PCoIP Ultra Technical Series Part 5: What About Bandwidth Consumption?

This is the fifth post in a technical blog series by Ian Main, Teradici's Technical Marketing Principal. In the series, he'll go through the PCoIP Ultra protocol enhancements from a technical perspective, and answer common questions that have come up since the release of the enhancements. 

And now without further delay, over to Ian!

PCoIP Ultra^TM Technical Series Part 5: What About Bandwidth Consumption?

So far, this technical blog series has explored the architecture of the PCoIP Ultra enhancements and how multi-core CPU scaling works for 4K/UHD video workloads. In this article, we’ll look at network bandwidth for video workloads at different quality settings as well as bandwidth efficiency compared to PCoIP-SSE4.2. Spoiler alert - some awesome improvements to see!

Using the same infrastructure as the CPU tests in the previous post, here is the network bandwidth consumption for Big Buck Bunny 1080p and 4K/UHD at 24 fps: 

Note the default image quality for PCoIP Ultra remains at Q80 as before, which approaches perceptually lossless in many cases, although discerning artists in the media and entertainment industry often select Q90 to eliminate any compression guesswork. 

This ability to scale image quality has always set the PCoIP protocol apart from alternative remote desktop technologies. With the Ultra enhancements, the quality can be now be dialed way beyond Q90 in a continuous fashion for remarkable reproduction where color accuracy and distortion-free playback are paramount. 

You can look forward to quality metrics in a later posting.

How does network bandwidth compare to traditional SSE4.2-based PCoIP encoding?

While I can’t repeat the 4K video test using SSE4.2 because AVX2 or NVENC is required to achieve 24 fps on a 4K/UHD display, I ran some benchmarks comparing a 24 fps workload on a standard 1920 x 1080 FHD display:

The PCoIP Ultra enhancements in the latest release exhibit some worthwhile network bandwidth savings for video workloads across the spectrum of image quality settings, mostly due to optimizations in our pixel block structures and differential encoding efficiencies. 

If you’ve deployed an NVIDIA vDWS environment and selected PCoIP-NVENC mode (calibrated to a default quality between Q80 and Q90), expect approximately 30 Mbps for the same 1080p 24 fps test case, offering a further 30% bandwidth savings over default PCoIP-AVX2 in this scenario.  These benefits will vary according to video content characteristics, so I’d recommend that you first run a sandbox with your own workload ahead of any detailed network capacity planning.

In Case You Missed It: 

Part 1: Top-Level Architecture

Part 2: PCoIP Ultra Client Capability

Part 3: Four Times the Pixel Rate

Part 4: PCoIP Ultra Efficient CPU Scaling