PRINT VERSION: PDF [2.3MB]
Multi-Core Processing & Virtualization: Enabling Cost Effective Servers
A number of recent technology advances bring revolutionary change to the computing landscape. Enterprises of all sizes face the challenge of increasing system performance and reliability while lowering costs. New technologies offer a quantum leap forward with those goals. Recent steps forward in hardware and software technology help deliver breakthrough performance, power savings, and cost advantages that support larger real-time workloads with fewer servers.
From the 3-D ultrasound that shows the first glimpse of a new life to the C-arm X-ray used in life-saving procedures, real-time medical imaging becomes more prevalent and advanced each year.
High-end medical-imaging equipment must handle very large data sets deterministically to gather data and produce images in real-time. The necessity for set response times and extremely low latency make these tasks wellsuited to an RTOS.
Simultaneously, the software must support non-real-time functions such as a user interface, display and camera control, and networking support. A general-purpose OS is better suited to these requirements, particularly because of the wide availability of well-understood APIs, libraries, device drivers, and other components.
To provide this combination of functionality, many medical-imaging systems employ two separate pieces of hardware, with one running an RTOS and the other running a general-purpose OS. These increased system requirements add to equipment and support costs, while increased system complexity can lead to less favorable MTBF. Those factors can put the equipment at a competitive disadvantage in the marketplace.
Recent advances in processor architecture and virtualization technology allow these systems to be simplified down to a single board, while actually increasing the robustness of the solution. An advanced software stack and hardwareassisted virtualization allow multiple OSs to be run on a single processor at the same time, with each assigned to specific processor cores. Together, with dramatic advances in processor throughput, these designs reduce system cost while increasing stability and performance against measures like frame rate and time to first image.