Students in James Armstrong’s Design of Systems on a Chip (ECE 5524) class jumped into bio-mathematical computation last fall when they designed hardware accelerators for a “competition between species” model. Until some Americans introduced gray squirrels into parts of England in the early 20th century, red squirrels had been the only species of squirrel in the country. The gray squirrels were larger and bred faster and successfully competed for resources. Within a couple years of overlap in an area, the red squirrels disappeared.
“There is enough data on the red squirrel/grey squirrel competition in England that it is an excellent introduction to computation for biological modeling,” Armstrong said. Using an ARM processor and a field programmable gate array (FPGA), student teams were challenged to develop a hardware/software accelerator that could beat a Matlab standard for the model. Students were able to choose their strategy as long as the final system was a partitioned model between software and hardware, running on the ARM ASIC development platform. The teams found challenges ranging from serial transfer bottlenecks to memory storage.
The students enjoyed whole-system design approach. “I enjoyed working with a bunch of different things that all work together,” said Brian Marshall.
Patrick La Fratta agreed, saying, “I’ve had classes on just FPGAs and others on just software. This involved the interaction of both types of components and was challenging and fun.”