Towards full-system energy-accuracy tradeoffs: A case study of an approximate smart camera system?

@article{Raha2017TowardsFE,
  title={Towards full-system energy-accuracy tradeoffs: A case study of an approximate smart camera system?},
  author={Arnab Raha and Vijay Raghunathan},
  journal={2017 54th ACM/EDAC/IEEE Design Automation Conference (DAC)},
  year={2017},
  pages={1-6}
}
  • Arnab Raha, V. Raghunathan
  • Published 18 June 2017
  • Computer Science
  • 2017 54th ACM/EDAC/IEEE Design Automation Conference (DAC)
The intrinsic error resilience exhibited by emerging application domains enables a new dimension for energy optimization of computing systems, namely the introduction of a controlled amount of approximations during system operation in exchange for substantial energy savings. Prior work in the area of approximate computing has focused on individual subsystems of a computing system, e.g., the computational subsystem or the memory subsystem. Since they focus only on individual subsystems, these… 

Figures and Tables from this paper

Approximating Beyond the Processor: Exploring Full-System Energy-Accuracy Tradeoffs in a Smart Camera System

A systematic methodology to perform joint approximations across different subsystems, leading to significant energy benefits compared to approximating individual subsystems in isolation is proposed.

Approximate Systems: Synergistically Approximating Sensing, Computing, Memory, and Communication Subsystems for Energy Efficiency

This chapter introduces the concept of an Approximate System that performs joint approximations across different subsystems, leading to significant energy benefits compared to approximating individual subsystems in isolation.

Approximate Computing for Iris Recognition Systems

This chapter explores a complex software/hardware pipeline system for iris code computation from live camera feed using an FPGA-based SoC and employs reinforcement learning technique with a recurrent neural network as the learning agent.

MIPAC: Dynamic Input-Aware Accuracy Control for Dynamic Auto-Tuning of Iterative Approximate Computing

A novel design framework is proposed that incorporates a lightweight quality controller that makes input-dependent predictions on the output quality and determines the optimal NOI at run-time, composed of accurate yet low-overhead NOI predictors, generated by a novel logic reduction technique.

SECO: A Scalable Accuracy Approximate Exponential Function Via Cross-Layer Optimization

  • Di WuTianen Chen Younghyun Kim
  • Computer Science
    2019 IEEE/ACM International Symposium on Low Power Electronics and Design (ISLPED)
  • 2019
This work proposes SECO, an approximate exponential function unit (EFU), a key operation in many signal processing applications and more importantly in spiking neuron models, and introduces a cross-layer design method for SECO to optimize the energy-accuracy trade-off.

Approximate computing for biometrie security systems: A case study on iris scanning

The case for a new direction for approximate computing in the field of biometric security with a comprehensive case study of iris scanning is made with an end-to-end flow from an input camera to the final iris encoding that produces sufficiently accurate final results despite relying on intermediate approximate computational steps.

BEACON: BEst Approximations for Complete BehaviOral HeterogeNeous SoCs

This paper investigates if lower-power designs can be found through mixing approximations across the different components in the SoC as opposed to only aggressively approximating a single component, and develops a framework that allows generating complete SoCs at the behavioral level through a bus generator and a library of synthesizable bus interfaces.

Weight-Oriented Approximation for Energy-Efficient Neural Network Inference Accelerators

A time-efficient automated framework for mapping the NN weights to the accuracy levels of the approximate reconfigurable accelerator that is able to satisfy tight accuracy loss thresholds, while significantly reducing energy consumption without any need for intensive NN retraining is proposed.

SAADI-EC: A Quality-Configurable Approximate Divider for Energy Efficiency

This paper proposes an approximate binary divider design that features dynamic configurability of accuracy and energy consumption and corrects the approximation error using simple, yet effective error compensation hardware to greatly improve the accuracy compared to the base implementation, SAADI.

References

SHOWING 1-10 OF 18 REFERENCES

Input-Based Dynamic Reconfiguration of Approximate Arithmetic Units for Video Encoding

Experimental results show that the approach of dynamically adjusting the degree of hardware approximation based on the input video respects the given quality bound across different videos while achieving a power saving up to 38% over a conventional nonapproximated MPEG encoder architecture.

Synergistic Approximation of Computation and Memory Subsystems for Error-Resilient Applications

This letter proposes a methodology to perform synergistic approximations across the computation and memory subsystems together that results in a significant improvement in energy consumption compared to the case when the subsystems are approximated independently.

Lax: Driver Interfaces for Approximate Sensor Device Access

Lax is presented, a device driver abstraction for interacting with sensors that enables power savings in exchange for occasionally returning erroneous sensor data, and can deliver significant system-level energy savings.

Quality configurable reduce-and-rank for energy efficient approximate computing

A runtime framework is developed that modulates the identified parameters during execution of RnR kernels to minimize their energy while meeting a given target quality, and identifies key parameters that can be used to tune the proposed approximation strategies dynamically.

Quality-aware data allocation in approximate DRAM*

Experimental results demonstrate a significant improvement in the energy-quality trade-off compared to previous work and show a reduction in DRAM refresh power of up to 73% with minimal loss in output quality.

IMPACT: IMPrecise adders for low-power approximate computing

This paper proposes logic complexity reduction as an alternative approach to take advantage of the relaxation of numerical accuracy, and demonstrates this concept by proposing various imprecise or approximate Full Adder cells with reduced complexity at the transistor level, and utilizing them to design approximate multi-bit adders.

Quality programmable vector processors for approximate computing

Quality programmable processors, in which the notion of quality is explicitly codified in the HW/SW interface, are suggested to be a significant step towards bringing approximate computing to the mainstream.

Scaling Performance via Self-Tuning Approximation for Graphics Engines

SAGE combines a static compiler that automatically generates a set of CUDA kernels with varying levels of approximation with a runtime system that iteratively selects among the available kernels to achieve speedup while adhering to a target output quality set by the user.

SAGE: Self-tuning approximation for graphics engines

Across a set of machine learning and image processing kernels, SAGE's approximation yields an average of 2.5× speedup with less than 10% quality loss compared to the accurate execution on a NVIDIA GTX 560 GPU.

Trading Accuracy for Power with an Underdesigned Multiplier Architecture

A novel multiplier architecture with tunable error characteristics, that leverages a modified inaccurate 2x2 building block, that can achieve 2X - 8X better Signal-Noise-Ratio (SNR) for the same power savings when compared to recent voltage over-scaling based power-error tradeoff methods is proposed.