Large Page Considerations
Compiler Methodology for Intel® MIC ArchitectureLarge Page ConsiderationsUse THP enabled by default in the MPSS Operating System:MPSS versions later than 2.1.4982-15 support “Transparent Huge Pages...
View ArticlePXFPUTC
POSIX Subroutine: Outputs a character to logical unit 6 (stdout).ModuleUSE IFPOSIXCALL PXFPUTC(ch,ierror)ch(Input) Character. The character to be written.ierror(Output) INTEGER(4). The error status.If...
View ArticlePXFPOSIXIO
POSIX Subroutine: Sets the current value of the POSIX I/O flag.ModuleUSE IFPOSIXCALL PXFPOSIXIO(new,old,ierror)new(Input) INTEGER(4). The new value for the POSIX I/O flag.old(Output) INTEGER(4). The...
View ArticlePXFPIPE (L*X, M*X)
POSIX Subroutine: Creates a communications pipe between two processes.ModuleUSE IFPOSIXCALL PXFPIPE(ireadfd,iwritefd,ierror)ireadfd(Output) INTEGER(4). The file descriptor for reading.iwritefd(Output)...
View ArticlePXFPAUSE
POSIX Subroutine: Suspends process execution.ModuleUSE IFPOSIXCALL PXFPAUSE(ierror)ierror(Output) INTEGER(4). The error status.If successful, ierror is set to zero; otherwise, an error code.The...
View ArticlePXFGETUID (L*X, M*X)
POSIX Subroutine: Gets the real user ID of the current process.ModuleUSE IFPOSIXCALL PXFGETUID(iuid,ierror)iuid(Output) INTEGER(4). The returned real user ID.ierror(Output) INTEGER(4). The error...
View ArticlePXFGETEGID (L*X, M*X)
POSIX Subroutine: Gets the effective group ID of the current process.ModuleUSE IFPOSIXCALL PXFGETEGID(iegid,ierror)iegid(Output) INTEGER(4). The returned effective group ID.ierror(Output) INTEGER(4)....
View ArticleCache Blocking Techniques
Compiler Methodology for Intel® MIC ArchitectureCache Blocking TechniquesOverviewAn important class of algorithmic changes involves blocking data structures to fit in cache. By organizing data memory...
View ArticleEfficient Parallelization
Efficient Parallelization Document Compiler Methodology for Intel® MIC Architecture Efficient Parallelization OverviewThis chapter covers topics in parallelization. There are links to various...
View ArticleOpenMP Thread Affinity Control
Intel® Composer XE for MIC Compi Compiler Methodology for Intel® MIC Architecture Compiler Methodology for Intel® MIC Architecture Efficient Parallelization, OpenMP Thread Affinity ControlOverviewThe...
View ArticleOverview of Vectorization Reports and new vec-report6
Overview of vectorization report Compiler Methodology for Intel® MIC ArchitectureVectorization Essentials, Vectorization and Optimization Reports, Overview of vectorization reports and new...
View ArticleMemory Allocation and First-Touch
Compiler Methodology for Intel® MIC ArchitectureMemory Allocation and First-TouchMemory allocation is expensive on the coprocessor compared to Xeon - so it is prudent to reuse already-allocated memory...
View ArticleGetting Started with Intel® Composer XE 2013, New User Compiler Basics
Compiler Methodology for Intel® MIC Architecture Getting Started with Intel® Composer XE 2013, New User Compiler BasicsOverviewModern compilers can be invoked with hundreds of options. From these, what...
View ArticleElement wise alignment requirements for data accesses to be ABI-compliant on...
Compiler Methodology for Intel® MIC ArchitectureUnlike the IA-32 and Intel® 64 architectures, the Intel® MIC Architecture requires all data accesses to be properly aligned according to their size,...
View ArticleResolving problem when building HDF5* with Intel® compiler 14.0
IntroductionTo build the latest HDF5* with Intel® compiler 14.0, a segmantation fault occurs when running "make check". This article is to provide a solution in resolving this issue. The information in...
View ArticleIntel® Visual Fortran Composer XE 2013 SP1 for Windows* with IMSL*, Update 1
Intel® Visual Fortran Composer XE 2013 SP1 includes the latest Intel Fortran compilers and performance libraries for IA-32 and Intel® 64 architecture systems. This new product release now includes:...
View Article面向英特尔® 至强融核™ 协处理器(和英特尔® 至强® 处理器)架构应用的浮点计算 R2R 再现性
问题如果在相同处理器上针对相同输入数据重新运行相同的程序，得到的结果相同吗？串行代码串行代码浮点运算结果的差异的唯一已知来源是由外部环境更改引发的数据调整。 例如，用户动态分配了一个包含日期的新字符串。 该字符串的大小随一年中的月份或周的日期而不同，因此需要对分配的日期进行后续调整。 调整范围决定着哪些循环迭代是循环开头或循环结尾(remainder...
View ArticleIntel® Compilers - Supported compiler versions
The following compiler versions are supported:Intel® C++ and Fortran Composer XE 2013 SP1 for Linux* or OS X* (Compiler version 14.0)Intel® C++ and Fortran Composer XE 2013 for Linux* or OS X*...
View Article内存分配和首次访问
面向英特尔® MIC 架构的编译器方法内存分配和首次访问相对至强而言，协处理器的内存分配成本较高——因此尽可能重复使用已非配的内存是非常明智的。例如，如果某函数被重复调用（假设在循环内），并且该函数使用数组作为临时存储，尝试初次分配足够大的数组（所需的最大大小），并在后续调用中重复使用该数组：static real *temp_array=0; void foo(..) {...
View ArticleOpenMP Related Tips
Compiler Methodology for Intel® MIC ArchitectureOpenMP Related TipsOpenMP* Loop Collapse DirectiveUse the OpenMP collapse-clause to increase the total number of iterations that will be partitioned...
View Article