Exploiting Adaptive Data Compression to Improve Performance and Energy-Efficiency of Compute Workloads in Multi-GPU Systems

2019 ◽  
Author(s):  
Mohammad Khavari Tavana ◽  
Yifan Sun ◽  
Nicolas Bohm Agostini ◽  
David Kaeli
Keyword(s):  
Energy Efficiency ◽  
Data Compression ◽  
Improve Performance

scholarly journals Ineffectual Neuron Free Deep Learning Accelerator

2019 ◽  
Vol 8 (9) ◽  
pp. 2819-2826
Keyword(s):  
Neural Network ◽  
Energy Efficiency ◽  
Deep Learning ◽  
Visual Cortex ◽  
Computational Complexity ◽  
Image Recognition ◽  
Deep Neural Network ◽  
Improve Performance ◽  
High Efficient ◽  
Accelerator Design

Convolutional neural network (CNN) is actually a deep neural network which plays an important role in image recognition. The CNN recognizes images similar to visual cortex in our eyes. In this proposed work, an accelerator is used for high efficient convolutional computations. The main aim of using the accelerator is to avoid ineffectusal computations and to improve performance and energy efficiency during image recognition without any loss in accuracy. However, the throughput of the accelerator is improved by adding max-pooling function only. Since the CNN includes multiple inputs and intermediate weights for its convolutional computation, the computational complexity is increased enormously. Hence, to reduce the computational complexity of the CNN, a CNN accelerator is proposed in this paper. The accelerator design is simulated and synthesized in Cadence RTL compiler tool with 90nm technology library.

scholarly journals C-Pack: Cache Compression for Microprocessor Performance

2011 ◽  
pp. 102-107
Author(s):  
T. Narasimhulu
Keyword(s):  
Power Consumption ◽  
Data Compression ◽  
Lossless Compression ◽  
Register Transfer Level ◽  
Improve Performance ◽  
Memory Hierarchies ◽  
Compression Algorithms ◽  
Cache Line ◽  
On Line ◽  
Cache Compression

Computer systems and micro architecture researchers have proposed using hardware data compression units within the memory hierarchies of microprocessors in order to improve performance, energy efficiency, and functionality. However, most past work, and all work on cache compression, has made unsubstantiated assumptions about the performance, power consumption, and area overheads of the proposed compression algorithms and hardware. In this work, I present a lossless compression algorithm that has been designed for fast on-line data compression, and cache compression in particular. The algorithm has a number of novel features tailored for this application, including combining pairs of compressed lines into one cache line and allowing parallel compression of multiple words while using a single dictionary and without degradation in compression ratio. We reduced the proposed algorithm to a register transfer level hardware design, permitting performance, power consumption, and area estimation.

On the Energy Efficiency of Dual Prime Mover Pump-Controlled Hydraulic Cylinders

2019 ◽  
Author(s):  
Petter H. Gøytil ◽  
Damiano Padovani ◽  
Michael R. Hansen
Keyword(s):  
Energy Efficiency ◽  
Operating Conditions ◽  
Prime Mover ◽  
Improve Performance ◽  
Servo Valve ◽  
System Parameters ◽  
Component Sizing ◽  
Controlled Systems ◽  
Hydraulic Cylinders ◽  
Prime Movers

Abstract This paper concerns the energy efficiency of a special class of pump-controlled hydraulic cylinders utilizing two prime movers. The performance of such circuits has been studied previously motivated by their capability of providing an actuator stiffness similar to that of servo valve-controlled systems. This characteristic may improve performance and robustness in applications requiring feedback control. In this paper, the presence of losses similar to that of fluid throttling, in the sense that they occur even in the absence of component inefficiencies, are demonstrated for such circuits and shown to degrade the overall energy efficiency of the system. The conditions under which such losses occur are derived analytically as a function of system parameters and operating conditions and two solutions for their elimination are proposed and verified analytically and numerically. Several implementation options are compared in terms of energy efficiency and component sizing and benchmarked to a conventional servo valve solution. It is shown that with the appropriate implementation, an energy efficiency up to ten times greater than that of a conventional servo valve system may be expected.

scholarly journals A new scheme for evaluating energy efficiency of data compression in wireless sensor networks

2018 ◽  
Vol 14 (5) ◽  
pp. 155014771877692
Author(s):  
Shaoqiang Liu ◽  
Yanfang Liu ◽  
Xiang Chen ◽  
Xiaoping Fan
Keyword(s):  
Energy Efficiency ◽  
Wireless Sensor Networks ◽  
Energy Consumption ◽  
Sensor Networks ◽  
Data Compression ◽  
Energy Saving ◽  
Data Communication ◽  
Wireless Sensor ◽  
Compression Algorithms ◽  
Wireless Module

Data communication incurs the highest energy cost in wireless sensor networks, and restricts the application of wireless sensor networks. Data compression is a promising technique that can reduce the amount of data exchanged between nodes and results in energy saving. However, there is a lack of effective methods to evaluate the efficiency of data compression algorithms and to increase nodes’ energy efficiency. The energy saving of nodes is related to both hardware and software, this article proposes a new scheme for evaluating energy efficiency of data compression in wireless sensor networks according to the node’s hardware and software. The relationship between the energy efficiency and the hardware and software factors is expressed by a formula. In this formula, energy efficiency can be improved by increasing the compression ratio and decreasing the ratio of s/ k, in which k represents the node’s hardware factor related to energy consumption of processor, wireless module, and so on and s represents the software factor that reflects the energy consumption of the algorithm. Based on the scheme, a mechanism is proposed to improve the node’s energy efficiency by selecting effective algorithms in accordance with the node’s radio frequency power. The feasibility of the scheme is demonstrated with lossless data compression algorithms on the MSP430F2618 processor.

Sign in / Sign up

Export Citation Format

Share Document