Skip to content

Compiler Support

CHAMP supports multiple Fortran and C compilers. The build system automatically detects your compiler and applies appropriate optimization flags.

Supported Compilers

GNU Compiler Collection (GCC)

Minimum version: GCC 9.3.0 or higher

Installation:

Ubuntu/Debian: 

sudo apt-get install -y gfortran gcc

Fedora/RHEL: 

sudo dnf install gcc-gfortran gcc

macOS (Homebrew): 

brew install gcc

Compiler flags auto-applied:

  • -O2 - Optimization level 2
  • -cpp - Enable preprocessor
  • -mcmodel=large - Large memory model support
  • -ffree-line-length-none - No limit on free-form line length
  • -D_MPI_ and -DCLUSTER - MPI and cluster support

Debug mode flags:

  • -fcheck=all - Runtime checks
  • -fbacktrace - Backtrace on errors
  • -g - Debug symbols
  • -Wall -Wextra - Enable warnings

Intel Fortran Compiler (Classic)

Minimum version: Intel Fortran 2020 or higher

Installation:

The Intel Fortran compiler is part of Intel oneAPI HPC Toolkit.

Without sudo access - download installers from Intel oneAPI Downloads:

chmod +x ./l_BaseKit_*.sh
./l_BaseKit_*.sh

chmod +x ./l_HPCKit_*.sh
./l_HPCKit_*.sh

With sudo access:

wget https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
sudo apt-key add GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
sudo add-apt-repository "deb https://apt.repos.intel.com/oneapi all main"
sudo apt-get update
sudo apt-get install intel-oneapi-compiler-fortran intel-oneapi-mkl intel-oneapi-mpi

Loading the compiler: 

source /opt/intel/oneapi/setvars.sh

Compiler flags auto-applied:

  • -O2 - Optimization level 2
  • -ipo - Interprocedural optimization
  • -fpp - Fortran preprocessor
  • -mcmodel=small - Memory model
  • -shared-intel - Dynamic linking for Intel libraries
  • Includes MKL optimizations for Intel10_64lp or Intel10_64_dyn

Intel Fortran Compiler (LLVM-based ifx)

Available in: Intel oneAPI 2021.1 and later

The newer LLVM-based Intel Fortran compiler (ifx) is also supported with similar flags to the classic compiler but without -ipo flag.

Installation: Same as Intel Classic (part of oneAPI HPC Toolkit)

Without sudo access:

chmod +x ./l_BaseKit_*.sh
./l_BaseKit_*.sh

chmod +x ./l_HPCKit_*.sh
./l_HPCKit_*.sh

With sudo access:

wget https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
sudo apt-key add GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
sudo add-apt-repository "deb https://apt.repos.intel.com/oneapi all main"
sudo apt-get update
sudo apt-get install intel-oneapi-compiler-fortran intel-oneapi-mkl intel-oneapi-mpi

Loading the compiler: 

source /opt/intel/oneapi/setvars.sh

LLVM Flang

Compilers supported:

  • Classic Flang (flang)
  • AMD Optimized Flang (AOCC aocc-flang)

Compiler flags applied:

  • -O2 - Optimization level 2
  • -cpp - Preprocessor support
  • -mcmodel=large - Large memory model

Note: Do not confuse with flang-new (LLVMFlang in CMake), which has different compatibility.

NVIDIA HPC SDK (NVHPC)

Use for: GPU-accelerated calculations with OpenACC

Installation: Download from NVIDIA HPC SDK website

Compiler flags applied:

  • -O2 - Optimization
  • -mp=multicore,gpu - OpenMP for multicore and GPU
  • -acc - OpenACC support
  • -gpu=cc80 - Target GPU compute capability
  • -Mcudalib=cublas - Use cuBLAS library
  • CUDA libraries: cuFFT, cuBLAS, cuBLASLt, cuDART

GPU support requirements:

  • CUDA toolkit
  • Set ENABLE_GPU=ON in CMake
  • Optionally set NVFORTRAN_PATH to CUDA executable path

Fujitsu Compiler

Use for: Fugaku supercomputer and other Fujitsu systems

Compiler flags auto-applied:

  • -O3 - High optimization
  • -Cpp - Preprocessor
  • -Kopenmp - OpenMP support
  • -Kparallel -Kfast - Auto-parallelization and fast math
  • -mcmodel=large - Large memory model

Note: Available on Fugaku supercomputer and ARM-based Fujitsu systems

Build Type Configuration

CHAMP supports different build types:

Release (default):

  • Optimized for performance
  • Minimal runtime checks

Debug:

  • Enabled with -DCMAKE_BUILD_TYPE=DEBUG
  • Additional runtime checks and error tracing
  • Debug symbols for debugging with gdb/lldb

Verifying Your Compiler

Check your compiler version:

gfortran --version    # GNU
ifort --version       # Intel Classic
ifx --version         # Intel LLVM
nvfortran --version   # NVIDIA
frt --version         # Fujitsu
flang --version       # LLVM Flang

HPC Systems

On HPC systems, compilers are typically available through module systems:

module avail gcc
module avail intel-compilers
module load gcc/11.2.0

Refer to your system's documentation for specific module names. Specific recipes for a select supercomputers are provided here in the documentation.