I’m just back from NAB Show in Las Vegas. While the show is mainly focused on the broadcast industry, I noticed a few interesting news around cloud supercomputing. In fact, several companies are presenting solutions to process large amount of data within the cloud. The processing power may be managed by different kinds of processors like CPU, GPU or FPGA. The latter one is getting a lot of traction and seems a perfect solution for such kinds of applications. NGCodec, a promising startup located in Sunnyvale, California, was presenting an innovative VR solution where the computing is located in the cloud. NGCodec and Celeno have actually joined their forces to deliver a Wireless and Cloud Virtual Reality (VR) solution using NGCodec’s RealityCodec™, a low delay H.265/HEVC encoder IP core, paired with Celeno’s 5GHz 802.11ac WiFi. RealityCodec enables wireless and cloud VR by replacing the now-ubiquitous HDMI cable with a video encoder to transmit video via WiFi. Celeno WiFi’s dynamic channel-switching capabilities circumvent problems that arise when the spectrum becomes busy. Together, the result is 0.4% (250:1) video compression without the use of HDMI cables and with no perceived latency. NGCodec enables an untethered VR experience, which delivers higher fidelity and longer battery life than traditional mobile VR. When low latency 4.5/5G and fiber networks are available, the same solution can be deployed to deliver a Cloud VR/AR/MR experience. A FPGA is used in the cloud to encode the signal, and another FPGA is used at the reception site to decode the signal and then coupled with the VR glasses. This technology is supported by Amazon Web Services which is a pro fan of FPGA capabilities in terms of supercomputing. NGCodec is a member of the Origami Ecosystem launched by the Belgian startup Image Matters, in which I’m involved. Image Matters develops and commercializes electronic modules based on FPGA. While most applications are in the field of video processing for various markets (broadcast, medical, earth observation, …), Image Matters modular technology has also numerous opportunities in cloud supercomputing.
I will follow up the developments in that area and will provide updates in future blogs.
