Title :
DySER: Unifying Functionality and Parallelism Specialization for Energy-Efficient Computing
Author :
Govindaraju, Venkatraman ; Ho, Chen-Han ; Nowatzki, Tony ; Chhugani, Jatin ; Satish, Nadathur ; Sankaralingam, Karthikeyan ; Kim, Changkyu
Abstract :
The DySER (Dynamically Specializing Execution Resources) architecture supports both functionality specialization and parallelism specialization. By dynamically specializing frequently executing regions and applying parallelism mechanisms, DySER provides efficient functionality and parallelism specialization. It outperforms an out-of-order CPU, Streaming SIMD Extensions (SSE) acceleration, and GPU acceleration while consuming less energy. The full-system field-programmable gate array (FPGA) prototype of DySER integrated into OpenSparc demonstrates a practical implementation.
Keywords :
field programmable gate arrays; parallel processing; power aware computing; resource allocation; DySER architecture; FPGA; GPU acceleration; OpenSparc; dynamically specializing execution resources; energy-efficient computing; field-programmable gate array; functionality specialization; out-of-order CPU; parallelism mechanism; parallelism specialization; streaming SIMD extension; Computer architecture; Energy efficiency; Field programmable gate arrays; Hardware; Parallel processing; Prototypes; DySER; accelerator; architecture; data-level parallelism; energy efficiency; specialization;
Journal_Title :
Micro, IEEE
