iray benchmark

It is only 3 cards, running at x8 speed, instead of x16 speed. I wish I had 24 GPUs, but alas it is only 3.

-Eric

We tested the iRay example in Max 2012 (we dont have 2011 installed on the new 12 core machine).

100 passes = 11 seconds GPU only

10 minute render. We stopped it after about 4 minutes as the cards were getting warm (97C) and we didn't want to "let any smoke out" of them.
At 4 minutes worth of render the image looked pretty good.

David

And what GPU (and how many) is being used in that system? That's over twice as fast as my system even with CPU.

Don't use CPU, if you have a capable GPU. It is actually slower to use GPU+CPU, than just GPU.

David is using Dual 590 GTX.

-Eric

Just tested on my machine in 2012 and it appears that 2012 is faster than 2011 iray. It appears that there is about a 16% increase in speed over 2011.

On the temperature of cards they peaked at 90°C, but held at 87°C once the fans spun up.

-Eric

can someone second:
the scene render here in full view 1min20sec but a region render of just 30% do also render 1min20sec
a crop of 10% aswell 1min20
it not matters what region or region size i choose - the rendertime is the very same

another iray file I tested have the same "problem" so I wonder what the matter is ?!

iray always renders the full frame buffer and crops afterwards. So use zoom, pan, and output size to render crop/regions.

-Eric

bad - there is no hidden script command or variable to avoid that ?
a zoom region pan way dosnt sounds reliable

I was reading on the Nvidia forum regarding the GTX-590:

"Correct. More important for CUDA is that these two GPUs are directly connected via a NF200 PCI-Express switch on the card. Once the second release candidate of CUDA 4.0 comes out, the two GPUs should be able to directly copy memory between each other without loading the external PCI-Express bus. This could have useful application to large multi-GPU algorithms that needs to sync data between devices periodically."

http://forums.nvidia.com/index.php?showtopic=196269

Thoughts?
(or course my hope is they will figure out a way to use all available card RAM)..
David

100 iterations: 16 seg (gpu+cpu)
100 iterations: 19 seg (gpu only)

10 min. render
here

cpu: i7 980x

The link in the first post fails, can someone post that file?


I got it, thx anyway...

Dual Quad XEON E5530 @ 2.4GHz (16 virtual cores)
Cubix GPU Xpander
- Quadro FX 5000
- 3 x Tesla 2050
12GB RAM

Temps: 175f / 79.4c (This is usually lower but we are having to cut back power consumption due to the high heatwave, my office is close to 80f today and outside reached 107f / 41.6c!)

100 iterations @ 9 seconds

10 minute render ( 6043 iterations ):

maybe iray needs an update soon!

Wow, with 9 seconds you win this thread... I think rendering with that machine is what is causing the heatwave!

actually i didnt understand yet.
if i select ' time 10 min then no chance to select iterations(passes) =100.
what i didnt understand?.