too is a comparison to crunching on guru.

http://www.guru3d.com/article/radeon...review-test/25

But that comparison is not really a useful one because they are not comparing the same applications / code in different hw, but different code on different hw (note that ATI cards are running code in ATi Stream while NVIDIA cards are running code in CUDA, so that means that the code they are running is different). That's similar to compare the Windows loading time with 2 ssd's (A and B) using Windows XP in ssd A and Windows Vista in ssd B.

1. I love how they leave out RV770, the real competition to GT200, and

2. That is single precision. GT300 will decimate RV870 in double precision.

They are testing HD5870, which is the reviewed card. They are using the most powerful NVIDIA solution, to compare. Which is exactly the point to include the RV770? About the GT300 decimating RV870 in double precision, we will see. That's too much fortune-telling for me. On the other hand I'm not all that interested in the applications that can get a benefit from double precision, since domestic applications and the kind of professional applications that may be useful to me, shouldn't do much about double precision floats.

[003]

At least talking about supporting technologies and so (the new C/C++ compiled code support, for example). Regarding computing power, I don't have it so clear.

The only example that I have been able to find comparing GPGPU performance in equality of conditions (that includes both running the same code, so it has to be a DirectCompute or OpenCL piece), the HD5870 is pulverizing the GTX285.

And the funniest part is that said example, is an NVIDIA demo of DirectCompute (run on both cards in AnandTech).

To be exact, this one:



But yes, I'm of the opinion that NVIDIA is one generation ahead when talking about GPGPUs technologies. We will see in terms of GPGPU performance when we have something else than this little demo to compare. I think it's a too specific program to make any conclusion from this.

1. I love how they leave out RV770, the real competition to GT200, and

2. That is single precision. GT300 will decimate RV870 in double precision.

[Helloworld_98]

1. I love how they leave out RV770, the real competition to GT200, and

2. That is single precision. GT300 will decimate RV870 in double precision.

they didn't use the RV770 because it doesn't have directcompute 5.0 which directly compares to CUDA, so it would be kinda unfair since RV770 doesn't have the same feature set.

[Xoulz]

1. I love how they leave out RV770, the real competition to GT200, and

2. That is single precision. GT300 will decimate RV870 in double precision.

I think you are a tad lost...!

The 5870 is out! It cost $379, which is $10 different than the GTX285... We will not see the GT300 for 4 months!

I hope the Nvidia GT300 will do something better, specially double precision, because thats the field in which Nvidia's moving their business model; Scientific Computing.... not games!

[003]

We will not see the GT300 for 4 months!

Try a month to a month and a half.

[tdream]

Try a month to a month and a half.

JHH wasn't even holding a real GF100 at the tesla launch. How in the name of god will that timeframe work?

[Xoulz]

Try a month to a month and a half.

Dude, are you paying attention to whats going on around you..? The GT300 will not be out for the Xmas holiday shopping season..

Nvidia concedes this, just read Anand's review of the Tesla architecture. ATi will have 3 months of un-fettered sales. When the GT300 is released, ATi will simply drop the prices on their current line-up and release (or announce) their X2 line, etc.. competing $ for $.



Do you really think Nvidia will be able to compete with Hemlock (5870x2) @ $599, (come this January or Febuary...) when Nvidia is due to release their GT300.

Or, a 5870 @ $299, or the GT300 for $499..? If Nvidia doesn't have a DX11 $199 part soon, they will loose massive market share and loose $Billions in stock. It doesn't matter how good the GT300 actually is, if only 40k uber-hardcore people ever buy it. The 5870, for the foreseeable future, is good enough for almost anyone needs.

Millions of people will be buying or upgrading their stuff for the Holiday Season. Nvidia will miss out on all those sales.


Timing is everything!

[Chumbucket843]

At least talking about supporting technologies and so (the new C/C++ compiled code support, for example). Regarding computing power, I don't have it so clear.

they win in programmability which is half the battle. usually the speed of a gpgpu app is compared to a cpu. here is an interesting article about implementing molecular dynamics on gpu's.http://www3.interscience.wiley.com/c...77402/PDFSTART it ends up being the exact opposite of the directcompute app with regards to speed.