Noise Budgeting in Multiple-Kernel Word-Length Optimization

Abstract : Word-Length Optimization (WLO) is a key step in the implementation of Digital Signal Processing applications on hardware platforms. Existing approaches face scalability problems for applications with several kernels. In this paper, we present our work-in-progress to address the scalability problems when performing multi-kernel WLO. Our approach uses application-wide analysis to derive noise budgets for each kernel, followed by independent WLO. The main idea is to characterize the impact of approximating each kernel to the accuracy/cost through simulation and regression analysis. The constructed models can be used to determine the appropriate noise budget for each kernel. When applied to WLO for two kernels in Image Signal Processor, the noise budgets given by our analysis closely match those found empirically with global WLO.
Complete list of metadatas

Cited literature [13 references]  Display  Hide  Download

https://hal.inria.fr/hal-02183936
Contributor : Olivier Sentieys <>
Submitted on : Monday, July 15, 2019 - 4:55:57 PM
Last modification on : Friday, September 13, 2019 - 9:49:43 AM

File

AxC19_paper_7.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02183936, version 1

Citation

Van-Phu Ha, Tomofumi Yuki, Olivier Sentieys. Noise Budgeting in Multiple-Kernel Word-Length Optimization. AxC 2019 - 4th Workshop on Approximate Computing, Mar 2019, Florence, Italy. pp.1-3. ⟨hal-02183936⟩

Share

Metrics

Record views

275

Files downloads

495