Tool Integration for Source-Level Mixed Precision

Lam, Michael O.; Vanderbruggen, Tristan; Menon, Harshitha; Schordan, Markus

doi:10.1109/correctness49594.2019.00009

Cited by 16 publications

(6 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In that, our approach is related to mixed-precision tuning techniques that mostly focus on implementations that mix single, double and quad floating-point precision. Some of these use dynamic analysis to estimate errors and thus do not provide sound guarantees [25,19,17,15], and others use static analysis with accuracy guarantees, but less scalability [5,9]. Mixed-precision tuning generally works well when the target error bounds are close to the error bounds of uniform-precision implementations [9,27].…”

Section: Related Workmentioning

confidence: 99%

Inferring Interval-Valued Floating-Point Preconditions

Krämer¹,

Blatter

Darulová

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Aggregated roundoff errors caused by floating-point arithmetic can make numerical code highly unreliable. Verified postconditions for floating-point functions can guarantee the accuracy of their results under specific preconditions on the function inputs, but how to systematically find an adequate precondition for a desired error bound has not been explored so far. We present two novel techniques for automatically synthesizing preconditions for floating-point functions that guarantee that user-provided accuracy requirements are satisfied. Our evaluation on a standard benchmark set shows that our approaches are complementary and able to find accurate preconditions in reasonable time.

show abstract

Section: Related Workmentioning

confidence: 99%

Inferring Interval-Valued Floating-Point Preconditions

Krämer¹,

Blatter

Darulová

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…Complications typically arise due to (i) the different treatment of built-ins compared to user-defined types by the C++ language causing compilation errors [19], and (ii) the usage of external libraries [18], [20] and C-language function calls [21] that are not compatible with the AD type (as seen in LULESH). While tools exist to (partially) automate the process of the type change [22], these may not be able to handle, e.g., external solver libraries, which require special treatment in the adjoint context [23].…”

Section: A Ad-enhancement Of Luleshmentioning

confidence: 99%

Towards compiler-aided correctness checking of adjoint MPI applications

Hück

Protze

Lehr

et al. 2020

2020 IEEE/ACM 4th International Workshop on Software Correctness for HPC Applications (Correctness)

View full text Add to dashboard Cite

Algorithmic Differentiation (AD) is a set of techniques to calculate derivatives of a computer program. In C++, AD typically requires (i) a type change of the built-in double, and (ii) a replacement of all MPI calls with AD-specific implementations. This poses challenges on MPI correctness tools, such as MUST, a dynamic checker, and TypeART, its memory sanitizer extension. In particular, AD impacts (i) memory layouts of the whole code, (ii) requires more memory allocations tracking by TypeART, and (iii) approximately doubles the MPI type checks of MUST due to an AD-specific communication reversal. To address these challenges, we propose a new callback interface for MUST to reduce the number of intercepted MPI calls, and, also, improve the filtering capabilities of TypeART to reduce tracking of temporary allocations for the derivative computation. We evaluate our approach on an AD-enhanced version of CORAL LULESH. In particular, we reduce stack variable tracking from 32 million to 13 thousand. MUST with TypeART and the callback interface reduces the runtime overhead to that of vanilla MUST.

show abstract

“…In [3] a benchmark suite of programs is introduced for mixed precision computing analysis. Moreover the authors present the performance of various precision auto-tuning algorithms such as combinational (used in FloatSmith [4]), compositional (used in FloatSmith [4]), Delta Debug (introduced in [5], used in Precimonious [6] and PROMISE [1]), hierarchical (used in CRAFT-HPC [7]), hierarchical-compositional (used in FloatSmith [4]) and a Genetic Search Algorithm (GA) (used in AMPT-GA [8]). The Delta Debug algorithm requires multiple executions of the user program to provide a suitable type configuration.…”

Section: Introductionmentioning

confidence: 99%

Performance of precision auto-tuned neural networks

Ferro,

Graillat,

Hilaire

et al. 2023

2023 IEEE 16th International Symposium on Embedded Multicore/Many-Core Systems-on-Chip (MCSoC)

View full text Add to dashboard Cite

While often used in embedded systems, neural networks can be costly in terms of memory and execution time.Reducing the precision used in neural networks can be beneficial in terms of performance and energy consumption. After having applied a floating-point auto-tuning tool, PROMISE, on various neural networks, we obtained versions using lower precision while keeping a required accuracy on the results. In this article, we present results regarding the memory and computation time gains obtained thanks to reduced precision, using vectorized and non-vectorized code. We also show the impact on the execution time of PROMISE of the parallelization of the Delta Debug algorithm it implements.

show abstract

Tool Integration for Source-Level Mixed Precision

Cited by 16 publications

References 21 publications

Inferring Interval-Valued Floating-Point Preconditions

Inferring Interval-Valued Floating-Point Preconditions

Towards compiler-aided correctness checking of adjoint MPI applications

Performance of precision auto-tuned neural networks

Contact Info

Product

Resources

About