A New Era for Video Processing: Technology Brief
Network traffic is exploding, placing unprecedented demands on media servers to increase workload density and throughput. Video is, and will continue to be, a driving force in this trend, due to the popularity of social media such as YouTube* and Facebook*, and consumer devices such as smart phones, tablets, and mobile TVs. It is predicted that, “two thirds of mobile data traffic will be video by 2015, and mobile video will more than double every year between 2010 and 2015.”
This tremendous growth will require signifcant infrastructure build-outs and leave service providers to balance demands for power, bandwidth, advanced traffic control, differing standards, and quality. These increases in network costs (capital and operating) can easily outpace revenue growth. As service providers position themselves for the next cycle of network upgrades, they must decide how to simplify the convergence of multiple media workloads onto one platform, while insuring flexibility in the network to deploy new intelligent services as they come to market.
Media processing workloads —for example, transcoding/transrating media streams in real-time— have traditionally been managed by a range of multicore DSP-, ASIC- or multicore MIPS-based solutions. However, software must often be optimized to partition workloads among the cores, requiring programmers with a specific skill set in order to achieve optimal performance. Moreover, these solutions normally host a local management processor for control and workload distribution, all of which adds to the cost of development, scaling, and time-to-market.
The 3rd generation Intel® Core™ processor family (codename Ivy Bridge) in mobile, desktop and workstation platforms, is extremely well-suited to handle media processing workloads through hardware acceleration. While leaving CPU headroom for other workloads such as cryptography or control plane, the on-chip processor graphics provide a low-power, high-density performance solution for media processing while minimizing power consumption and maximizing performance.