Merge branch 'master' into doxygen_doc_fixes

Jenkinsfile

@@ -237,7 +237,7 @@ for(b = 0; b < builderNodes.size(); b++) {
                             virtualenv virtualenv
                             . virtualenv/bin/activate
 
-                            pip install "numpy>1.6, < 1.15"
+                            pip install "numpy>=1.10.0,!=1.16.*"
 
                             python setup.py clean --all
                             python setup.py bdist_wheel -d . 1>> "${uniqueMsg}" 2>> "${uniqueMsg}"
@@ -282,7 +282,7 @@ for(b = 0; b < builderNodes.size(); b++) {
                             call activate
                             popd
 
-                            pip install numpy
+                            pip install "numpy>1.6, < 1.15"
 
                             copy /Y lib\\genn*Release_DLL.* pygenn\\genn_wrapper
                             

bin/genn-buildmodel.bat

@@ -10,6 +10,7 @@ echo genn-buildmodel.bat [cdho] model
 echo -c             only generate simulation code for the CPU
 echo -d             enables the debugging mode
 echo -h             shows this help message
+echo -s             build GeNN without SDL checks
 echo -o outpath     changes the output directory
 echo -i includepath add additional include directories (seperated by semicolons)
 goto :eof
@@ -19,7 +20,7 @@ rem :: define genn-buildmodel.bat options separated by spaces
 rem :: -<option>:              option
 rem :: -<option>:""            option with argument
 rem :: -<option>:"<default>"   option with argument and default value
-set "OPTIONS=-o:"%CD%" -i:"" -d: -c: -h:"
+set "OPTIONS=-o:"%CD%" -i:"" -d: -c: -h: -s:"
 for %%O in (%OPTIONS%) do for /f "tokens=1,* delims=:" %%A in ("%%O") do set "%%A=%%~B"
 
 :genn_option
@@ -54,54 +55,60 @@ if not defined MODEL (
 )
 for /f %%I in ("%-o%") do set "-o=%%~fI"
 for /f %%I in ("%-i%") do set "-i=%%~fI"
-for /f %%I in ("%MODEL%") do set "MACROS=/p:ModelFile=%%~fI /p:GeneratePath=%-o% /p:BuildModelInclude=%-i%"
+for /f %%I in ("%MODEL%") do set "MACROS=/p:ModelFile="%%~fI" /p:GeneratePath="%-o%" /p:BuildModelInclude="%-i%""
 
-if defined -d (
-	set "BACKEND_MACROS= /p:Configuration=Debug"
-	if defined -c (
-		set "BACKEND_PROJECT=single_threaded_cpu_backend"
-		set "MACROS=%MACROS% /p:Configuration=Debug"
-		set GENERATOR=.\generator_Debug.exe
-	) else (
-		set "BACKEND_PROJECT=cuda_backend"
-		set "MACROS=%MACROS% /p:Configuration=Debug_CUDA"
-		set GENERATOR=.\generator_Debug_CUDA.exe
-	)    
+rem set SDL check flag
+if defined -s (
+    set "MACROS=%MACROS% /p:BuildModelSDLCheck=false"
 ) else (
-	set "BACKEND_MACROS= /p:Configuration=Release"
-	if defined -c (
-		set "BACKEND_PROJECT=single_threaded_cpu_backend"
-		set "MACROS=%MACROS% /p:Configuration=Release"
-		set GENERATOR=.\generator_Release.exe
-	) else (
-		set "BACKEND_PROJECT=cuda_backend"
-		set "MACROS=%MACROS% /p:Configuration=Release_CUDA"
-		set GENERATOR=.\generator_Release_CUDA.exe
-	)
+    set "MACROS=%MACROS% /p:BuildModelSDLCheck=true"
 )
 
+if defined -d (
+    set "BACKEND_MACROS= /p:Configuration=Debug"
+    if defined -c (
+        set "BACKEND_PROJECT=single_threaded_cpu_backend"
+        set "MACROS=%MACROS% /p:Configuration=Debug"
+        set GENERATOR=.\generator_Debug.exe
+    ) else (
+        set "BACKEND_PROJECT=cuda_backend"
+        set "MACROS=%MACROS% /p:Configuration=Debug_CUDA"
+        set GENERATOR=.\generator_Debug_CUDA.exe
+    )    
+    ) else (
+    set "BACKEND_MACROS= /p:Configuration=Release"
+    if defined -c (
+        set "BACKEND_PROJECT=single_threaded_cpu_backend"
+        set "MACROS=%MACROS% /p:Configuration=Release"
+        set GENERATOR=.\generator_Release.exe
+    ) else (
+        set "BACKEND_PROJECT=cuda_backend"
+        set "MACROS=%MACROS% /p:Configuration=Release_CUDA"
+        set GENERATOR=.\generator_Release_CUDA.exe
+    )
+)
 
 rem :: build backend
 msbuild "%GENN_PATH%..\genn.sln" /m /verbosity:minimal /t:%BACKEND_PROJECT% %BACKEND_MACROS% /p:BuildProjectReferences=true && (
-	echo Successfully built GeNN
+    echo Successfully built GeNN
 ) || (
-	echo Unable to build GeNN
-	goto :eof
+    echo Unable to build GeNN
+    goto :eof
 )
 
 
 rem :: build generator
 msbuild "%GENN_PATH%..\src\genn\generator\generator.vcxproj" /m /verbosity:minimal %MACROS%&& (
-	echo Successfully built code generator
+    echo Successfully built code generator
 ) || (
-	echo Unable to build code generator
-	goto :eof
+    echo Unable to build code generator
+    goto :eof
 )
 
 if defined -d (
-	devenv /debugexe "%GENERATOR%" "%-o%"
+    devenv /debugexe "%GENERATOR%" "%-o%"
 ) else (
-	"%GENERATOR%" "%-o%"
+    "%GENERATOR%" "%-o%"
 )
 
 echo Model build complete

doxygen/00_MainPage.dox

@@ -18,7 +18,7 @@ GeNN is a software package to enable neuronal network simulations on
 NVIDIA GPUs by code generation. Models are defined in a simple C-style
 API and the code for running them on either GPU or CPU hardware is
 generated by GeNN. GeNN can also be used through external
-interfaces. Currently there are interfaces for \ref SpineML and for [Brian](http://brian2.readthedocs.io) via \ref brian2genn "Brian2GeNN".
+interfaces. Currently there are interfaces for \ref SpineML, a \ref PyGeNN and a [Brian](http://brian2.readthedocs.io) interface via \ref brian2genn "Brian2GeNN".
 
 GeNN is currently developed and maintained by
 

doxygen/06_Brian2GeNN.dox

@@ -1,9 +1,8 @@
 //----------------------------------------------------------------------------
 /*! \page brian2genn Brian interface (Brian2GeNN)
 
-GeNN can simulate models written for the [Brian simulator](http://brian2.readthedocs.io) via the [Brian2GeNN](http://brian2genn.readthedocs.io) interface \cite  brian2genn2018 .
-The easiest way to install everything needed is to install the [Anaconda](https://www.anaconda.com/download) or [Miniconda](https://conda.io/miniconda.html) Python distribution and then follow the [instructions to install Brian2GeNN](http://brian2genn.readthedocs.io/en/latest/introduction/index.html#installing-the-brian2genn-interface) with the conda package manager.
-When Brian2GeNN is installed in this way, it comes with a bundled version of GeNN and no further configuration is required. In all other cases (e.g. an installation from source), the path to GeNN and the CUDA libraries has to be configured via the `GENN_PATH` and `CUDA_PATH` environment variables as described in \ref Installation or via the `devices.genn.path` and `devices.genn.cuda_path` [Brian preferences](http://brian2.readthedocs.io/en/stable/advanced/preferences.html).
+GeNN can simulate models written for the [Brian simulator](http://brian2.readthedocs.io) via the [Brian2GeNN](http://brian2genn.readthedocs.io) interface \cite  brian2genn2018.
+In order for Brian2GeNN to find GeNN, GeNN must either be in your path with the location of the CUDA libraries configured via the `CUDA_PATH` environment variable (as described in \ref Installation) or the `devices.genn.path` and `devices.genn.cuda_backend.cuda_path` [Brian preferences](http://brian2.readthedocs.io/en/stable/advanced/preferences.html) should be set.
 
 To use GeNN to simulate a Brian script, import the `brian2genn` package and switch Brian to the `genn` device.
 As an example, the following Python script will simulate Leaky-integrate-and-fire neurons with varying input currents to construct an f/I curve:

doxygen/09_ReleaseNotes.dox

@@ -1,4 +1,39 @@
 /*! \page ReleaseNotes Release Notes
+Release Notes for GeNN v4.2.1 (PyGeNN 0.3.1)
+====
+This release fixes several small issues including several relating to Brian2GeNN compatibility.
+
+User Side Changes
+----
+1. Added -s option to genn-buildmodel.bat on Windows to turn off Visual C++ additional security checks (SDL), allowing Brian2GeNN libraries to be included in code generator.
+
+Bug fixes:
+----
+1. Fixed bug where \$(sT_pre) and \$(sT_post) were incorrect when accessed in weight update model pre and postsynaptic spike code respectively when using the single-threaded CPU backend.
+2. Fixed a corner case where valid models might result in compiler errors about ``Isyn`` not being defined.
+3. Fixed a bug preventing multiple include paths being passed to genn-buildmodel.bat on Windows.
+
+Release Notes for GeNN v4.2.0 (PyGeNN 0.3)
+====
+This release adds a number of new features to GeNN and its Python interface as well as fixing a number of bugs that have been identified since the 4.1.0 release.
+
+User Side Changes
+----
+1. Kernel timings can now be enabled from python with ``pygenn.genn_model.GeNNModel.timing_enabled`` and subsequently accessed with ``pygenn.genn_model.GeNNModel.neuron_update_time``, ``pygenn.genn_model.GeNNModel.init_time``, ``pygenn.genn_model.GeNNModel.presynaptic_update_time``, ``pygenn.genn_model.GeNNModel.postsynaptic_update_time``, ``pygenn.genn_model.GeNNModel.synapse_dynamics_time`` and ``pygenn.genn_model.GeNNModel.init_sparse_time``.
+2. Backends now generate ``getFreeDeviceMemBytes()`` function to allow free device memory to be queried from user simulation code. This is also exposed to Python via ``GeNNModel.free_device_mem_bytes`` property.
+3. GeNN preferences are now fully exposed to PyGeNN by passing kwargs to ``pygenn.genn_model.GeNNModel.__init__``.
+4. Logging level can now be seperately specified for GeNN, the code generator, the SpineML generator and the backend and is accessible from PyGeNN.
+5. ``CodeGenerator::PreferencesBase::enableBitmaskOptimisations`` flag enables an alternative algorithm for updating synaptic matrices implemented with ``SynapseMatrixConnectivity::BITMASK`` which performs better on smaller GPUs and CPUs. If you are manually initialising matrices this adds padding to align words to rows of the matrix.
+6. ``SynapseMatrixConnectivity::PROCEDURAL`` and ``SynapseMatrixWeight::PROCEDURAL`` allow connectivity and synaptic weights to be generated on the fly rather than stored in memory.
+7. ``CodeGenerator::PreferencesBase::automaticCopy`` flag allows models to be built without the need for explicitly copying data between host and device. For CUDA backend this uses unified memory (https://devblogs.nvidia.com/unified-memory-cuda-beginners/).
+8. Speed of code compilation can be improved by building using multiple threads. This is now done everywhere where make or MSBuild is invocated automatically.
+
+Bug fixes:
+----
+1. Fixed several bugs in extra global parameter implementation in PyGeNN.
+2. Floating point min and max should be calculated with ``fmax`` and ``fmin`` in code snippets - fixed in several models and user projects.
+3. Fixed issues with version of numpy required in PyGeNN (previously held back by an issue with PyNN)
+
 Release Notes for GeNN v4.1.0
 ====
 This release adds a number of new features to GeNN and its SpineML interface as well as fixing a number of bugs that have been identified since the 4.0.2 release.

generate_swig_interfaces.py

@@ -406,8 +406,9 @@ def generateBackend(swigPath, folder, namespace):
         # Add plog severity header so logging severities can be passed to backend
         mg.addSwigInclude('<plog/Severity.h>')
 
-        # Parse backend base, ignore BackendBase itself to get PreferencesBase definition
+        # Parse backend base, ignore BackendBase itself and MemAlloc to get PreferencesBase definition
         mg.addSwigIgnore("BackendBase")
+        mg.addSwigIgnore("MemAlloc")
         mg.addSwigInclude('"code_generator/backendBase.h"')
 
         # Parse backend, ignore Backend itself to get PreferencesBase definition
@@ -500,18 +501,27 @@ def generateConfigs(gennPath, backends):
             for header in (NEURONMODELS, POSTSYNMODELS,
                            WUPDATEMODELS, CURRSOURCEMODELS, INITVARSNIPPET, SPARSEINITSNIPPET):
                 pygennSmg.addCppInclude( '"' + header + 'Custom.h"' )
+            pygennSmg.addCppInclude( '"code_generator/backendBase.h"' )
             pygennSmg.addCppInclude( '"code_generator/generateAll.h"' )
             pygennSmg.addCppInclude( '"code_generator/generateMakefile.h"' )
             pygennSmg.addCppInclude( '"code_generator/generateMSBuild.h"' )
             pygennSmg.addCppInclude( '"path.h"' )
         pygennSmg.addSwigImport( '"StlContainers.i"' )
 
+        # Enable C++ camelCase to python underscore_case conversion
+        pygennSmg.addSwigEnableUnderCaseConvert()
+
         # Include genn export header so GENN_EXPORT macros can be correctly parsed
         pygennSmg.addSwigInclude( '"gennExport.h"' )
 
         # Add plog severity header so logging severities can be passed to backend
         pygennSmg.addSwigInclude('<plog/Severity.h>')
 
+        # Add backend base header but ignore everything apart from MemAlloc
+        pygennSmg.addSwigIgnore("BackendBase")
+        pygennSmg.addSwigIgnore("PreferencesBase")
+        pygennSmg.addSwigInclude('"code_generator/backendBase.h"')
+
         # define and wrap three functions which replace main in generateALL.cc
         with SwigInlineScope( pygennSmg ):
             pygennSmg.write( '''
@@ -521,22 +531,23 @@ def generateConfigs(gennPath, backends):
                 Logging::init(gennLevel, codeGeneratorLevel, consoleAppender, consoleAppender);
             }
 
-            void generate_code(ModelSpecInternal &model, CodeGenerator::BackendBase &backend, const std::string &path, int localHostID)
+            CodeGenerator::MemAlloc generate_code(ModelSpecInternal &model, CodeGenerator::BackendBase &backend, const std::string &path, int localHostID)
             {
                 const filesystem::path outputPath(path);
 
                 // Generate code, returning list of module names that must be build
-                const auto moduleNames = CodeGenerator::generateAll(model, backend, outputPath);
+                const auto output = CodeGenerator::generateAll(model, backend, outputPath);
 
             #ifdef _WIN32
                 // Create MSBuild project to compile and link all generated modules
                 std::ofstream makefile((outputPath / "runner.vcxproj").str());
-                CodeGenerator::generateMSBuild(makefile, backend, "", moduleNames);
+                CodeGenerator::generateMSBuild(makefile, backend, "", output.first);
             #else
                 // Create makefile to compile and link all generated modules
                 std::ofstream makefile((outputPath / "Makefile").str());
-                CodeGenerator::generateMakefile(makefile, backend, moduleNames);
+                CodeGenerator::generateMakefile(makefile, backend, output.first);
             #endif
+                return output.second;
             }
             ''' )
 
@@ -621,8 +632,6 @@ def generateConfigs(gennPath, backends):
             %}''')
 
         # wrap NeuronGroup, SynapseGroup and CurrentSource
-        pygennSmg.addSwigEnableUnderCaseConvert()
-
         pygennSmg.addSwigInclude( '"neuronGroup.h"' )
         pygennSmg.addSwigInclude( '"synapseGroup.h"' )
         pygennSmg.addSwigInclude( '"currentSource.h"' )
@@ -669,7 +678,19 @@ def generateConfigs(gennPath, backends):
                 'ModelSpec::addSynapsePopulation<WeightUpdateModels::{0}, PostsynapticModels::{1}>'.format(
                     wu_model, ps_model),
                 'add_synapse_population_{}_{}'.format(wu_model, ps_model))
-
+
+        # Loop through all postsynaptic models
+        for ps_model in mgs[1].models:
+        #SynapseGroup *addSlaveSynapsePopulation(const std::string &name, const std::string &weightSharingMasterName, unsigned int delaySteps, const std::string &src, const std::string &trg,
+                                            #const typename PostsynapticModel::ParamValues &postsynapticParamValues, const typename PostsynapticModel::VarValues &postsynapticVarInitialisers)
+            # Ignore the overload of the functions which automatically get instance from class name
+            pygennSmg.addSwigIgnore("ModelSpec::addSlaveSynapsePopulation<PostsynapticModels::{0}>(std::string const &, std::string const &,unsigned int,std::string const &,std::string const &,PostsynapticModels::{0}::ParamValues const &,PostsynapticModels::{0}::VarValues const &)".format(ps_model))
+
+            # Add template expansion
+            pygennSmg.addSwigTemplate(
+                'ModelSpec::addSlaveSynapsePopulation<PostsynapticModels::{0}>'.format(ps_model),
+                'add_slave_synapse_population_{}'.format(ps_model))
+
         for cs_model in mgs[3].models:
             # Ignore the overload of the function which automatically gets instance from class name
             pygennSmg.addSwigIgnore("ModelSpec::addCurrentSource<CurrentSourceModels::{0}>(std::string const &,std::string const &,CurrentSourceModels::{0}::ParamValues const &,CurrentSourceModels::{0}::VarValues const &)".format(cs_model))