PDA

View Full Version : Laptop selection



ndingler
10-08-2012, 03:44 PM
I tried hijacking someone else's thread, but haven't seen any replies yet, so I thought I'd start a new thread...

I am shopping a new laptop and I am wondering about the GTX 680M

I could get a dell Precision M6600 with a 4G Quadro 5010, 384 Cuda Cores
http://www.nvidia.com/content/PDF/product-comparison/Product-Comparison-Quadro-mobile-series.pdf

or for about the same price an alienware with dual 2G GTX 680Ms SLI linked together.
Each GTX 680M has 2G memory, 1344 CUDA Cores
http://www.geforce.com/hardware/notebook-gpus/geforce-gtx-680m/specifications

Overall price for each is fairly comparable at around $3500 similar processors / memory / drives, etc.

Any recommendations?


Anyone have experience with the Alienware / gaming computers being used for a CAD / Rendering machine, or the gtx680m cards? also wondering if Alias, ProE, Solidworks will play nicely with the alienware.

Thanks for any input!

david.randle
10-08-2012, 11:30 PM
The first thing you should consider is what kind of data you're dealing with. Our products support about 5 million polygons per GB of VRAM. This alone may force you down the Quadro path.
In terms of performance, the dual GTX680m's should be quite a bit faster than the Quadro (more than twice as fast), however, iray isn't fully optimized for Kepler yet. So, that kind of performance may take some time to achieve. I"m not sure about Pro/E but Alias and SW will probably be just fine on GeForce. Also, if you go down the GeForce route, you will want to disable SLI as that hurts rendering performance in iray. It will still use both GPU's by default without SLI on.

I hope this helps,

Dave

ndingler
10-09-2012, 10:37 PM
Great!
Thanks for the input!
Now you said 5M polys per G of Vram... the 5010 has 4G and the other system has 2 2G cards... while they have the same amount overall, do the GTX cards 2G per card bottleneck the whole system... or would it be a wash vram wise?

andy
10-10-2012, 12:30 PM
The 2GB would be the bottleneck if you reach it. If you use engineering data, plan on reaching that pretty quick if you don't remove B-side surfaces before rendering.

david.randle
10-12-2012, 06:54 AM
Also, remember that VRAM is not additive...so although you have 2 x 2GB cards, that doesn't mean iray sees 4GB. It still sees and utilizes 2 as the data gets loaded into each cards memory.