Arkouda allows a user to interactively issue massively parallel computations on distributed data using functions and syntax that mimic NumPy, the underlying computational library used in the vast majority of Python data science workflows. The computational heart of Arkouda is a Chapel interpreter that accepts a pre-defined set of commands from a client and uses Chapel's built-in machinery for multi-locale and multithreaded execution. Arkouda has benefited greatly from Chapel's distinctive features and has also helped guide the development of the language. For more details see https://github.com/Bears-R-Us/arkouda.
Clone the Arkouda repo. The following steps will take place inside the repo directory.
git clone https://github.com/Bears-R-Us/arkouda.git
cd arkoudaWe recommend using GCC version 11 or later. For example, Spack is an easy way to install GCC 12.1.0:
# Install spack, if you haven't already
git clone https://github.com/spack/spack.git
# Install gcc+binutils, if you haven't already
spack install [email protected]+binutils
spack load [email protected]+binutils
spack compiler findWe will use Miniconda to provide a Python environment and manage Python dependencies.
# Install Miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-aarch64.sh
chmod +x Miniconda3-latest-Linux-aarch64.sh
./Miniconda3-latest-Linux-aarch64.sh -b
# Add Miniconda3 to the environment
source ~/.bashrc
# Make sure you're in the Arkouda repo directory.
cd arkouda
# Developer conda env
conda env create -f arkouda-env-dev.yml
conda activate arkouda-dev
# User conda env
conda env create -f arkouda-env.yml
conda activate arkouda
# Install client and server dependencies
conda install -y jupyter
conda install -y cmake>=3.11.0
# Install the Arkouda Client Package
pip install -e .# Download and unpack the Chapel source in the location where you wish to install Chapel
curl -L https://github.com/chapel-lang/chapel/releases/download/1.27.0/chapel-1.27.0.tar.gz | tar xvzf -
cd chapel-1.27.0
# Set CHPL_HOME
export CHPL_HOME=$PWD
# Add chpl to PATH
source $CHPL_HOME/util/setchplenv.bash
# Set remaining env variables and execute make
# Arkouda documentation recommends adding these variables to a ~/.chplconfig file to prevent having to export them again
cat > ~/.chplconfig <<EOF
CHPL_COMM=gasnet
CHPL_COMM_SUBSTRATE=smp
CHPL_TARGET_CPU=native
CHPL_RE2=bundled
CHPL_LLVM=bundled
EOF
source ~/.chplconfig
# Build Chapel
export GASNET_QUIET=Y
export CHPL_RT_OVERSUBSCRIBED=yes
cd $CHPL_HOME
make -j80 && make -j80 chpldoc
# Add Chapel to your environment
export PATH=$CHPL_HOME/bin/linux64-aarch64:$PATH# Make sure you're in the Arkouda repo directory.
cd arkouda
# Confirm Chapel version 1.25 or higher
(arkouda-dev) [jlinford@amp001 arkouda]$ chpl --version
chpl version 1.27.0
built with LLVM version 14.0.0
Copyright 2020-2022 Hewlett Packard Enterprise Development LP
Copyright 2004-2019 Cray Inc.
(See LICENSE file for more details)
# Edit Makefile.paths and add the location of the Arkouda conda environment
# to the first line as shown below
# For example, `cat Makefile.paths` should produce something like:
# $(eval $(call add-path,/global/home/groups/amp/miniconda3/envs/arkouda))
# ^
# Note that there is no space after 'add-path'
cat Makefile.paths
# Build the server
make -j80Here's an an example of the Chapel pass report for a successful build on an Ampere Altra Q80-30 CPU:
Pass Name Main Check Clean Time % Accum %
---- --------------------------------- ------- ------- ------- ------- ----- ------- -----
42 makeBinary 607.002 0.000 3.069 610.072 47.9 610.072 47.9
13 resolve 462.449 0.204 3.705 466.358 36.6 1076.430 84.6
41 codegen 67.254 0.000 0.219 67.473 5.3 1143.903 89.9
22 parallel 16.878 0.141 0.433 17.452 1.4 1161.355 91.3
20 callDestructors 14.703 0.000 0.367 15.070 1.2 1176.425 92.4
29 copyPropagation 13.750 0.101 1.210 15.061 1.2 1191.486 93.6
26 inlineFunctions 9.320 0.224 2.563 12.107 1.0 1203.593 94.6
36 insertWideReferences 10.182 0.103 0.339 10.623 0.8 1214.216 95.4
33 loopInvariantCodeMotion 10.219 0.086 0.204 10.509 0.8 1224.726 96.2
21 lowerIterators 7.300 0.148 1.128 8.576 0.7 1233.302 96.9
27 scalarReplace 4.158 0.168 1.758 6.084 0.5 1239.386 97.4
28 refPropagation 4.556 0.131 1.364 6.052 0.5 1245.438 97.9
40 denormalize 4.866 0.000 0.878 5.744 0.5 1251.182 98.3
23 prune 1.710 0.123 1.861 3.694 0.3 1254.876 98.6
30 deadCodeElimination 3.328 0.084 0.211 3.622 0.3 1258.498 98.9
37 optimizeOnClauses 2.077 0.103 0.239 2.419 0.2 1260.917 99.1
39 insertLineNumbers 1.164 0.105 0.253 1.522 0.1 1262.439 99.2
34 prune2 1.132 0.086 0.223 1.441 0.1 1263.880 99.3
32 localizeGlobals 0.701 0.084 0.201 0.986 0.1 1264.866 99.4
9 normalize 0.940 0.000 0.031 0.971 0.1 1265.836 99.5
19 lowerErrorHandling 0.749 0.000 0.142 0.891 0.1 1266.728 99.5
18 cullOverReferences 0.620 0.011 0.155 0.786 0.1 1267.513 99.6
17 flattenFunctions 0.629 0.011 0.138 0.777 0.1 1268.291 99.7
1 parse 0.691 0.000 0.013 0.704 0.1 1268.995 99.7
35 returnStarTuplesByRefArgs 0.230 0.086 0.209 0.525 0.0 1269.520 99.8
7 scopeResolve 0.470 0.001 0.027 0.499 0.0 1270.018 99.8
25 removeUnnecessaryAutoCopyCalls 0.032 0.082 0.252 0.366 0.0 1270.385 99.8
38 addInitCalls 0.003 0.103 0.240 0.345 0.0 1270.730 99.9
24 bulkCopyRecords 0.005 0.083 0.243 0.331 0.0 1271.061 99.9
15 checkResolved 0.176 0.000 0.137 0.313 0.0 1271.374 99.9
31 removeEmptyRecords 0.000 0.083 0.196 0.280 0.0 1271.653 99.9
14 resolveIntents 0.122 0.011 0.136 0.269 0.0 1271.922 99.9
11 buildDefaultFunctions 0.131 0.000 0.028 0.159 0.0 1272.081 100.0
16 replaceArrayAccessesWithRefTemps 0.000 0.011 0.136 0.147 0.0 1272.228 100.0
init 0.098 0.000 0.000 0.098 0.0 1272.326 100.0
2 checkParsed 0.076 0.001 0.007 0.085 0.0 1272.410 100.0
6 cleanup 0.045 0.000 0.009 0.055 0.0 1272.465 100.0
12 createTaskFunctions 0.017 0.010 0.023 0.050 0.0 1272.515 100.0
10 checkNormalized 0.009 0.001 0.017 0.026 0.0 1272.541 100.0
driverCleanup 0.019 0.000 0.000 0.019 0.0 1272.560 100.0
5 expandExternArrayCalls 0.001 0.000 0.006 0.007 0.0 1272.568 100.0
8 flattenClasses 0.000 0.000 0.007 0.007 0.0 1272.574 100.0
3 docs 0.000 0.000 0.007 0.007 0.0 1272.581 100.0
4 readExternC 0.000 0.000 0.007 0.007 0.0 1272.587 100.0
startup 0.000 0.000 0.000 0.000 0.0 1272.587 100.0
total time 1247.814 2.384 22.390 1272.587
Use make check to run the builtin correctness checks. For example:
(arkouda) [jlinford@amp001 arkouda]$ make check
INFO:root:Starting "['/global/home/groups/amp/arkouda/arkouda_server', '--trace=false', '--serverConnectionInfo=/global/home/groups/amp/arkouda/ak-server-info', '-nl 2', '--Ser
verPort=5555']"
<jemalloc>: Unsupported system page size
INFO:root:Running client "['python3', '/global/home/groups/amp/arkouda/tests/check.py', 'localhost', '5555']"
_ _ _
/ \ _ __| | _____ _ _ __| | __ _
/ _ \ | '__| |/ / _ \| | | |/ _` |/ _` |
/ ___ \| | | < (_) | |_| | (_| | (_| |
/_/ \_\_| |_|\_\___/ \__,_|\__,_|\__,_|
Client Version: v2022.07.08+6.g9d84d1ea.dirty
<jemalloc>: Unsupported system page size
>>> Sanity checks on the arkouda_server
connected to arkouda server tcp://*:5555
check boolean : Passed
check arange : Passed
check linspace : Passed
check ones : Passed
check zeros : Passed
check argsort : Passed
check coargsort : Passed
check sort : Passed
check get slice [::2] : Passed
check set slice [::2] = value: Passed
check set slice [::2] = pda: Passed
check (compressing) get bool iv : Passed
check (expanding) set bool iv = value: Passed
check (expanding) set bool iv = pda: Passed
check (gather) get integer iv: Passed
check (scatter) set integer iv = value: Passed
check (scatter) set integer iv = pda: Passed
check get integer idx : Passed
check set integer idx = value: Passed
disconnected from arkouda server tcp://*:5555
INFO:root:Running client "['python3', '/global/home/groups/amp/arkouda/util/test/shutdown.py', 'localhost', '5555']"
<jemalloc>: Unsupported system page size
connected to arkouda server tcp://*:5555
Success running checks