Here's some footage we took at NAB2014. We showcased our High Performance Computing (HPC) Appliances. For more information about the High Density Compute Accelerator (HDCA), click here. For more information about the Flash Storage Array (FSA), click here. For more information about The CUBEs, click here.
Jaan: How's it going?
Interviewer: Tell me about your product.
Jaan: So we do all things expansion. We do rackmount expansion and also desktop expansion using PCI Express and Thunderbolt as the interconnect.
Interviewer: So how much compute power does that have?
Jaan: This is the densest piece of equipment that we make. It's a high-density compute accelerator. It holds 16 double wide graphics cards or GPUs. And it has up to four PCI Express Gen 3 by 16 uplinks that can connect to an upstream server or up to four servers. It can do a couple different types of configurations. One of the more popular ones if you notice in the back here has four different cable inputs so this is full PCI Express Gen 3 by 16. Each of these cable inputs can go to a different server and each server would have access to four GPUs. And you're basically getting four 4 to 1 configurations in one box. You could also hook up two servers and get two 8 to 1 configurations in one box. Or one server, for all 16.
Interviewer: What applications can you use with it?
Jaan: So applications we see now are things in data centers. High performance computing as far as seismic is concerned they like when the seismic get information from the ground to produce an image and find out where the oil is. They use GPUs to run their RTM algorithms, or their reverse time migration algorithms. So we see a real common one is having a server that runs 8 GPUs. And another server that runs 8 GPUs So, you're getting the two 8 to 1 configurations in one box. Medical imaging is another popular application where we're seeing this being used. We're experimenting here at the show to try to get it into post-production and rendering applications as well.
Interviewer: Great, thank you.