Dynamic Selective Devectorization for Efficient Power Gating of SIMD Units in a HW/SW Co-Designed Environment

Abstract

Abstract-Leakage power is a growing concern in current and future microprocessors. Functional units of microprocessors are responsible for a major fraction of this power. Therefore, reducing functional unit leakage has received much attention in the recent years. Power gating is one of the most widely used techniques to minimize leakage energy. Power gating turns off the functional units during the idle periods to reduce the leakage. Therefore, the amount of leakage energy savings is directly proportional to the idle time duration. This paper focuses on increasing the idle interval for the higher SIMD lanes. The applications are profiled dynamically, in a HW/SW co-designed environment, to find the higher SIMD lanes usage pattern. If the higher lanes need to be turned-on for small time periods, the corresponding portion of the code is devectorized to keep the higher lanes off. The devectorized code is executed on the lowest SIMD lane. Our experimental results show average SIMD accelerator energy savings of 12% and 24% relative to power gating, for SPECFP2006 and Physicsbench. Moreover, the slowdown caused due to devectorization is less than 1%.

Morphable hundred‐core heterogeneous architecture for energy‐aware computation

“…To tackle the development of energy‐efficient platforms, many techniques have been proposed. This includes turning‐off parts of the processor [1], dynamic selective de‐vectorisation [2] or using dynamic voltage and frequency scaling [3, 4] to decrease energy consumption whenever the computational requirements decrease. Recently, researchers have also turned to multi‐core heterogeneous systems composed of high‐performance ‘big’ cores and low‐power ‘small’ cores (e.g.…”

Design of low power shift register using activity-driven optimized clock gating and run-time power gating

“…The amount of leakage energy saved is directly proportional to the length of time interval for which the circuit remains idle. [3] Clock gating (CG) is the most common and widely used technique to reduce dynamic power, and Power gating (PG) is the dominant technique to reduce standby leakage power. Clock Gating can reduce power consumption of registers by switching off unnecessary clock signals based on a control signal when the values of these registers are not changing.…”

“…It can be thought of as a basic memorycell. The figure below depicts the internal schematic of a D Flip-Flop using NAND gates [3]. Fig.…”