Intel Fortran Composer XE FCX999ASGE01 Manuel D’Utilisation
Codes de produits
FCX999ASGE01
Top Features
Example of Coarray Fortran
Intel® Fortran – Great Standards Supports
The compiler in Intel Fortran Composer XE 2013 offers great Fortran
standards coverage, now with expanded support for Fortran 2003 and
2008 and full source code compatibility with Compaq Visual Fortran. On
Windows and Linux, it includes support for coarray Fortran, providing
support for multi-cpu shared-memory nodes. Cluster support is available
in Intel® Cluster Studio XE product. The benefit is performance for your
Fortran applications as they run on individual computers or clusters.
standards coverage, now with expanded support for Fortran 2003 and
2008 and full source code compatibility with Compaq Visual Fortran. On
Windows and Linux, it includes support for coarray Fortran, providing
support for multi-cpu shared-memory nodes. Cluster support is available
in Intel® Cluster Studio XE product. The benefit is performance for your
Fortran applications as they run on individual computers or clusters.
Intel® Fortran compiler supports Intel® Core® & Xeon®
Processors and Intel® Xeon Phi™ Coprocessors
Intel Fortran supports Intel and compatible processors, including application
offloading to Intel Xeon Phi coprocessors and the graphics processing units
offloading to Intel Xeon Phi coprocessors and the graphics processing units
that are part of the Intel Core i3, i5 and i7 processors.
Vectorization, Parallelism Made Powerful
Intel Fortran, like Intel C++, includes support for three industry-leading
vectorization and parallelism features associated with Intel Core
Processors, Intel Xeon Processors and Intel Xeon Phi coprocessors.
First, the compiler automatically vectorizes and/or automatically
parallelizes code for use on systems with Intel® Core™, Intel® Xeon, and
compatible processors. These capabilities have been enhanced for
broader applicability and improved application performance. And they
support the wider vectors in the latest Intel Architecture processors.
Second, when a loop can’t be vectorized, GAP – Guided Auto Parallelism
can generate a report suggesting source code changes, use of pragmas,
or use of compiler options. You remain in control in applying suggestions.
In cases where you know your code well, GAP can offer great
opportunities for performance improvement.
Third, Intel Fortran supports application development for Intel® Xeon Phi
coprocessors using simple directives. Support is extended to Xeon Phi
coprocessors hosted on Xeon-based systems using Linux and, new with
SP1, Windows. The benefit? Choose your operating environment,
maintain your existing programming model, and use of Fortran, to take
advantage of outstanding application performance.
vectorization and parallelism features associated with Intel Core
Processors, Intel Xeon Processors and Intel Xeon Phi coprocessors.
First, the compiler automatically vectorizes and/or automatically
parallelizes code for use on systems with Intel® Core™, Intel® Xeon, and
compatible processors. These capabilities have been enhanced for
broader applicability and improved application performance. And they
support the wider vectors in the latest Intel Architecture processors.
Second, when a loop can’t be vectorized, GAP – Guided Auto Parallelism
can generate a report suggesting source code changes, use of pragmas,
or use of compiler options. You remain in control in applying suggestions.
In cases where you know your code well, GAP can offer great
opportunities for performance improvement.
Third, Intel Fortran supports application development for Intel® Xeon Phi
coprocessors using simple directives. Support is extended to Xeon Phi
coprocessors hosted on Xeon-based systems using Linux and, new with
SP1, Windows. The benefit? Choose your operating environment,
maintain your existing programming model, and use of Fortran, to take
advantage of outstanding application performance.
Intel® MKL – Optimization Made Even Easier
One of the easiest ways to take advantage of vectorization and
threading is to make calls in your applications to the pre-optimized
functions in the Intel® MKL. A wealth of routines is included to optimize
math-intensive application performance. Applications using Intel® MKL
functions automatically scale on previous, current and future processor
architectures. Just re-link to the latest library version and your code is
ready to take advantage of the latest processor features available.
Intel® MKL offers the highest-performing implementations of LAPaCK and
BLAS linear algebra functions and are 100% compatible with these de
facto industry standard APIs. Fast Fourier Transforms functions include
FFTW*-compatible APIs to ease adoption. Additional routines include
vectorized transcendental functions, random number generators for
several probability distributions, convolution and correlation routines, and
summary statistics functions.
The Rogue Wave* IMSL* Numerical Library for Intel Visual Fortran
(Windows only) is also available. This leading library offers the largest
collection of commercially-available math and statistical functions for
science, technical, and business environments.
threading is to make calls in your applications to the pre-optimized
functions in the Intel® MKL. A wealth of routines is included to optimize
math-intensive application performance. Applications using Intel® MKL
functions automatically scale on previous, current and future processor
architectures. Just re-link to the latest library version and your code is
ready to take advantage of the latest processor features available.
Intel® MKL offers the highest-performing implementations of LAPaCK and
BLAS linear algebra functions and are 100% compatible with these de
facto industry standard APIs. Fast Fourier Transforms functions include
FFTW*-compatible APIs to ease adoption. Additional routines include
vectorized transcendental functions, random number generators for
several probability distributions, convolution and correlation routines, and
summary statistics functions.
The Rogue Wave* IMSL* Numerical Library for Intel Visual Fortran
(Windows only) is also available. This leading library offers the largest
collection of commercially-available math and statistical functions for
science, technical, and business environments.
real
,
ALLOCATABLE
:: grid(:,:)[:,:]
...
ALLOCATE
(grid(0:N+1,0:M+1)[1:P,1:*])
! with halo
...
SYNC ALL
!...wait for all iterations
grid( 0, 1:M) = grid( N, 1:M)[north_P,me_Q]
! north
grid( N+1, 1:M) = grid( 1, 1:M)[south_P,me_Q]
! south
grid( 1:N, M+1) = grid( 1:N, 1 )[me_P, east_Q]
! east
grid( 1:N, 0 ) = grid( 1:N, M )[me_P, west_Q]
! west