The Portland Compiler Suite contains Fortran 95, Fortran 90, Fortran 77, C, C++, HPF and CUDA Accelerator compilers for machines running 64bit Linux.<\/p>\n
Version 13.6 is the most up to date installed on the CSF<\/p>\n
Note: the PGI CUDA Accelerator supports OpenACC compiler directives. This is available via the PGI C compiler ( The PGI compiler can be used to compile for Intel (Westmere, Sandybridge), AMD (Magny-Cours, Bulldozer) and Nvidia CUDA GPU architectures.<\/p>\n There are only two network licenses available site wide for each of Fortran and C\/C++. If you get a license related error, it is almost certainly because all of our licenses are in use and you should try again at a later time. For further licensing information see the following webpage:<\/p>\n To gain access to these compilers, run one<\/strong> of the following command after logging into the login node.<\/p>\n If your code is using the AMD Core Math Libraries (ACML) for routines such as BLAS, LAPACK, FFT and you are compiling for AMD Bulldozer nodes then a version of the compiler containing an optmized ACML utilizing the Fused Multiply Add instruction in Bulldozer (FMA4) is available by loading<\/p>\n Basic compilation commands are as follows:<\/p>\n Note: Use To compile specifically for architectures such as Intel Sandybridge or AMD Bulldozer use the Further optimization flags may be appropriate for different architectures. For example, on AMD Bulldozer it is recommended to use the following:<\/p>\n See below<\/a> for compiling for Nvidia GPUs (which uses the Sample code can be found in three subfolders of Each subfolder contains makefiles to compile the sample code. If you copy these folders and see an errors like <\/a><\/p>\n Note, this method uses PGI-specific directives, not OpenACC directives. We recommend using OpenACC for portability (see below<\/a>).<\/p>\n It is not necessary to be logged into a node hosting a GPU in order to compile CUDA executables. Copy one of the samples to your current working directory:<\/p>\n Compile it for NVIDIA CUDA either using the makefile provided in The following SGE script, Submit this as a batch job using:<\/p>\n If successful, the output file out.txt<\/strong> will contain the following:<\/p>\n Here is the source code for the example referred to above:<\/p>\n Note that all that is required in order to run a loop on the GPU is the line:<\/p>\n <\/a><\/p>\n Copy one of the samples to your current working directory:<\/p>\n Compile it either using the makefile provided in The code in this example is very similar to that given above. However, the OpenACC pragma<\/em> is slightly different as it introduces the kernel<\/em> keyword. You should examine carefully the The following SGE script, Submit this as a batch job using:<\/p>\n If successful, the output file out.txt<\/strong> will contain the following:<\/p>\n Once you have loaded the PGI module (see set up procedure<\/strong>), basic documentation is available as a manpage:<\/p>\n Documentation in pdf files is available in directory $PGI\/linux86-64\/$PGIVER\/doc and can be viewed using evince, e.g. <\/p>\n or<\/p>\n The following links may be useful:<\/p>\n Overview The Portland Compiler Suite contains Fortran 95, Fortran 90, Fortran 77, C, C++, HPF and CUDA Accelerator compilers for machines running 64bit Linux. Version 13.6 is the most up to date installed on the CSF Note: the PGI CUDA Accelerator supports OpenACC compiler directives. This is available via the PGI C compiler (pgcc) from version 12.10 and the PGI C++ and fortran compilers (pgcpp and pgf90) from version 13.6. The PGI compiler can be.. Read more »<\/a><\/p>\n","protected":false},"author":2,"featured_media":0,"parent":31,"menu_order":0,"comment_status":"open","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-105","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/ri.itservices.manchester.ac.uk\/csf-apps\/wp-json\/wp\/v2\/pages\/105","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ri.itservices.manchester.ac.uk\/csf-apps\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/ri.itservices.manchester.ac.uk\/csf-apps\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/ri.itservices.manchester.ac.uk\/csf-apps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/ri.itservices.manchester.ac.uk\/csf-apps\/wp-json\/wp\/v2\/comments?post=105"}],"version-history":[{"count":13,"href":"https:\/\/ri.itservices.manchester.ac.uk\/csf-apps\/wp-json\/wp\/v2\/pages\/105\/revisions"}],"predecessor-version":[{"id":2947,"href":"https:\/\/ri.itservices.manchester.ac.uk\/csf-apps\/wp-json\/wp\/v2\/pages\/105\/revisions\/2947"}],"up":[{"embeddable":true,"href":"https:\/\/ri.itservices.manchester.ac.uk\/csf-apps\/wp-json\/wp\/v2\/pages\/31"}],"wp:attachment":[{"href":"https:\/\/ri.itservices.manchester.ac.uk\/csf-apps\/wp-json\/wp\/v2\/media?parent=105"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}pgcc<\/code>) from version 12.10 and the PGI C++ and fortran compilers (pgcpp<\/code> and pgf90<\/code>) from version 13.6.<\/p>\nRestrictions on use<\/h2>\n
\n
Set up procedure<\/h2>\n
\r\nmodule load compilers\/PGI\/13.6\r\nmodule load compilers\/PGI\/12.10\r\nmodule load compilers\/PGI\/12.8\r\n<\/pre>\n
\r\nmodule load compilers\/PGI\/14.10-acml-fma4\r\nmodule load compilers\/PGI\/13.6-acml-fma4\r\n<\/pre>\n
Basic compilation commands<\/h2>\n
\n
\r\npgcc hello.c -o hello\r\n<\/pre>\n
\n
\r\npgCC hello.cpp -o hello\r\npgcpp hello.cpp -o hello\r\n<\/pre>\n
pgc++<\/code> if link compatibility with the GNU C++ compiler is required (see man pgc++<\/code>)<\/p>\n\n
\r\npgf90 hello.f90 -o hello\r\npgf95 hello.f90 -o hello\r\npgfortran hello.f90 -o hello\r\n<\/pre>\n
\n
\r\npghpf hello.hpf -o hello\r\n<\/pre>\n
Target Architectures<\/h3>\n
-tp=<\/code>architecture<\/em> flag. For example:<\/p>\n\r\npgcc -tp=sandybridge hello.c -o hello_sb\r\npgcc -tp=bulldozer hello.c -o hello_bd\r\n<\/pre>\n
\r\npgcc -tp=bulldozer -O3 -fast \r\n<\/pre>\n
-ta=<\/code>accelerator<\/em> flag to specify nvidia<\/code> accelerators).<\/p>\nSample code<\/h2>\n
$PGI\/linux86-64\/$PGIVER\/etc\/samples<\/code><\/p>\n\n
\/accel<\/code><\/li>\n\/cudafor<\/code><\/li>\n\/openacc<\/code><\/li>\n<\/ul>\n\/usr\/bin\/ld: cannot open output file c1.exe: Permission denied<\/code> when compiling ensure you have the necessary permissions to write files to your folders.<\/p>\nCompiling and submitting a test CUDA accelerator application in C<\/h2>\n
\r\ncp $PGI\/linux86-64\/$PGIVER\/etc\/samples\/accel\/c1.c .\r\n<\/pre>\n
\/accel<\/code> or use the following command:<\/p>\n\r\npgcc -ta=nvidia -Minfo=accel -fast c1.c -o c1.exe\r\n<\/pre>\n
pgi.sge<\/code>, is suitable for submitting the above executable:<\/p>\n\r\n#!\/bin\/bash\r\n#$-cwd\r\n#$-S \/bin\/bash\r\n#$-l nvidia\r\n.\/c1.exe > out.txt\r\n<\/pre>\n
\r\nqsub pgi.sge\r\n<\/pre>\n
\r\n100000 iterations completed\r\n<\/pre>\n
Source code for an example application<\/h3>\n
\r\n#include <stdio.h>\r\n#include <stdlib.h>\r\n#include <assert.h>\r\n\r\nint main( int argc, char* argv[] )\r\n{\r\n int n; \/* size of the vector *\/\r\n float *restrict a; \/* the vector *\/\r\n float *restrict r; \/* the results *\/\r\n float *restrict e; \/* expected results *\/\r\n int i;\r\n if( argc > 1 )\r\n n = atoi( argv[1] );\r\n else\r\n n = 100000;\r\n if( n <= 0 ) n = 100000;\r\n\r\n a = (float*)malloc(n*sizeof(float));\r\n r = (float*)malloc(n*sizeof(float));\r\n e = (float*)malloc(n*sizeof(float));\r\n for( i = 0; i < n; ++i ) a[i] = (float)(i+1);\r\n\r\n #pragma acc region\r\n {\r\n for( i = 0; i < n; ++i ) r[i] = a[i]*2.0f;\r\n }\r\n \/* compute on the host to compare *\/\r\n for( i = 0; i < n; ++i ) e[i] = a[i]*2.0f;\r\n \/* check the results *\/\r\n for( i = 0; i < n; ++i )\r\n assert( r[i] == e[i] );\r\n printf( \"%d iterations completed\\n\", n );\r\n return 0;\r\n}\r\n<\/pre>\n\r\n#pragma acc region\r\n<\/pre>\n
Compiling and submitting OpenACC C code<\/h2>\n
\r\ncp $PGI\/linux86-64\/$PGIVER\/etc\/samples\/openacc\/acc_c1.c .\r\n<\/pre>\n
\/openacc<\/code> or use the following command:<\/p>\n\r\npgcc -o acc_c1.exe acc_c1.c -acc<\/strong> -Minfo=accel -fast # notice the additional -acc flag\r\n<\/pre>\n
acc_c1.c<\/code> file.<\/p>\n\r\n#pragma acc kernels loop\r\n<\/pre>\n
pgi.sge<\/code>, is suitable for submitting the above executable:<\/p>\n\r\n#!\/bin\/bash\r\n#$-cwd\r\n#$-S \/bin\/bash\r\n#$-l nvidia\r\n.\/acc_c1.exe > out.txt\r\n<\/pre>\n
\r\nqsub pgi.sge\r\n<\/pre>\n
\r\n100000 iterations completed\r\n<\/pre>\n
Further information<\/h2>\n
\r\nman pgcc\r\n #\r\n # replace pgcc with the names of the other compilers as required\r\n #\r\n<\/pre>\n
\r\nevince $PGI\/linux86-64\/$PGIVER\/doc\/pgi12ug.pdf\r\n<\/pre>\n
\r\nevince $PGI\/linux86-64\/$PGIVER\/doc\/pgi12ref.pdf\r\n<\/pre>\n
\n