nV will demo Fermi based Tesla on Dec.17

jvroig

Platinum Member
Nov 4, 2009
2,394
1
81
While I'm sure Tesla is great and all, I am of the impression that it is quite different from the GeForce series. If this new generation Tesla proves to be 2x faster (for example) than the previous one, could performance from there be used to extrapolate Fermi GeForce performance? (I'm just throwing it out there, so to speak, but I personally believe it won't be related at all, except for the very generic "Yes, naturally it will be faster than previous gen")
 

Lonyo

Lifer
Aug 10, 2002
21,939
6
81
While I'm sure Tesla is great and all, I am of the impression that it is quite different from the GeForce series. If this new generation Tesla proves to be 2x faster (for example) than the previous one, could performance from there be used to extrapolate Fermi GeForce performance? (I'm just throwing it out there, so to speak, but I personally believe it won't be related at all, except for the very generic "Yes, naturally it will be faster than previous gen")

You can't extrapolate Fermi results to GPU performance because Fermi has been developed to make up for the shortfalls in HPC areas that previous cards have. I wouldn't be surprised to see more than double the performance, assuming a doubling of execution units and similar clockspeeds.

Specifically:
http://www.anandtech.com/video/showdoc.aspx?i=3651&p=3

It wouldn't be surprising to see pretty massive improvements in some areas when it comes to Tesla applications. That doesn't necessarily mean much for 3D gaming though.
 

Daedalus685

Golden Member
Nov 12, 2009
1,386
1
0
While I'm sure Tesla is great and all, I am of the impression that it is quite different from the GeForce series. If this new generation Tesla proves to be 2x faster (for example) than the previous one, could performance from there be used to extrapolate Fermi GeForce performance? (I'm just throwing it out there, so to speak, but I personally believe it won't be related at all, except for the very generic "Yes, naturally it will be faster than previous gen")

I'd assume they are going ot be showing off a lot of the improved DP performance. This won't make a lick of difference in gaming. We could see an order of magnitude increase clock for clock on some tesla apps. If we assume the Tesla and Geforce are the same core and use a number crunching program very heavey on SP we could perhaps guess at the performance icnrease from a 280 to a 380 very roughly (no better than we can from the spec icrease). But you are correct to say that it means little to nothing.

Tesla usually uses the exact same cores as the gaming counterparts. I'm not sure if that is the case here or not. They won't be much different but will probably be binned differently, and obviousy cost a LOT more.
 
Last edited:

Creig

Diamond Member
Oct 9, 1999
5,171
13
81
"Now, I have one right here. This, Ladies and Gentlemen, this puppy here is Tesla" ;)
 

Idontcare

Elite Member
Oct 10, 1999
21,118
58
91
I thought they had already done this. :hmm:

They showed a presentation containing results that had been previously generated with Fermi hardware, the output was captured and then "replayed" on a non-Fermi system for demonstration purposes.

I suspect this time around the demo will actually run "live" on Fermi hardware.
 

ronnn

Diamond Member
May 22, 2003
3,918
0
71
Hope they sneak in a dx11 benchmark. Faster compute is only for smart people and not sure if I qualify.
 

Genx87

Lifer
Apr 8, 2002
41,095
513
126
I'd assume they are going ot be showing off a lot of the improved DP performance. This won't make a lick of difference in gaming. We could see an order of magnitude increase clock for clock on some tesla apps. If we assume the Tesla and Geforce are the same core and use a number crunching program very heavey on SP we could perhaps guess at the performance icnrease from a 280 to a 380 very roughly (no better than we can from the spec icrease). But you are correct to say that it means little to nothing.

Tesla usually uses the exact same cores as the gaming counterparts. I'm not sure if that is the case here or not. They won't be much different but will probably be binned differently, and obviousy cost a LOT more.

My guess is they will be just like their quadro line. The same GPU but the driver exposes functionality you dont get on the gaming side. Of course the memory configurations and speeds and interconnects will most likely be tighter as well. I think the rumor is Tesla GPUs come with 4GB of local memory each.
 

PlasmaBomb

Lifer
Nov 19, 2004
11,815
2
81
It was on. There was a thread at XS about it.

There are a couple of threads.

Did you mean this one?
http://www.xtremesystems.org/forums/showthread.php?t=241120

Edit: Actual details!

http://www.techpowerup.com/104942/NVIDIA_GT300__Fermi__Detailed.html

* Transistor count of over 3 billion
* Built on the 40 nm TSMC process
* 512 shader processors (which NVIDIA may refer to as "CUDA cores")
* 32 cores per core cluster
* 384-bit GDDR5 memory interface
* 1 MB L1 cache memory, 768 KB L2 unified cache memory
* Up to 6 GB of total memory, 1.5 GB can be expected for the consumer graphics variant
* Half Speed IEEE 754 Double Precision floating point
* Native support for execution of C (CUDA), C++, Fortran, support for DirectCompute 11, DirectX 11, OpenGL 3.1, and OpenCL

^Which was pulled from BSN, which was pulled from the table linked in the other thread, which was pulled from wikipedia and changed slightly, which was pulled from random forum rumours.
randomizer
Xtreme Enthusiast

77a.jpg


^Look see it is real!
 
Last edited:

thilanliyan

Lifer
Jun 21, 2005
11,848
2,051
126
One of the threads had questions that had been asked and answered but I think all the Fermi threads were merged into that one.