|
PathScale Compiler Suite64-Bit Compilers for Linux |
|
User login |
Building HPCC 1.0.0 with the PathScale compilers
HPCC stands for High Performance Computing Challenge, and is a benchmark
incorporating Linpack and a number of other benchmarks to provide a more complete test of supercomputers than the standalone Linpack test currently used for the Top-500 list. You can find out more at the HPC Challenge web site.
Steps to building
To build HPCC, you will need to download the source, configure it to use
the PathScale compilers, and then build it. HPCC uses a BLAS library for a few of its benchmarks; we recommend the free GOTO BLAS library appropriate for your architecture.
Downloading HPCC
Building GOTO BLAS with the PathScale Fortran compiler Building HPCC with the PathScale Compilers OPTIONAL STREAM performance boost Known limitations
Downloading HPCC
HPCC can be downloaded from the website:
http://icl.cs.utk.edu/hpcc/software/index.html These notes are specific to the 1.0.0 version of HPCC released on
Building GOTO BLAS with the PathScale Fortran compiler
The following steps are done in the top-level GotoBLAS directory.
After downloading and unpacking the source code, you will need to modify Attached is a Makefile.rule from the GOTO 1.0.0 After modifying the file Makefile.rule, run the following commands to make cd exports make so cd ..
This should create a library that looks something like:
libgoto_opteron-r1.0.0.so where opteron will be replaced by your architecture if you are using
Building HPCC with the PathScale Compilers
After downloading, unpack the source code:
#example for release hpcc-1.0.0 tar -xvzf hpcc-1.0.0.tar.gz You will now have a top-level directory named hpcc-1.0.0. Configure the make system to build with PathScale compilers. To do If you want to customize the compilation flags used, this is the file to At this point, you may want skip to the OPTIONAL STREAM performance make arch=Linux_PSC_Opteron_GOTO
The final step before running is to customize the input file hpccinf.txt
file for the number of processes you will be running on, and how much memory they have. P * Q must be equal to your np. Tuning information can be found at: http://icl.cs.utk.edu/hpcc/faq/index.html As a start, we note that you can calculate memory usage per process mem/process = (N^2 * 8 bytes) / num processes The FAQ referenced above suggests aiming for about 80% of your total
OPTIONAL STREAM performance boost
The version of STREAM in HPCC was reworked in such a way as to give
compilers difficulty. It has become common for people wanting to show the very best STREAM performance to modify the make system to allow STREAM to be compiled with additional flags that are not applied to the rest of HPCC. This is completely permitted by the rules for a base submission as found at: http://icl.cs.utk.edu/hpcc/overview/index.html The flags we recommend for this are are defined in the files # from top-level directory for hpcc (hpcc-1.0.0 for current version) # (assuming we previously put Makefile.hpcc.patch in this directory) patch hpl/lib/arch/build/Makefile.hpcc Makefile.hpcc.patch
Known limitations
None known.
We recommend you record what steps you took, along with their output, make foo 2>&1 | tee make-foo-log.txt If you have any comments or suggestions about additions to these
categories [ ]
|
Active forum topicsShopping cart
|