AE CS6 11.0.1 CUDA BENCHMARK PROJECT – test your graphics cards!

[John Cuevas]

As I’m deciding on my new system(and video card) I’d be very interested in seeing the results.

Johnny Cuevas, Editor
Thinkck.com

“I have not failed 700 times. I have succeeded in proving that those 700 ways will not work. When I have eliminated the ways that will not work, I will find the way that will work.”
—THOMAS EDISON on inventing the light bulb.

[Juan Salvo]

We’ve seen the same results in Resolve, where the 580 is on par or slightly ahead with regards to CUDA processing. It has to do with a shift to a software task processor in the Kepler systems. The 690 may out perform the 580. But ofcourse so would the 590.

Hopefully the Kepler based Quadros will have a hardware task controller and improved CUDA performance.

Online Editor | Colorist | Post Super | VFX Artist | BD Author

https://JuanSalvo.com

[Erik Mickelson]

Nope, not happening on my mac.

I stopped the render after 15 minutes. Time remaining was over 7 hours. Obviously there is zero Open CL support(cannot select GPU option). Specs are in my sig.

CrippleBook Pro 2.3Ghz i7, 16GB ram, Lion 10.7.4, FCPStudio 3, Adobe CS6. QT 7

[Mathew Lear]

Hey guys, just tried this on my mac and getting slightly faster speeds with my GTX 480 5 minutes 41 seconds – Had to make a few edits to supported cards list as shown here https://www.tonymacx86.com/viewtopic.php?f=15&t=62702&start=0

[Ben Jones]

System 1:
3930K (stock) (no idea why it is, it must have reset itself)
470 GTX (448 CUDA cores)
8min 17 sec

System 2:
2 x 8 Core Xeon E5 Sandy Bridge @ 3.3GHz (16 Cores / 32 threads)
Quadro 4000 (256 CUDA cores)
17 minutes 24 sec

conclusion: Test seems completely processor independent
CUDA cores are key, Quadros are useless at raytracing!

Thanks Teddy… you just answered a question that has been bugging me for months.
Q: Are Quadros worth the money?
A: Not even slightly!

[Mathew Lear]

Quadros are definitely worth the money if your working in a engineering environment with multiple workstations and need support from nvidia…. But for home users and single user environments then geforce works just fine and in most cases better…..

[Mark Deherrera]

Question about your test machine. Did it use pci 3.0? Because i imagine that may give an edge to the 680 under those circumstances sicne the 680 supports 3.0 and the 580 does not. However, if it was PCI 2.o then this is still a valid benchmark for those with 2.0 slots on their motherboard.

[Mathew Lear]

PCIExpress 2.0 has more than enough bandwidth for a 680.. There would be no visible difference unless you were rendering with 3 of these cards……

[Mark Deherrera]

Good point, I was thinking it may make a difference but now that you said that I agree with you.

Also I ran the test on my quadro 2000 , soon to be replaced with a 680, and it rendered that test in 28min, 11sec

[Roy Foliente]

Is it the amount of RAM that makes a difference? I wonder how a Maximus combo with a Quadro card and Tesla C2075 with 6GB of RAM would fare?