fix HPC build, documentation

.github/workflows/c-cpp.yml

@@ -29,7 +29,7 @@ jobs:
       run: meson setup build
     - name: configure
       working-directory: ./core/source/build
-      run:  meson configure -Domp=true -Docl=false
+      run:  meson configure -Domp=true -Docl=false -Dhpc=false
     - name: build
       working-directory: ./core/source/build
       run: meson compile

.vscode/c_cpp_properties.json

@@ -23,6 +23,29 @@
             "intelliSenseMode": "macos-clang-x64",
             "configurationProvider": "mesonbuild.mesonbuild"
         },
+        {
+            "name": "BrewGCC-ARM",
+            "includePath": [
+                "${workspaceFolder}/core/source/**"
+            ],
+            "defines": ["USE_DESKTOP","USE_OPENMP","VBITS=128"],
+            "compilerArgs": [
+                "-I${workspaceFolder}/core/source/base",
+                "-I${workspaceFolder}/core/source/concurrency",
+                "-I${workspaceFolder}/core/source/io",
+                "-I${workspaceFolder}/core/source/particles",
+                "-I${workspaceFolder}/core/source/solver",
+                "-I${workspaceFolder}/core/source/sparc"
+            ],
+            "macFrameworkPath": [
+                "/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/System/Library/Frameworks"
+            ],
+            "compilerPath": "/opt/homebrew/bin/g++-13",
+            "cStandard": "c17",
+            "cppStandard": "c++17",
+            "intelliSenseMode": "macos-clang-x64",
+            "configurationProvider": "mesonbuild.mesonbuild"
+        },
         {
             "name": "WindowsLLVM",
             "includePath": [

core/examples/pgc/beatwave-3d.tw

@@ -2,6 +2,7 @@
 // NERSC Edison: 11 nodes, 7.5 minutes
 // NERSC Cori-Haswell: 8 nodes, 7.2 minutes
 // NERSC Cori-KNL: 8 nodes, 10 minutes
+// NERSC Perlmutter-CPU: 2 nodes, 10 minutes
 steps = 1000
 timestep = 0.08
 dump period = 0
@@ -71,7 +72,7 @@ new species 'e'
 	mass = 1
 	emission temperature = ( 0.1 , 0.1 , 0.1 )
 	particles per cell = ( 2 , 2 , 4 ) when density = 1
-	zboundary = emitting emitting
+	zboundary = (emitting emitting)
 }
 
 generate uniform 'e'

core/source/base/Main.cpp

@@ -263,7 +263,7 @@ void CommandHandler::Evaluate(const Arg& arg,
 // (used with internal MPI) //
 //////////////////////////////
 
-#ifdef USING_TW_MPI
+#ifdef USE_TW_MPI
 
 struct Launcher : tw::Thread
 {
@@ -328,7 +328,7 @@ void TW_Interactive::Run()
 // Standard MPI Launch //
 /////////////////////////
 
-#ifndef USING_TW_MPI
+#ifndef USE_TW_MPI
 int main(int argc,char *argv[])
 {
 	int numOMPThreads=0; // indicates -c argument was not given
@@ -382,7 +382,7 @@ int main(int argc,char *argv[])
 // Internal MPI Launch //
 /////////////////////////
 
-#ifdef USING_TW_MPI
+#ifdef USE_TW_MPI
 int main(int argc,char *argv[])
 {
 	int numMPIThreads=1,numOMPThreads=1;

core/source/base/Simulation.cpp

@@ -214,7 +214,7 @@ void Simulation::SetupIO()
 
 void Simulation::Run()
 {
-	#ifdef USING_TW_MPI
+	#ifdef USE_TW_MPI
 	// Wait until master thread is done
 	// TODO: fold this into TW_MPI's MPI_init
 	TW_MPI_Lock();
@@ -306,7 +306,7 @@ void Simulation::Run()
 
 void Simulation::Test()
 {
-	#ifdef USING_TW_MPI
+	#ifdef USE_TW_MPI
 	// Wait until master thread is done
 	// TODO: fold this into TW_MPI's MPI_init
 	TW_MPI_Lock();

core/source/concurrency/Tasks.cpp

@@ -218,9 +218,9 @@ void Task::Initialize(tw::Int *doms,tw::Int *gcells,tw::Int *cyclic)
 	gpu = devices[whichDevice];
 	context = clCreateContext(NULL,1,&gpu,NULL,NULL,&err);
 	// Following is the updated function
-	commandQueue = clCreateCommandQueueWithProperties(context,gpu,NULL,&err);
+	//commandQueue = clCreateCommandQueueWithProperties(context,gpu,NULL,&err);
 	// Following is the deprecated function
-	//commandQueue = clCreateCommandQueue(context,gpu,0,&err);
+	commandQueue = clCreateCommandQueue(context,gpu,0,&err);
 
 	// Create kernels
 	std::string buildLog;
@@ -278,7 +278,10 @@ void Task::InitializeCLProgram(cl_program& program,const std::string& fileName,s
 	}
 	std::string header_and_source = CLDefinitions(gpu) + sourceString;
 	sourceText.resize(header_and_source.size()+1);
-	strcpy_s(&sourceText[0],sourceText.size(),header_and_source.c_str());
+	// Following is the updated function
+	//strcpy_s(&sourceText[0],sourceText.size(),header_and_source.c_str());
+	// Following is the deprecated function
+	strcpy(&sourceText[0],header_and_source.c_str());
 	sourceList[0] = &sourceText[0];
 
 	program = clCreateProgramWithSource(context,1,(const char**)&sourceList[0],NULL,&err);

core/source/concurrency/tasks.h

@@ -245,19 +245,19 @@ struct Task
 
 	void Lock()
 	{
-		#ifdef USING_TW_MPI
+		#ifdef USE_TW_MPI
 		TW_MPI_Lock();
 		#endif
 	}
 	void Unlock()
 	{
-		#ifdef USING_TW_MPI
+		#ifdef USE_TW_MPI
 		TW_MPI_Unlock();
 		#endif
 	}
 	void Complete()
 	{
-		#ifdef USING_TW_MPI
+		#ifdef USE_TW_MPI
 		TW_MPI_ThreadRef(strip[0].comm_cart,strip[0].Get_rank())->Complete();
 		#endif
 	}

core/source/concurrency/tw_mpi.h

@@ -1,6 +1,6 @@
 // TurboWAVE non-conforming implementation of MPI
 
-#define USING_TW_MPI
+#define USE_TW_MPI
 
 static const int MAX_THREADS = 256;
 static const int MPI_PROC_NULL = -3;

core/source/meson.build

@@ -1,7 +1,20 @@
-project('turbowave','cpp')
+project('turbowave','cpp','c',
+    meson_version: '>=1.1.0',  
+    version: '"4.8.0"',
+    default_options: ['cpp_std=c++17','buildtype=release']
+)
+
+# Meson philosophy is to give the source list explicitly.
+# We can always copy and paste from a glob elsewhere.
+src_base = ['base/3DFields.cpp', 'base/ComputeTool.cpp', 'base/DiscreteSpace.cpp', 'base/FFT.cpp', 'base/Functions.cpp', 'base/Injection.cpp', 'base/Injection_Test.cpp', 'base/Main.cpp', 'base/MetricSpace.cpp', 'base/MetricSpace_Test.cpp', 'base/Module.cpp', 'base/Region.cpp', 'base/Simulation.cpp']
+src_con = ['concurrency/Tasks.cpp']
+src_io = ['io/Diagnostics.cpp', 'io/Input.cpp', 'io/Units.cpp']
+src_par = ['particles/Bundle.cpp', 'particles/Mover.cpp', 'particles/Mover_Test.cpp', 'particles/Particles.cpp', 'particles/Particles_Test.cpp', 'particles/Pusher.cpp', 'particles/Pusher_Test.cpp', 'particles/Slicer.cpp', 'particles/Tiler.cpp']
+src_solv = ['solver/Electrostatic.cpp', 'solver/Elliptic.cpp', 'solver/Elliptic_Test.cpp', 'solver/FieldSolve.cpp', 'solver/Hyperbolic.cpp', 'solver/LaserSolve.cpp', 'solver/Numerics.cpp', 'solver/Parabolic.cpp', 'solver/Qstate.cpp', 'solver/Quantum.cpp', 'solver/SolidState.cpp']
+src_sparc = ['sparc/Chemistry.cpp', 'sparc/Fluid.cpp', 'sparc/Fluid_Test.cpp', 'sparc/Physics.cpp', 'sparc/Physics_Test.cpp']
 
 conf_data = configuration_data()
-conf_data.set('TW_VERSION_STRING','"4.8.0"')
+conf_data.set('TW_VERSION_STRING',meson.project_version())
 conf_data.set('BIGENDIAN',get_option('bigendian'))
 conf_data.set('VBITS',get_option('vbits'))
 conf_data.set('USE_DESKTOP',not get_option('hpc'))
@@ -22,25 +35,19 @@ if get_option('omp')
 endif
 if get_option('ocl')
     deps += [dependency('opencl')]
-    conf_data.set('CL_TARGET_OPENCL_VERSION','300')
+    conf_data.set('CL_TARGET_OPENCL_VERSION','120')
+endif
+if get_option('hpc')
+    deps += [dependency('mpi')]
+else
+    src_con += ['concurrency/TW_MPI.cpp']
 endif
-
-# Meson philosophy is to give the source list explicitly.
-# We can always copy and paste from a glob elsewhere.
-src_base = ['base/3DFields.cpp', 'base/ComputeTool.cpp', 'base/DiscreteSpace.cpp', 'base/FFT.cpp', 'base/Functions.cpp', 'base/Injection.cpp', 'base/Injection_Test.cpp', 'base/Main.cpp', 'base/MetricSpace.cpp', 'base/MetricSpace_Test.cpp', 'base/Module.cpp', 'base/Region.cpp', 'base/Simulation.cpp']
-src_con = ['concurrency/Tasks.cpp', 'concurrency/TW_MPI.cpp']
-src_io = ['io/Diagnostics.cpp', 'io/Input.cpp', 'io/Units.cpp']
-src_par = ['particles/Bundle.cpp', 'particles/Mover.cpp', 'particles/Mover_Test.cpp', 'particles/Particles.cpp', 'particles/Particles_Test.cpp', 'particles/Pusher.cpp', 'particles/Pusher_Test.cpp', 'particles/Slicer.cpp', 'particles/Tiler.cpp']
-src_solv = ['solver/Electrostatic.cpp', 'solver/Elliptic.cpp', 'solver/Elliptic_Test.cpp', 'solver/FieldSolve.cpp', 'solver/Hyperbolic.cpp', 'solver/LaserSolve.cpp', 'solver/Numerics.cpp', 'solver/Parabolic.cpp', 'solver/Qstate.cpp', 'solver/Quantum.cpp', 'solver/SolidState.cpp']
-src_sparc = ['sparc/Chemistry.cpp', 'sparc/Fluid.cpp', 'sparc/Fluid_Test.cpp', 'sparc/Physics.cpp', 'sparc/Physics_Test.cpp']
-src_all = src_base + src_con + src_io + src_par + src_solv + src_sparc
 
 configure_file(output: 'config.h', configuration: conf_data)
 executable(
     'tw3d',
-    src_all,
+    src_base + src_con + src_io + src_par + src_solv + src_sparc,
     include_directories : incdir,
-    cpp_args: ['-std=c++17'],
     dependencies: deps,
     install: true
 )

docs/index.rst

@@ -53,7 +53,6 @@ User Reference
 	ref-fluid
 	ref-QO
 	ref-twutils
-	ref-boot
 
 Developer Reference
 -------------------

docs/install-easy.rst

@@ -20,15 +20,15 @@ The following should work in any operating system, assuming compilers are instal
 
 #. Use the installer to complete the sequence of steps in the ``Tasks`` area.
 
-	* You can usually accept default responses.
-	* On Cray, the platform and vector type popups should be set manually.
-	* For now use OpenMP for the accelerator.
-	* The installer can configure for GPGPU, but you may need to fulfill some prerequisites as root for the compiler to succeed.
-
 .. note:: Pretty PowerShell
 
 	To get color working try ``Set-ItemProperty HKCU:\Console VirtualTerminalLevel -Type DWORD 1``.  To get Unicode characters working try ``$OutputEncoding = [console]::InputEncoding = [console]::OutputEncoding = New-Object System.Text.UTF8Encoding`` (put this in the ``$PROFILE`` file to make it persist). If unicode is still not displayed correctly, start cycling fonts (right-click title bar and choose ``Properties``).
 
+Runtime
+-------
+
+When ``twinstall`` is used, the executable installs to the conda environment.  As a result, when outside the environment, the executable will not be found.  However, one should understand that the executable itself does not depend on Anaconda or Python in any way.  It can be manually installed anywhere.
+
 Upgrade
 -------