PROJECT TITLE :

Cache Impacts of Datatype Acceleration

ABSTRACT:

Hardware acceleration may be a widely accepted solution for performance and energy efficient computation as a result of it removes unnecessary hardware for general computation whereas delivering exceptional performance via specialized control methods and execution units. The spectrum of accelerators available today ranges from coarse-grain off-load engines like GPUs to fine-grain instruction set extensions such as SSE. This research explores the advantages and challenges of managing memory at the data-structure level and exposing those operations on to the ISA. We tend to decision these instructions Abstract Datatype Directions (ADIs). This paper quantifies the performance and energy impact of ADIs on the instruction and data cache hierarchies. For instruction fetch, our measurements indicate that ADIs will lead to 21̵one;48% and sixteen–27% reductions in instruction fetch time and energy respectively. For information delivery, we observe a 22̵one;forty% reduction in total data read/write time and nine̵one;thirty% in total information scan/write energy.