Setting up Your Environment for Chapel¶
To get started with Chapel, there are three environment settings that are strongly recommended for effective use of the release, and a number of other optional settings that are useful for cross-compiling or overriding the default settings. To check the values of the Chapel environment variables that are set or can be inferred, run the command:
$CHPL_HOME/util/printchplenv --all
The setchplenv.*
source scripts in the $CHPL_HOME/util/quickstart/
and
$CHPL_HOME/util/
directories contain commands that set the following
variables for various shells and host platforms when they are sourced from the
$CHPL_HOME
directory. Frequent Chapel users may want to add such settings
to their shell’s dotfile(s); but for getting started the setchplenv.* scripts
can be convenient.
Recommended Settings¶
CHPL_HOME¶
Setting
CHPL_HOME
is important if you have not installed Chapel and are instead working from a source directory. In that event, set theCHPL_HOME
environment variable to point to the location of the chapel/ directory that was created when you unpacked the release. For example:export CHPL_HOME=~/chapel-1.28.0Note
This, and all other examples in the Chapel documentation, assumes you’re using the
bash
shell. If usingcsh
ortcsh
, mentally changeexport FOO=BAR
tosetenv FOO BAR
. If using some other shell, make the appropriate adjustment.
PATH¶
Updating
PATH
is important if you have not installed Chapel and are instead working from a source directory. Otherwise it might be necessary to use the full path tochpl
when compiling programs. In that event, you can set path using the following command:CHPL_BIN_SUBDIR=`"$CHPL_HOME"/util/chplenv/chpl_bin_subdir.py` export PATH="$PATH":"$CHPL_HOME/bin/$CHPL_BIN_SUBDIR"
Optional Settings¶
CHPL_HOST_PLATFORM¶
You can set the
CHPL_HOST_PLATFORM
environment variable to represent the platform on which you’re working. For standard UNIX workstations, the default is sufficient, and is equivalent toexport CHPL_HOST_PLATFORM=`$CHPL_HOME/util/chplenv/chpl_platform.py`For other platforms that appear very similar to a UNIX workstation from the shell prompt (e.g., a Cray CS™), the value may need to be set explicitly. The strings for our currently-supported host platforms are as follows:
Value
Description
cygwin32
x86 Cygwin (Windows) platforms
cygwin64
x86_64 Cygwin (Windows) platforms
darwin
Macintosh OS X platforms
linux32
32-bit Linux platforms
linux64
64-bit Linux platforms
netbsd32
32-bit NetBSD platforms
netbsd64
64-bit NetBSD platforms
pwr6
IBM Power6 SMP cluster
cray-cs
Cray CS™
cray-xc
Cray XC™
hpe-cray-ex
HPE Cray EX
hpe-apollo
HPE Apollo
Platform-specific documentation is available for most of these platforms in Platform-Specific Notes.
The Chapel Makefiles and sources are designed to work for any UNIX-compatible environment that supports a GNU-compatible make utility. The list above represents the set of platforms that we have access to and can test easily. We are interested in making our code framework portable to other platforms—if you are using Chapel on a platform other than the ones listed above, please contact us for help with the effort.
CHPL_TARGET_PLATFORM¶
If you are cross-compiling for a platform other than your
$CHPL_HOST_PLATFORM
, set theCHPL_TARGET_PLATFORM
environment variable to describe that platform. See CHPL_HOST_PLATFORM above for legal values (though whether or not a given setting will support cross-compilation depends on your specific environment).Note
If
CHPL_TARGET_PLATFORM
is not set, the target platform defaults to the same value as$CHPL_HOST_PLATFORM
.
CHPL_HOST_ARCH¶
Optionally, set the
CHPL_HOST_ARCH
environment variable to indicate the architecture type of the current machine. Normally, the default value is sufficient.
Value
Description
x86_64
64-bit AMD and Intel processors
aarch64
64-bit ARM processors
If unset, the default will be computed. The command
uname -m
should produce the same value as the default.
CHPL_TARGET_ARCH¶
Optionally, set the
CHPL_TARGET_ARCH
environment variable to indicate the architecture type of the target machine. See the table above forCHPL_HOST_ARCH
for values this might be set to.If unset,
CHPL_TARGET_ARCH
will be inferred. IfCHPL_TARGET_CPU
isnative
,unknown
, ornone
thenCHPL_TARGET_ARCH
will be set toCHPL_HOST_ARCH
. Otherwise,CHPL_TARGET_ARCH
will be set based on the architecture type specified inCHPL_TARGET_CPU
.
CHPL_*_COMPILER¶
Optionally, you can set
CHPL_HOST_COMPILER
and/orCHPL_TARGET_COMPILER
to indicate the compiler suite to use in building the sources.CHPL_HOST_COMPILER
is the compiler used to build the Chapel compiler itself so that it will run onCHPL_HOST_PLATFORM
.CHPL_TARGET_COMPILER
is the compiler used to build the runtime libraries and generated code forCHPL_TARGET_PLATFORM
. Currently supported values are as follows:
Value
Description
allinea
The Allinea ARM compiler suite – clang and clang++
clang
The Clang compiler suite – clang and clang++
cray-prgenv-allinea
The Cray PrgEnv compiler using the Allinea backend
cray-prgenv-cray
The Cray PrgEnv compiler using the Cray CCE backend
cray-prgenv-gnu
The Cray PrgEnv compiler using the GNU backend
cray-prgenv-intel
The Cray PrgEnv compiler using the Intel backend
cray-prgenv-pgi
The Cray PrgEnv compiler using the PGI backend
gnu
The GNU compiler suite – gcc and g++
ibm
The IBM compiler suite – xlc and xlC
intel
The Intel compiler suite – icc and icpc
llvm
LLVM code generation
pgi
The PGI compiler suite – pgcc and pgc++
The default for
CHPL_HOST_COMPILER
depends on the value of the correspondingCHPL_HOST_PLATFORM
environment variable:
Host Platform
Compiler
hpe-cray-ex
cray-xc
gnu
darwin
freebsd
clang if available, otherwise gnu
pwr6
ibm
other
gnu
The default for
CHPL_TARGET_COMPILER
is:
llvm
if the compiler is configured with LLVM support (see CHPL_LLVM)
cray-prgenv-$PE_ENV
oncray-xc
andhpe-cray-ex
platforms (wherePE_ENV
is set byPrgEnv-*
modules)
CHPL_HOST_COMPILER
if the host and target platforms are the same
gnu
otherwise.In cases where the LLVM code generation strategy is the default, setting
CHPL_TARGET_COMPILER
to something other thanllvm
will request that the C backend be used with that compiler. For example, to select the C backend with the PrgEnv-gnu compiler, setCHPL_TARGET_COMPILER=cray-prgenv-gnu
.It is sometimes important to be able to provide a particular command to run for C or C++ compilation. The following variables are available to help with that:
Variable
Description
CC
indicates the C compiler to use (but see note below)
CXX
indicates the CXX compiler to use (but see note below)
CHPL_HOST_CC
indicates the C compiler for building
chpl
itselfCHPL_HOST_CXX
indicates the C++ compiler for building
chpl
itselfCHPL_TARGET_CC
indicates the C compiler used by
chpl
CHPL_TARGET_CXX
indicates the C++ compiler used by
chpl
Note
If the
CC
andCXX
variables are set, the other variables in the above table can be inferred. When these variables are used, the following variables can be inferred from them:
CHPL_HOST_COMPILER
,CHPL_HOST_CC
,CHPL_HOST_CXX
CHPL_TARGET_COMPILER
,CHPL_TARGET_CC
,CHPL_TARGET_CXX
However:
Setting any of these inferred variables will disable the inference for all of them
The
*TARGET*
variables above are not inferred fromCC
/CXX
whenCHPL_TARGET_COMPILER=llvm
or when working with a PrgEnv compiler.
CHPL_TARGET_CPU¶
Optionally, set the
CHPL_TARGET_CPU
environment variable to indicate that the target executable should be specialized to the given architecture when using--specialize
(and--fast
). Valid options are:
Value
Description
native
The C compiler will attempt to detect the architecture on the machine that is compiling the target executable. This is a good choice if you will be running on the same machine that you are compiling on. If you are not, see the options below.
unknown
No specialization will be performed
none
No specialization will be performed (will not warn)
Architecture-specific values
intel
amd
arm
core2
k8
aarch64
nehalem
k8sse3
thunderx
westmere
barcelona
thunderx2t99
sandybridge
bdver1
ivybridge
bdver2
haswell
bdver3
broadwell
bdver4
skylake
knl
These values are defined to be the same as in GCC 7:
If you do not want
CHPL_TARGET_CPU
to have any effect, you can set it to eitherunknown
ornone
. Both will disable specialization, but the latter will not warn if--specialize
is used.Setting
CHPL_TARGET_CPU
to an incorrect value for your processor may result in an invalid binary that will not run on the intended machine. Special care should be taken to select the lowest common denominator when running on machines with heterogeneous processor architectures.The default value for this setting will vary based on settings in your environment, in order of application these rules are:
If CHPL_TARGET_COMPILER is
cray-prgenv-*
you do not need to set anything inCHPL_TARGET_CPU
. One of thecraype-*
modules (e.g.craype-sandybridge
) should be loaded to provide equivalent functionality. Once the proper module is loaded,CRAY_CPU_TARGET
will have the architecture being used in it.If
CHPL_TARGET_COMPILER
iscray
,pgi
, oribm
,CHPL_TARGET_CPU
will be set tonone
and no specialization will occur.If CHPL_COMM is set, no attempt to set a useful value will be made and
CHPL_TARGET_CPU
will beunknown
.If CHPL_TARGET_PLATFORM is
darwin
,linux*
, orcygwin*
CHPL_TARGET_CPU
will benative
, passing the responsibility off to the backend C compiler to detect the specifics of the hardware.
CHPL_MAKE¶
Optionally, set the
CHPL_MAKE
environment variable to indicate the GNU-compatible make utility that you want the compiler back-end to invoke when compiling the generated C code. If not set, this will default to a value based on$CHPL_HOST_PLATFORM
:
platform
make utility
cygwin*, darwin
make
linux32, linux64
gmake if available, otherwise make
other
gmake
CHPL_MODULE_PATH¶
Optionally, set the
CHPL_MODULE_PATH
environment variable to provide a list of directories to be added to the Module Search Paths. The value of this environment variable should be a colon-separated list of directory paths.The module search path is used to satisfy ‘use’ statements in the Chapel program. The complete search path can be displayed using the compiler option
--print-search-dirs
. It will also include the compiler’s standard module search paths, those introduced by the-M
flag on the command line and directories containing the .chpl files named explicitly on the compiler command line.
CHPL_LOCALE_MODEL¶
Optionally, set the
CHPL_LOCALE_MODEL
environment variable to indicate the locale model you want to use. Current options are:
Value
Description
flat
top-level locales are not further subdivided
numa
top-level locales are further subdivided into sublocales, each one a NUMA domain
If unset,
CHPL_LOCALE_MODEL
defaults toflat
.Warning
The NUMA locale model is deprecated and will be removed in a future release.
CHPL_TASKS¶
Optionally, set the
CHPL_TASKS
environment variable to indicate what tasking layer you want to use to implement intra-locale parallelism (see Chapel Tasks for more information on this option). Current options are:
Value
Description
qthreads
use Sandia’s Qthreads package
fifo
use POSIX threads
If
CHPL_TASKS
is not set it defaults toqthreads
in all cases except for a few specific configurations in which it defaults tofifo
:
target platform is
cygwin*
target platform is
netbsd*
Note
Note that the Chapel
util/quickstart/setchplenv.*
source scripts setCHPL_TASKS
tofifo
to reduce build-time and third-party dependences, while theutil/setchplenv.*
versions leave it unset, resulting in the behavior described just above.See Chapel Tasks for more information about executing using the various
CHPL_TASKS
options.
CHPL_COMM¶
Optionally, set the
CHPL_COMM
environment variable to indicate what communication layer you want to use to implement inter-locale communication. Current options are:
Value
Description
none
only supports single-locale execution
gasnet
use the GASNet-based communication layer
ofi
use the libfabric-based communication layer
ugni
Cray-specific native communication layer
If unset,
CHPL_COMM
defaults tonone
in most cases. On Cray XC systems it defaults tougni
. On Cray CS systems it defaults togasnet
. See Multilocale Chapel Execution for more information on executing Chapel programs using multiple locales. See Using Chapel with libfabric for more information about the ofi communication layer. See Using Chapel on Cray Systems for more information about Cray-specific runtime layers.
CHPL_MEM¶
Optionally, the
CHPL_MEM
environment variable can be used to select a memory management layer. Current options are:
Value
Description
cstdlib
use the standard C malloc/free commands
jemalloc
use Jason Evan’s memory allocator
If unset,
CHPL_MEM
defaults tojemalloc
for most configurations. If the target platform iscygwin*
it defaults tocstdlib
CHPL_TARGET_MEM
will be replacingCHPL_MEM
in the future.CHPL_TARGET_MEM
takes precedence overCHPL_MEM
.Note
Certain
CHPL_COMM
settings (e.g. ugni, gasnet segment fast/large, ofi with the gni provider) register the heap to improve communication performance. Registering the heap requires special allocator support that not all allocators provide. Currently onlyjemalloc
is capable of supporting configurations that require a registered heap.
CHPL_HOST_MEM¶
Optionally, the
CHPL_HOST_MEM
environment variable can be used to select a memory management layer for thechpl
compiler.
Value
Description
cstdlib
use the standard C malloc/free commands
jemalloc
use Jason Evan’s memory allocator
If unset,
CHPL_HOST_MEM
defaults tocstdlib
.
CHPL_HOST_JEMALLOC¶
Optionally, the
CHPL_HOST_JEMALLOC
environment variable can select between no jemalloc, or using the jemalloc distributed with Chapel in third-party. This setting is intended to elaborate uponCHPL_HOST_MEM=jemalloc
.
Value
Description
none
do not build or use jemalloc
bundled
use the jemalloc distribution bundled with Chapel in third-party
system
use jemalloc found on system. requires
jemalloc-config
in PATHCurrently, the only supported combinations of host target are:
Host
Source
darwin
system
linux
bundled
If unset,
CHPL_HOST_JEMALLOC
defaults to one of the above support combinations, ornone
if CHPL_HOST_MEM iscstdlib
.
CHPL_LAUNCHER¶
Optionally, the
CHPL_LAUNCHER
environment variable can be used to select a launcher to get your program up and running. See Chapel Launchers for more information on this variable’s default and possible settings.
CHPL_ATOMICS¶
Optionally, the
CHPL_ATOMICS
environment variable can be used to select an implementation for atomic operations in the runtime. Current options are:
Value
Description
cstdlib
implement atomics with C standard atomics (from C11)
intrinsics
implement atomics with target compiler intrinsics
locks
implement atomics with mutexes
If
CHPL_ATOMICS
is not set, it defaults tocstdlib
when the target compiler isgnu
,clang
,allinea
,llvm
, orcray
. It defaults tointrinsics
when the target compiler isintel
. It defaults tolocks
when the target compiler ispgi
.See the Chapel Language Specification for more information about atomic operations in Chapel or Runtime Support for Atomics for more information about the runtime implementation.
CHPL_TIMERS¶
Optionally, the
CHPL_TIMERS
environment variable can be used to select an implementation for Chapel’s timers. Current options are:
- generic
use a
gettimeofday()
-based implementationIf unset,
CHPL_TIMERS
defaults togeneric
CHPL_GMP¶
Optionally, the
CHPL_GMP
environment variable can select between no GMP support, using the GMP distributed with Chapel in third-party, or using a system GMP. Current options are:
Value
Description
system
use a system install of GMP (#include gmp.h, -lgmp)
none
do not build GMP support into the Chapel runtime
bundled
use the GMP distribution bundled with Chapel in third-party
If unset, Chapel will attempt to build GMP using CHPL_TARGET_COMPILER (noting that the bundled version may not be supported by all compilers). Based on the outcome, Chapel will default to:
Value
Description
bundled
if the build was successful
system
if unsuccessful and CHPL_TARGET_PLATFORM is cray-x*
none
otherwise
Note
Note that the Chapel
util/quickstart/setchplenv.*
source scripts setCHPL_GMP
tonone
while theutil/setchplenv.*
versions leave it unset, resulting in the behavior described just above.
CHPL_HWLOC¶
Optionally, the
CHPL_HWLOC
environment variable can select between no hwloc support or using the hwloc package distributed with Chapel in third-party.
Value
Description
none
do not build hwloc support into the Chapel runtime
bundled
use the hwloc distribution bundled with Chapel in third-party
If unset,
CHPL_HWLOC
defaults tobundled
if CHPL_TASKS isqthreads
. In all other cases it defaults tonone
. In the unlikely event the bundled hwloc distribution does not build successfully, it should still be possible to use qthreads. To do this, manually setCHPL_HWLOC
tonone
and rebuild (and please file a bug with the Chapel team.) Note that building without hwloc will have a negative impact on performance.
CHPL_RE2¶
Optionally, the
CHPL_RE2
environment variable can be used to enable regular expression operations as defined inRegex
. Current options are:
Value
Description
bundled
use the re2 distribution in third-party
none
do not support regular expression operations
If unset, Chapel will attempt to build RE2 using CHPL_TARGET_COMPILER (noting that the bundled version may not be supported by all compilers). Based on the outcome, Chapel will default to:
Value
Description
bundled
if the build was successful
none
otherwise
Note
Note that the Chapel
util/quickstart/setchplenv.*
source scripts setCHPL_RE2
to'none
while theutil/setchplenv.*
versions leave it unset, resulting in the behavior described just above.
CHPL_AUX_FILESYS¶
Optionally, the
CHPL_AUX_FILESYS
environment variable can be used to request runtime support for certain filesystems.
Value
Description
none
only support traditional Linux filesystems
lustre
enable I/O improvements specific to Lustre filesystems
If unset,
CHPL_AUX_FILESYS
defaults tonone
.
CHPL_LLVM¶
The
CHPL_LLVM
environment variable enables support for the LLVM back-end to the Chapel compiler (see LLVM Support) and support for extern blocks in Chapel code via the Clang compiler (see C Interoperability). Current options are:
Value
Description
bundled
use the llvm/clang distribution in third-party
system
find a compatible LLVM and clang in system libraries; but note that it must be a version supported by Chapel
none
do not support llvm/clang-related features (but note that the LLVM Support library will still be used – see
CHPL_LLVM_SUPPORT
below)unset
indicates that no reasonable default has been inferred, requiring the user to intentionally select another option
If unset,
CHPL_LLVM
defaults to:
none
on linux32 where Chapel LLVM support is not yet implemented
bundled
if you’ve already built the bundled llvm in third-party/llvm
system
if a compatible system-wide installation of LLVM and clang is detected
unset
otherwiseIf CHPL_LLVM is
unset
you will need to either add a system-wide installation of LLVM or setCHPL_LLVM
tobundled
ornone
.See Chapel Prerequisites for more information about currently supported LLVM versions and how to install them. If you are having trouble getting the build system to recognize your system install of LLVM, try setting
CHPL_LLVM=system
and setCHPL_LLVM_CONFIG
to thellvm-config
command from the LLVM version you have installed. Temporarily setting these can help produce a different error message that may may help you to diagnose the problem.
CHPL_LLVM_SUPPORT¶
This variable indicates where to find the LLVM support library. The LLVM support library is required to build the
chpl
compiler. It can only have two values:
Value
Description
bundled
build the LLVM support library from source using the bundled version in third-party
system
use a system-wide install of LLVM to get the LLVM support library
If unset,
CHPL_LLVM_SUPPORT
defaults to the same value asCHPL_LLVM
ifCHPL_LLVM=system
orCHPL_LLVM=bundled
. Otherwise:
system
if a compatible system-wide installation of LLVM is detected
bundled
otherwise
CHPL_LLVM_CONFIG¶
In some cases, it is useful to be able to select a particular LLVM installation for use with
CHPL_LLVM=system
or withCHPL_LLVM_SUPPORT=system
. In that event, in addition to setting one of those variables, you can setCHPL_LLVM_CONFIG
to the llvm-config command from the LLVM installation you wish to use.Inspecting the value of this variable from
printchplenv --all
can also help to identify problems with detection of a system install of LLVM and clang.
CHPL_LLVM_GCC_PREFIX¶
In some cases, the configured
clang
will not work correctly without a--gcc-toolchain
flag. The Chapel compiler tries to infer this flag based upon thegcc
currently available in yourPATH
but sometimes that strategy does not work. As a result, it is sometimes necessary to indicate the path to the GCC libraries. You can setCHPL_LLVM_GCC_PREFIX
tonone
to disable passing the--gcc-toolchain
flag; or you can set it to a particular directory to pass toclang
with the--gcc-toolchain
flag.
CHPL_UNWIND¶
Optionally, the
CHPL_UNWIND
environment variable can be used to select an unwind library for stack tracing. Current options are:
Value
Description
bundled
use the libunwind bundled with Chapel in third-party
system
assume libunwind is already installed on the system
none
don’t use an unwind library, disabling stack tracing
If unset,
CHPL_UNWIND
defaults tonone
CHPL_LIB_PIC¶
Optionally, the
CHPL_LIB_PIC
environment variable can be used to build position independent or position dependent code. This is intended for use when Calling Chapel Code from Other Languages, especially when Using Your Library in Python or when building with--dynamic
. Current options are:
Value
Description
pic
build position independent code
none
use the system default, which might be position independent or position dependent
If unset,
CHPL_LIB_PIC
defaults tonone
Character Set¶
Chapel works with the Unicode character set with the UTF-8 encoding. Chapel programs will use the UTF-8 encoding regardless of the LANG and LC_ALL environment variable settings that the C library uses.
Compiler Command Line Option Defaults¶
Most of the compiler’s command line options support setting a default value for
the option via an environment variable. To see a list of the environment
variables that support each option, run the compiler with the --help-env
flag. For boolean flags and toggles, setting the environment variable to any
value selects that flag.
Chapel Configuration Files¶
The Chapel configuration file is a file named either chplconfig
or
.chplconfig
that can store overrides of the inferred environment variables
listed as a result of executing printchplenv
.
Syntax¶
Below are the valid forms of syntax for Chapel configuration files. All other usages will result in a syntax error.
Definitions
Users can define variables with the following format:
CHPL_ENV=value
Above, the default value of CHPL_ENV
will be overridden to be value
.
All white space is stripped away from definitions.
Ignored Lines
Any lines containing nothing or only white space will be ignored. Comments,
which are denoted by the #
character, similar to bash
or python
,
are also ignored.
Example¶
Below is an example of a Chapel configuration file with comments:
# ~/.chplconfig
# Default to multi-locale
CHPL_COMM=gasnet
CHPL_TASKS=qthreads # Use Qthreads
# System GMP is available on these machines
CHPL_GMP=system
To confirm the configuration file is written correctly, you can run
printchplenv --all --overrides
, which will show a list of variables that are
currently being overridden. Values followed by a
+
have been overridden by the Chapel configuration file, whereas
values followed by a *
have been overridden by an environment variable.
Generating Configuration Files¶
To generate a configuration file based on the current configuration, use
printchplenv
or ./configure
.
When using printchplenv
, run it with the --simple
format flag to get a
format compatible with Chapel configuration files.
The --overrides
filter flag can be used to print only the variables
currently overridden by either environment variables or Chapel
configuration file.
For example, to save the current overrides into a Chapel configuration file:
printchplenv --all --simple --overrides > ~/.chplconfig
The printchplenv --all --simple
flag can be used to print all the variables
of the current configuration. For example:
printchplenv --all --simple > ~/.chplconfig
For more information on using printchplenv
, see the printchplenv -h
output.
Alternatively, the ./configure
script will generate a chplconfig
file. See Installing Chapel.
Search Paths and File Names¶
Though you can put your Chapel configuration file anywhere by setting the
$CHPL_CONFIG
environment variable to its enclosing directory, you can also
place it in your $HOME
or $CHPL_HOME
directory and Chapel will be able to
find it.
The search priority for Chapel configuration files is as follows:
$CHPL_CONFIG
$HOME
(~/
)$CHPL_HOME
When both a chplconfig
and .chplconfig
are present, the visible
chplconfig
will be prioritized.
Only a single chplconfig
file will be used. That is, as soon as a valid
Chapel configuration file is found, the definitions of that file are used.
Note
The $CHPL_CONFIG
variable is the path to the enclosing
directory - not the full path including chplconfig
itself.
Variable Priority¶
Variable precedence goes in the following order:
Explicit compiler flags:
chpl --env=value
Environment variables:
CHPL_ENV=value
Chapel configuration file:
~/.chplconfig
Inferred environment variables:
printchplenv