Quick Multichannel Question

[JimboG]

Hello there,



I wonder if I may ask a quick question regarding multi-channel configurations, thank you. Apologies if it's a bit noob!



Could a multichannel configuration be used to create a system where there were two machines (identical hardware specifications) linked together to share the load for rendering for a single display output on the master machine?



In other words, without wanting to set up a multi-display system (at this time) is there any performance benefit with using two machines together for a single display, connected to the master?



I am trying to weigh up the relative costs involved with having a very expensive single machine with mutliple CPUs/Cores Vs two less expensive machines in Mutlichannel configuration sharing the load?



Given that the Xeon E5s are just around the corner, and the EVGA SR-X arriving onto the market with it I wonder if a machine with a couple of E5-2690's and a couple of GTX690's would outperform say two machines running with an i7 3960X and whatever graphics card would sit in the master (let's say a GTX690 or two in SLI) - in the context of P3D.



There would be arguments for both but it would be good to get someone else's opinion on this forum as to what benefit if any using a Multichannel configuration would offer in this scenario?



Thanks very much for your advice.



Kind regards,



James Gillies

[jimcooper1]

Quote:
Quote from JimboG on April 29, 2012, 09:07

........Could a multichannel configuration be used to create a system where there were two machines (identical hardware specifications) linked together to share the load for rendering for a single display output on the master machine



No

[JimboG]

Thanks Jim.



Kind regards,



James

[jimcooper1]

James,

The best advice I can offer is choose the best MB & CPU config you can afford together with SSD to speed up data access. GPUs are not so critical. Bear in mind that P3D is a 32-bit application and that too much Memory can cause it problems as it will not be able to internally address more than 4GB. Consequently GPUs with high memory may actually overload the application. The P3D team are working on optimising the load sharing so things will improve. Consequently get the best you can afford but, for now, you may need to limit the GPU memory in the config file if you find you get OOM errors. Once DX11 is implemented and CPU load sharing is improved your P3D will get faster on a high-end PC.



Regards



Jim

[JimboG]

Hi Jim,



Thanks very much for this, some good advice - duly noted. It's certainly an exciting time where we can talk about how we might need to account for the hardware being "too powerful" for the software!



I look forward to seeing how the optimisations are introduced by the P3D team.



Thanks again for your time to answer my question.



KInd regards,



James