Code-Partitioning for a Concise Characterization of Programs for Decoupled Code Tuning

نویسندگان

  • Eric Petit
  • François Bodin
چکیده

Astex, Automatic Speculative Thread EXtractor, is a compiler front-end we devellop for automatic C-code partitioning in speculative codelets, i.e. pure function that can be distributed over computing units of a system. It is an hybrid approach mixing static analysis and speculative data from profiles to offline generating a partition of the input C sequential code. In addition to the generated C partitioned application, we provide all the information on extracted kernels for a distributed memory heterogeneous system implementation and optimization. Astex makes a full abstraction of the target system for the programmer and provide useful guideline for code refactoring, if needed. It includes time profiling, communication, data layout information and data value prediction. Each codelet can be generated as a stand-alone program with associated input data set for effective iterative optimisation. Astex results on NAS, SPEC 2006 and H264 benchmarks are relevant, and future works are on the way toward automatic specialized coprocessor programming.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A Novel Self-tuning Zone PID Controller for Temperature Control via a PLC code

S7-1200 with Tia Portal technology has become a Standard function of distributed controlsystems. Self-Tuning methods belong to Programmable Controllers (PLC) techniques. PLCtechniques contain software packages for advanced control based on mathematical methods. S7-1200 tools are designed to increase the Process Capacity, yield, and quality of products. Most ofthe present time digital industry r...

متن کامل

Architecture-dependent partitioning of dependence graphs

Performance tuning of non-blocking threads is based on graph partitioning algorithms that create serial code blocks from dependence graphs. Previously existing algorithms are directed toward deadlock-avoidance and maximization of run-length. The latter criterion often generates a high synchronization overhead. This paper presents a partitioning algorithm for dependence graphs that uses a heuris...

متن کامل

A Message-Passing Distributed Memory Parallel Algorithm for a Dual-Code Thin Layer, Parabolized Navier-Stokes Solver

In this study, the results of parallelization of a 3-D dual code (Thin Layer, Parabolized Navier-Stokes solver) for solving supersonic turbulent flow around body and wing-body combinations are presented. As a serial code, TLNS solver is very time consuming and takes a large part of memory due to the iterative and lengthy computations. Also for complicated geometries, an exceeding number of grid...

متن کامل

Some notes on the characterization of two dimensional skew cyclic codes

‎‎A natural generalization of two dimensional cyclic code ($T{TDC}$) is two dimensional skew cyclic code‎. ‎It is well-known that there is a correspondence between two dimensional skew cyclic codes and left ideals of the quotient ring $R_n:=F[x,y;rho,theta]/_l$‎. ‎In this paper we characterize the left ideals of the ring $R_n$ with two methods and find the generator matrix for two dimensional s...

متن کامل

Decoupled Access/Execute Metaprogramming for GPU-Accelerated Systems

We describe the evaluation of several implementations of a simple image processing filter on an NVIDIA GTX 280 card. Our experimental results show that performance depends significantly on low-level details such as data layout and iteration space mapping which complicate code development and maintenance. We propose extending a CUDA or OpenCL like model with decoupled Access/Execute (“Æcute” [1]...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2010