Advanced algorithms for performing signal and image processing on medical ultrasound signals often use the RF ultrasound signal and perform parallel adaptive signal processing. One example is ultrasonic flow estimation with the recursive least-squares lattice filter. Real-time processing is necessary for proper clinical evaluation of many algorithms, but this is generally impossible with conventional equipment.We present a multi-processor system capable of performing 1.2 billion flops, consisting of sixteen ADSP 21060 processors. Our system is based on four boards, each with four processors and shared internal and external memory. The processors can be flexibly configured using parallel communications links, so that almost any conceivable network of the sixteen processors can be made. The same type of links are used to interface with the data sampling system and with the PC used to display the processed data. The generous bandwidth of the links makes it easy to balance the computational load among the processors.In order to manage the shared system memory and to make use of the parallel processing capabilities of the system, a real-time multitasking kernel has been developed. The kernel uses a client-server model: the server manages semaphores to avoid memory conflicts and handles a queue of tasks and parameter addresses for scheduling the four DSP’s on the board.As an example, we show how to use the system to perform two parallel processes: imaging and flow estimation.
|Title of host publication||Proceedings of DSP Scandinavia'97|
|Publication status||Published - 1997|
|Event||DSP Scandinavia'97 - Stockholm|
Duration: 5 Nov 1829 → …
|Period||05/11/1829 → …|