4. Compilers that generate binaries for login nodes¶

4.1. Intel oneAPI ¶

Login node (Intel) can use Intel oneAPI Base & HPC Toolkit.

4.1.1. Environment setting ¶

To use Intel oneAPI compiler, it is needed to set the required path.

Please Use a shell script or modulefile to set the required paths.

To use a shell script for setting paths

Do the following:.

[_LNIlogin]$ source /opt/intel/oneapi/setvars.sh intel64

:: initializing oneAPI environment ...
   bash: BASH_VERSION = 4.4.20(1)-release
   args: Using "$@" for setvars.sh arguments: intel64
:: advisor -- latest
:: ccl -- latest
:: clck -- latest
:: compiler -- latest
:: dal -- latest
:: debugger -- latest
:: dev-utilities -- latest
:: dnnl -- latest
:: dpcpp-ct -- latest
:: dpl -- latest
:: inspector -- latest
:: ipp -- latest
:: ippcp -- latest
:: ipp -- latest
:: itac -- latest
:: mkl -- latest
:: mpi -- latest
:: tbb -- latest
:: vpl -- latest
:: vtune -- latest
:: oneAPI environment initialized ::

To use modulefile for setting paths

If you use a modulefile, you must load the modulefile that corresponds to the feature you want to use. If you are using 2024.1.0 or later, set MODULES_AUTO_HANDLING=1 beforehand.

For example, to make use of MKL, load mkl/latest and compiler/latest.
```
[_LNIlogin]$ export MODULES_AUTO_HANDLING=1
[_LNIlogin]$ module load mkl/latest
Loading mkl/latest
  Loading requirement: tbb/latest compiler-rt/latest
[_LNIlogin]$ module load compiler/latest
Loading compiler/latest
  Loading requirement: oclfpga/latest
```
Check the supplied modulefile with the module avail command.

4.1.2. To compile ¶

Compile command is as following.

Compiler type	Command name
DPC++/C++ Compiler	icx, icpx, dpcpp
Fortran Compiler	ifx

Sequential (Example of Intel Fortran)
```
[_LNIlogin]$ ifx -O3 source file name
```

OpenMP (Example of Intel Fortran)

[_LNIlogin]$ ifx -O3 -qopenmp source file name

Example of Intel oneAPI DPC++ compiler (oneAPI Example Compilation)

[_LNIlogin]$ dpcpp -I${MKLROOT}/include -c axpy.cpp -o axpy.o
[_LNIlogin]$ dpcpp axpy.o -fsycl-device-code-split=per_kernel \
"${MKLROOT}/lib/intel64"/libmkl_sycl.a -Wl,-export-dynamic -Wl,--start-group \
"${MKLROOT}/lib/intel64"/libmkl_intel_ilp64.a \
"${MKLROOT}/lib/intel64"/libmkl_sequential.a \
"${MKLROOT}/lib/intel64"/libmkl_core.a -Wl,--end-group -lsycl -lOpenCL \
-lpthread -lm -ldl -o axpy.out

4.1.3. About Intel oneAPI compiler ¶

For more information about the Intel oneAPI compiler, refer to the Intel home page.

https://www.intel.com/content/www/us/en/developer/tools/oneapi/overview.html

For information on the DPC++ (Data Parallel C++) compiler, see:

https://www.intel.com/content/www/us/en/developer/tools/oneapi/dpc-compiler.html

For information on the Intel Fortran compiler classic, see:

https://www.intel.com/content/www/us/en/developer/tools/oneapi/fortran-compiler.html

4.2. Arm Compiler for Linux ¶

Users can use Arm Compiler for Linux on the Arm login node.

4.2.1. Environment setting ¶

To use Arm compiler, it is needed to set the required path.

Please check the modulefile with the module avail command and set the environment appropriately.

[_LNAlogin]$ module avail

--------------------------------------- /opt/arm/modulefiles ---------------------------------------
arm22/22.0.2    arm23/23.10     binutils/11.2.0 binutils/9.2.0  gnu/13.2.0
arm23/23.04     arm24/24.04     binutils/12.2.0 gnu/11.2.0      gnu10/10.2.0
arm23/23.04.1   binutils/10.2.0 binutils/13.2.0 gnu/12.2.0      gnu9/9.2.0


[_LNAlogin]$ module load arm24/24.04

Modulefile name contains ‘gcc’ is used, GCC compiler (included in Arm Compiler for Linux) will be used.

When generating executable file for login node(Arm), you can expedite the optimization by specifying -mcpu=thunderx2t99 option.

4.2.2. To compile ¶

Compile command is as follows.

Language	Command name
Fortran	armflang
C	armclang
C++	armclang++

Sequential (Example of C)

[_LNAlogin]$ armclang -Ofast <source file name>

OpenMP (Example of C)

[_LNAlogin]$ armclang -Ofast -fopenmp <source file name>

For the compiler options that can be specified, please see the man command or Reference Guide.

Please refer to man command or reference guide for specifiable compiler options in the Arm compiler.

Arm Fortran Compiler Developer and Reference Guide: https://developer.arm.com/documentation/101380/2404/?lang=en
Arm C/C++ Compiler Developer and Reference Guide: https://developer.arm.com/documentation/101458/2404/?lang=en
Arm Compiler for Linux: https://developer.arm.com/Tools%20and%20Software/Arm%20Compiler%20for%20Linux

4.2.3. Arm Performance Libraries ¶

A licensed version of Arm Performance Libraries (ArmPL) is available as part of Arm Compiler for Linux. ArmPL includes the following libraries,

BLAS : Basic Linear Algebra Subprograms (including XBLAS, the extended precision BLAS).
LAPACK 3.11.0 : a comprehensive package of higher level linear algebra routines.
FFT functions : a set of Fast Fourier Transform routines for real and complex data using the FFTW interface.
Sparse linear algebra.
libamath : a subset of libm, which is a set of optimized mathematical functions.

Please refer to the reference guide for ArmPL instruction.

Arm Performance Libraries Reference guide: https://developer.arm.com/Tools%20and%20Software/Arm%20Performance%20Libraries

4.3. LLVM (Arm loign node)¶

Shows how to use the C/C++ compiler based on LLVM. LLVM is available on Arm login nodes.

4.3.1. Environment setting ¶

Use environment of LLVM cross compiler is set using modulefile.

Set the use environment by executing the following before executing compile.

[_LNAlogin]$ module load LLVM-cross-LN

The usable environment is as following.

Software name

Language

Version

Command

Clang

C

10.0.0

clang

Clang++

C++

10.0.0

clang++

4.3.2. To compile ¶

Here is an example of compiling a Fortran program using LLVM.

[Sequential execution]

[_LNAlogin]$ clang -O3  source_file_name

[OpenMP]

[_LNAlogin]$ clang -O3 -fopenmp source_file_name

Table of Contents

Previous topic

Next topic