TAFFO: The compiler-based precision tuner

Cattaneo, Daniele; Chiari, Michele; Agosta, Giovanni; Cherubin, Stefano

doi:10.1016/j.softx.2022.101238

Cited by 6 publications

(3 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our solution for precision tuning with GPGPUs exploits the well-established taffo framework [3,6]. The five pipeline stages of the taffo architecture are called Initializer (init), Value Range Analysis (vra) Data Type Allocation (dta), Conversion (conv), and Feedback Estimator (fe).…”

Section: Methodology For Gpgpu Precision Tuningmentioning

confidence: 99%

Mixed Precision in Heterogeneous Parallel Computing Platforms via Delayed Code Analysis

Cattaneo,

Maggioli,

Magnani

et al. 2023

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Mixed Precision techniques have been successfully applied to improve the performance and energy efficiency of computation in embedded and high performance systems. However, few solutions have been proposed that address precision tuning of both GPGPU code and its corresponding CPU code, limiting the gains achievable by mixed precision. We propose an extension to the taffo precision tuning toolset that enables Mixed Precision across the space of floating and fixed point data types on GPGPUs, leveraging static analysis and providing seamless interface adaptation between host and GPGPU kernel code. The proposed tool achieves speedups exceeding 2× by exploiting the optimization of both kernel and host code.

show abstract

Section: Methodology For Gpgpu Precision Tuningmentioning

confidence: 99%

Mixed Precision in Heterogeneous Parallel Computing Platforms via Delayed Code Analysis

Cattaneo,

Maggioli,

Magnani

et al. 2023

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…Very few tools discussed currently in the literature satisfy these points. TAFFO [4] is a precision tuning tool implemented as a plugin for LLVM compiler framework. It works with programs written in C/C++ and compiles them into transprecision binaries trading off accuracy of result for the execution time.…”

Section: Precision Tuning Support At Compiler Levelmentioning

confidence: 99%

“…A precision tuning compiler, therefore, needs such information either from the programmer (by means of compiler hints expressed as annotations or pragmas) or from profiling (as in profile-guided optimization [18]). Our set of plugins for the LLVM compiler framework, TAFFO [4], takes the former approach, leveraging programmer annotations that express the value ranges of the input data. Figure 3 shows the TAFFO compiler pipeline, compared with the standard compilation flow performed by the LLVM.…”

Section: Precision Tuning Compiler: Architecturementioning

confidence: 99%

Hardware and Software Support for Mixed Precision Computing: a Roadmap for Embedded and HPC Systems

Fornaciari

Agosta

Cattaneo

et al. 2023

2023 Design, Automation &Amp; Test in Europe Conference &Amp; Exhibition (DATE)

Self Cite

View full text Add to dashboard Cite

Mixed precision is an approximate computing technique that can be used to trade-off computation accuracy for performance and/or energy. It can be applied to many error-tolerant applications, but manual precision tuning is both tedious and error-prone. Furthermore, the effectiveness of the technique heavily depends on hardware characteristics. Therefore, a hardware/software co-design approach is necessary for an effective exploitation of precision tuning opportunities offered by the applications. In this paper, we propose, based on the state of the art of precision tuning software and mixed precision hardware, a roadmap for the evolution of hardware designs and compiler-based precision tuning support, which is ongoing in the context of the European projects TEXTAROSSA and APROPOS. CCS CONCEPTS• Software and its engineering → Compilers; • Computer systems organization → Reduced instruction set computing; Parallel architectures.

show abstract

Design-time methodology for optimizing mixed-precision CPU architectures on FPGA

Denisov,

Galimberti,

Cattaneo

et al. 2024

Journal of Systems Architecture

View full text Add to dashboard Cite

TAFFO: The compiler-based precision tuner

Cited by 6 publications

References 22 publications

Mixed Precision in Heterogeneous Parallel Computing Platforms via Delayed Code Analysis

Mixed Precision in Heterogeneous Parallel Computing Platforms via Delayed Code Analysis

Hardware and Software Support for Mixed Precision Computing: a Roadmap for Embedded and HPC Systems

Design-time methodology for optimizing mixed-precision CPU architectures on FPGA

Contact Info

Product

Resources

About