Support Website Contact Support Blog

Dell C410x

We are looking to build a powerful imagery processing server.

We currently have two Dell PowerEdge R620 dual Xeon servers, each with 128gb of RAM.  Still many of our jobs take weeks to process.  We are hoping to increase the amount of data we collect and process next year.

This Dell PowerEdge R620 technical document, on pages 18/19, declares the R620’s compatibility with an external device called a “C410x”, which is essentially an external enclosure housing 16 nVidia Tesla m2070 cards.

Would any portion of the Pix4d imagery processing workflow benefit from the addition of a c410 device?  Are we better off purchasing addition R620 servers instead, if cost was not a factor?

Thank you.

This is question that spans this forum!

One fast, cheap way to increase processing density in house systems is to utilize GPUs.

I want to know my self what pros/cons are.

Second is the out sourcing of cores from on line resources.

Not sure how P4d folks feel about cloud licensing.

!?!?

Hello Joseph and Gary,

First of all, we have not tested such a set up, so we do not have any reported processing times to provide you. However, we believe that each machine will use 1-1.5 GPU for such a set up. So, depending on the number of projects you want to run simultaneously, you should decide the number of GPUs to mount on the C410x device.

Moreover, you can decide if it is better to buy another server or the C410x device based on the projects that you want to process. If you will have more projects in the future, then it might be better to have more processing power than GPU processing, since we only implement GPU processing during step 1 and during a part of step 2. If you plan to process bigger projects then maybe it is better to combine CPU processing with GPU processing.

Also note, that if you decide to go for the C410x device, you should consider purchasing GPUs with at least 4 GB of RAM (ideally 6 GB) . Moreover, the CPUs mounted on the server and their processing power will also have an impact on processing time.

Best regards,