Began implementing CudaPlatform

olla/include/KernelFactory.h

@@ -48,8 +48,9 @@ public:
      * Create a KernelImpl.
      * 
      * @param name the name of the kernel to create
+     * @param context the context the kernel will belong to
      */
-    virtual KernelImpl* createKernelImpl(std::string name, const Platform& platform) const = 0;
+    virtual KernelImpl* createKernelImpl(std::string name, const Platform& platform, OpenMMContextImpl& context) const = 0;
     virtual ~KernelFactory() {
     }
 };

olla/include/Platform.h

@@ -40,6 +40,7 @@ namespace OpenMM {
 
 class Kernel;
 class KernelFactory;
+class OpenMMContextImpl;
 class StreamFactory;
 
 /**
@@ -83,6 +84,16 @@ public:
      * different StreamFactory has not been registered for the requested stream name.
      */
     virtual const StreamFactory& getDefaultStreamFactory() const = 0;
+    /**
+     * This is called whenever a new OpenMMContext is created.  It gives the Platform a chance to initialize
+     * the context and store platform-specific data in it.
+     */
+    virtual void contextCreated(OpenMMContextImpl& context) const;
+    /**
+     * This is called whenever an OpenMMContext is deleted.  It gives the Platform a chance to clean up
+     * any platform-specific data that was stored in it.
+     */
+    virtual void contextDestroyed(OpenMMContextImpl& context) const;
     /**
      * Register a KernelFactory which should be used to create Kernels with a particular name.
      * The Platform takes over ownership of the factory, and will delete it when the Platform itself
@@ -110,19 +121,20 @@ public:
      */
     bool supportsKernels(std::vector<std::string> kernelNames) const ;
     /**
-     * Create a Kernel object.  If you call this method multiple times with the same name,
+     * Create a Kernel object.  If you call this method multiple times for different contexts with the same name,
      * the returned Kernels are independent and do not interact with each other.  This means
      * that it is possible to have multiple simulations in progress at one time without them
      * interfering.
      * 
      * If no KernelFactory has been registered for the specified name, this will throw an exception.
      * 
-     * @param the name of the Kernel to get
+     * @param name the name of the Kernel to get
+     * @param context the context for which to create a Kernel
      * @return a newly created Kernel object
      */
-    Kernel createKernel(std::string name) const;
+    Kernel createKernel(std::string name, OpenMMContextImpl& context) const;
     /**
-     * Create a Stream object.  If you call this method multiple times with the same name,
+     * Create a Stream object.  If you call this method multiple times for different contexts with the same name,
      * the returned Streams are independent and do not interact with each other.  This means
      * that it is possible to have multiple simulations in progress at one time without them
      * interfering.
@@ -130,10 +142,11 @@ public:
      * If a StreamFactory has been registered for the specified name, it will be used to create
      * the Stream.  Otherwise, the default StreamFactory will be used.
      * 
-     * @param the name of the Stream to get
+     * @param name the name of the Stream to get
+     * @param context the context for which to create a Stream
      * @return a newly created Stream object
      */
-    Stream createStream(std::string name, int size, Stream::DataType type) const;
+    Stream createStream(std::string name, int size, Stream::DataType type, OpenMMContextImpl& context) const;
     /**
      * Register a new Platform.
      */

olla/include/StreamFactory.h

@@ -50,8 +50,9 @@ public:
      * @param name the name of the stream to create
      * @param size the number of elements in the stream
      * @param type the data type of each element in the stream
+     * @param context the context the kernel will belong to
      */
-    virtual StreamImpl* createStreamImpl(std::string name, int size, Stream::DataType type, const Platform& platform) const = 0;
+    virtual StreamImpl* createStreamImpl(std::string name, int size, Stream::DataType type, const Platform& platform, OpenMMContextImpl& context) const = 0;
     virtual ~StreamFactory() {
     }
 };

olla/src/Platform.cpp

@@ -68,6 +68,12 @@ Platform::~Platform() {
         delete *iter;
 }
 
+void Platform::contextCreated(OpenMMContextImpl& context) const {
+}
+
+void Platform::contextDestroyed(OpenMMContextImpl& context) const {
+}
+
 void Platform::registerKernelFactory(std::string name, KernelFactory* factory) {
     kernelFactories[name] = factory;
 }
@@ -83,16 +89,16 @@ bool Platform::supportsKernels(std::vector<std::string> kernelNames) const {
     return true;
 }
 
-Kernel Platform::createKernel(std::string name) const {
+Kernel Platform::createKernel(std::string name, OpenMMContextImpl& context) const {
     if (kernelFactories.find(name) == kernelFactories.end())
         throw PlatformException("Called createKernel() on a Platform which does not support the requested kernel");
-    return Kernel(kernelFactories.find(name)->second->createKernelImpl(name, *this));
+    return Kernel(kernelFactories.find(name)->second->createKernelImpl(name, *this, context));
 }
 
-Stream Platform::createStream(std::string name, int size, Stream::DataType type) const {
+Stream Platform::createStream(std::string name, int size, Stream::DataType type, OpenMMContextImpl& context) const {
     if (streamFactories.find(name) == streamFactories.end())
-        return Stream(getDefaultStreamFactory().createStreamImpl(name, size, type, *this));
-    return Stream(streamFactories.find(name)->second->createStreamImpl(name, size, type, *this));
+        return Stream(getDefaultStreamFactory().createStreamImpl(name, size, type, *this, context));
+    return Stream(streamFactories.find(name)->second->createStreamImpl(name, size, type, *this, context));
 }
 
 void Platform::registerPlatform(Platform* platform) {

openmmapi/include/internal/OpenMMContextImpl.h

@@ -148,6 +148,14 @@ public:
      * Delete all ForceImpl objects that have been created and create new ones.
      */
     void reinitialize();
+    /**
+     * Get the platform-specific data stored in this context.
+     */
+    void* getPlatformData();
+    /**
+     * Set the platform-specific data stored in this context.
+     */
+    void setPlatformData(void* data);
 private:
     friend class OpenMMContext;
     OpenMMContext& owner;
@@ -159,6 +167,7 @@ private:
     Platform* platform;
     Stream positions, velocities, forces;
     Kernel kineticEnergyKernel;
+    void* platformData;
 };
 
 } // namespace OpenMM

openmmapi/src/AndersenThermostatImpl.cpp

@@ -43,7 +43,7 @@ AndersenThermostatImpl::AndersenThermostatImpl(AndersenThermostat& owner) : owne
 }
 
 void AndersenThermostatImpl::initialize(OpenMMContextImpl& context) {
-    kernel = context.getPlatform().createKernel(ApplyAndersenThermostatKernel::Name());
+    kernel = context.getPlatform().createKernel(ApplyAndersenThermostatKernel::Name(), context);
     const System& system = context.getSystem();
     vector<double> masses(system.getNumAtoms());
     for (int i = 0; i < system.getNumAtoms(); ++i)

openmmapi/src/BrownianIntegrator.cpp

@@ -47,7 +47,7 @@ BrownianIntegrator::BrownianIntegrator(double temperature, double frictionCoeff,
 
 void BrownianIntegrator::initialize(OpenMMContextImpl& contextRef) {
     context = &contextRef;
-    kernel = context->getPlatform().createKernel(IntegrateBrownianStepKernel::Name());
+    kernel = context->getPlatform().createKernel(IntegrateBrownianStepKernel::Name(), contextRef);
     const System& system = context->getSystem();
     vector<double> masses(system.getNumAtoms());
     vector<std::vector<int> > constraintIndices(system.getNumConstraints());

openmmapi/src/CMMotionRemoverImpl.cpp

@@ -43,7 +43,7 @@ CMMotionRemoverImpl::CMMotionRemoverImpl(CMMotionRemover& owner) : owner(owner)
 }
 
 void CMMotionRemoverImpl::initialize(OpenMMContextImpl& context) {
-    kernel = context.getPlatform().createKernel(RemoveCMMotionKernel::Name());
+    kernel = context.getPlatform().createKernel(RemoveCMMotionKernel::Name(), context);
     const System& system = context.getSystem();
     vector<double> masses(system.getNumAtoms());
     for (int i = 0; i < system.getNumAtoms(); ++i)

openmmapi/src/GBSAOBCForceFieldImpl.cpp

@@ -41,7 +41,7 @@ GBSAOBCForceFieldImpl::GBSAOBCForceFieldImpl(GBSAOBCForceField& owner) : owner(o
 }
 
 void GBSAOBCForceFieldImpl::initialize(OpenMMContextImpl& context) {
-    kernel = context.getPlatform().createKernel(CalcGBSAOBCForceFieldKernel::Name());
+    kernel = context.getPlatform().createKernel(CalcGBSAOBCForceFieldKernel::Name(), context);
     vector<vector<double> > atomParameters(owner.getNumAtoms());
     for (int i = 0; i < owner.getNumAtoms(); ++i) {
         double charge, radius, scalingFactor;

openmmapi/src/LangevinIntegrator.cpp

@@ -47,7 +47,7 @@ LangevinIntegrator::LangevinIntegrator(double temperature, double frictionCoeff,
 
 void LangevinIntegrator::initialize(OpenMMContextImpl& contextRef) {
     context = &contextRef;
-    kernel = context->getPlatform().createKernel(IntegrateLangevinStepKernel::Name());
+    kernel = context->getPlatform().createKernel(IntegrateLangevinStepKernel::Name(), contextRef);
     const System& system = context->getSystem();
     vector<double> masses(system.getNumAtoms());
     vector<std::vector<int> > constraintIndices(system.getNumConstraints());

openmmapi/src/OpenMMContextImpl.cpp

@@ -45,7 +45,7 @@ using std::vector;
 using std::string;
 
 OpenMMContextImpl::OpenMMContextImpl(OpenMMContext& owner, System& system, Integrator& integrator, Platform* platform) :
-            owner(owner), system(system), integrator(integrator), platform(platform) {
+            owner(owner), system(system), integrator(integrator), platform(platform), platformData(NULL) {
     vector<string> kernelNames;
     kernelNames.push_back(CalcKineticEnergyKernel::Name());
     for (int i = 0; i < system.getNumForces(); ++i) {
@@ -61,12 +61,13 @@ OpenMMContextImpl::OpenMMContextImpl(OpenMMContext& owner, System& system, Integ
         this->platform = platform = &Platform::findPlatform(kernelNames);
     else if (!platform->supportsKernels(kernelNames))
         throw OpenMMException("Specified a Platform for an OpenMMContext which does not support all required kernels");
-    positions = platform->createStream("atomPositions", system.getNumAtoms(), Stream::Double3);
-    velocities = platform->createStream("atomVelocities", system.getNumAtoms(), Stream::Double3);
-    forces = platform->createStream("atomForces", system.getNumAtoms(), Stream::Double3);
+    platform->contextCreated(*this);
+    positions = platform->createStream("atomPositions", system.getNumAtoms(), Stream::Double3, *this);
+    velocities = platform->createStream("atomVelocities", system.getNumAtoms(), Stream::Double3, *this);
+    forces = platform->createStream("atomForces", system.getNumAtoms(), Stream::Double3, *this);
     double zero[] = {0.0, 0.0, 0.0};
     velocities.fillWithValue(&zero);
-    kineticEnergyKernel = platform->createKernel(CalcKineticEnergyKernel::Name());
+    kineticEnergyKernel = platform->createKernel(CalcKineticEnergyKernel::Name(), *this);
     vector<double> masses(system.getNumAtoms());
     for (int i = 0; i < masses.size(); ++i)
         masses[i] = system.getAtomMass(i);
@@ -79,6 +80,7 @@ OpenMMContextImpl::OpenMMContextImpl(OpenMMContext& owner, System& system, Integ
 OpenMMContextImpl::~OpenMMContextImpl() {
     for (int i = 0; i < (int) forceImpls.size(); ++i)
         delete forceImpls[i];
+    platform->contextDestroyed(*this);
 }
 
 double OpenMMContextImpl::getParameter(std::string name) {
@@ -126,3 +128,11 @@ void OpenMMContextImpl::reinitialize() {
     }
     integrator.initialize(*this);
 }
+
+void* OpenMMContextImpl::getPlatformData() {
+    return platformData;
+}
+
+void OpenMMContextImpl::setPlatformData(void* data) {
+    platformData = data;
+}

openmmapi/src/StandardMMForceFieldImpl.cpp

@@ -45,7 +45,7 @@ StandardMMForceFieldImpl::~StandardMMForceFieldImpl() {
 }
 
 void StandardMMForceFieldImpl::initialize(OpenMMContextImpl& context) {
-    kernel = context.getPlatform().createKernel(CalcStandardMMForceFieldKernel::Name());
+    kernel = context.getPlatform().createKernel(CalcStandardMMForceFieldKernel::Name(), context);
     vector<vector<int> > bondIndices(owner.getNumBonds());
     vector<vector<double> > bondParameters(owner.getNumBonds());
     vector<vector<int> > angleIndices(owner.getNumAngles());

openmmapi/src/VerletIntegrator.cpp

@@ -45,7 +45,7 @@ VerletIntegrator::VerletIntegrator(double stepSize) {
 
 void VerletIntegrator::initialize(OpenMMContextImpl& contextRef) {
     context = &contextRef;
-    kernel = context->getPlatform().createKernel(IntegrateVerletStepKernel::Name());
+    kernel = context->getPlatform().createKernel(IntegrateVerletStepKernel::Name(), contextRef);
     const System& system = context->getSystem();
     vector<double> masses(system.getNumAtoms());
     vector<std::vector<int> > constraintIndices(system.getNumConstraints());

platforms/brook/include/BrookKernelFactory.h

@@ -42,7 +42,7 @@ namespace OpenMM {
 
 class BrookKernelFactory : public KernelFactory {
 public:
-    KernelImpl* createKernelImpl(std::string name, const Platform& platform) const;
+    KernelImpl* createKernelImpl(std::string name, const Platform& platform, OpenMMContextImpl& context) const;
 };
 
 } // namespace OpenMM

platforms/brook/include/BrookStreamFactory.h

@@ -42,7 +42,7 @@ namespace OpenMM {
 
 class BrookStreamFactory : public StreamFactory {
 public:
-    StreamImpl* createStreamImpl(std::string name, int size, Stream::DataType type, int streamWidth, const Platform& platform) const;
+    StreamImpl* createStreamImpl(std::string name, int size, Stream::DataType type, int streamWidth, const Platform& platform, OpenMMContextImpl& context) const;
 };
 
 } // namespace OpenMM

platforms/brook/src/BrookKernelFactory.cpp

@@ -34,7 +34,7 @@
 
 using namespace OpenMM;
 
-KernelImpl* BrookKernelFactory::createKernelImpl(std::string name, const Platform& platform) const {
+KernelImpl* BrookKernelFactory::createKernelImpl(std::string name, const Platform& platform, OpenMMContextImpl& context) const {
     if (name == CalcStandardMMForceFieldKernel::Name())
         return new BrookCalcStandardMMForceFieldKernel(name, platform);
     if (name == CalcGBSAOBCForceFieldKernel::Name())

platforms/brook/src/BrookStreamFactory.cpp

@@ -36,7 +36,7 @@
 
 using namespace OpenMM;
 
-StreamImpl* BrookStreamFactory::createStreamImpl(std::string name, int size, Stream::DataType type, int streamWidth, const Platform& platform) const {
+StreamImpl* BrookStreamFactory::createStreamImpl(std::string name, int size, Stream::DataType type, int streamWidth, const Platform& platform, OpenMMContextImpl& context) const {
     switch (type) {
     case Stream::Float:
     case Stream::Float2:

platforms/cuda/include/CudaKernelFactory.h

+#ifndef OPENMM_CUDAKERNELFACTORY_H_
+#define OPENMM_CUDAKERNELFACTORY_H_
+
+/* -------------------------------------------------------------------------- *
+ *                                   OpenMM                                   *
+ * -------------------------------------------------------------------------- *
+ * This is part of the OpenMM molecular simulation toolkit originating from   *
+ * Simbios, the NIH National Center for Physics-Based Simulation of           *
+ * Biological Structures at Stanford, funded under the NIH Roadmap for        *
+ * Medical Research, grant U54 GM072970. See https://simtk.org.               *
+ *                                                                            *
+ * Portions copyright (c) 2008 Stanford University and the Authors.           *
+ * Authors: Peter Eastman                                                     *
+ * Contributors:                                                              *
+ *                                                                            *
+ * Permission is hereby granted, free of charge, to any person obtaining a    *
+ * copy of this software and associated documentation files (the "Software"), *
+ * to deal in the Software without restriction, including without limitation  *
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,   *
+ * and/or sell copies of the Software, and to permit persons to whom the      *
+ * Software is furnished to do so, subject to the following conditions:       *
+ *                                                                            *
+ * The above copyright notice and this permission notice shall be included in *
+ * all copies or substantial portions of the Software.                        *
+ *                                                                            *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR *
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,   *
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL    *
+ * THE AUTHORS, CONTRIBUTORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,    *
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR      *
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE  *
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.                                     *
+ * -------------------------------------------------------------------------- */
+
+#include "KernelFactory.h"
+
+namespace OpenMM {
+
+/**
+ * This KernelFactory creates all kernels for CudaPlatform.
+ */
+
+class CudaKernelFactory : public KernelFactory {
+public:
+    KernelImpl* createKernelImpl(std::string name, const Platform& platform, OpenMMContextImpl& context) const;
+};
+
+} // namespace OpenMM
+
+#endif /*OPENMM_CUDAKERNELFACTORY_H_*/

platforms/cuda/include/CudaPlatform.h

+#ifndef OPENMM_CUDAPLATFORM_H_
+#define OPENMM_CUDAPLATFORM_H_
+
+/* -------------------------------------------------------------------------- *
+ *                                   OpenMM                                   *
+ * -------------------------------------------------------------------------- *
+ * This is part of the OpenMM molecular simulation toolkit originating from   *
+ * Simbios, the NIH National Center for Physics-Based Simulation of           *
+ * Biological Structures at Stanford, funded under the NIH Roadmap for        *
+ * Medical Research, grant U54 GM072970. See https://simtk.org.               *
+ *                                                                            *
+ * Portions copyright (c) 2008 Stanford University and the Authors.           *
+ * Authors: Peter Eastman                                                     *
+ * Contributors:                                                              *
+ *                                                                            *
+ * Permission is hereby granted, free of charge, to any person obtaining a    *
+ * copy of this software and associated documentation files (the "Software"), *
+ * to deal in the Software without restriction, including without limitation  *
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,   *
+ * and/or sell copies of the Software, and to permit persons to whom the      *
+ * Software is furnished to do so, subject to the following conditions:       *
+ *                                                                            *
+ * The above copyright notice and this permission notice shall be included in *
+ * all copies or substantial portions of the Software.                        *
+ *                                                                            *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR *
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,   *
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL    *
+ * THE AUTHORS, CONTRIBUTORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,    *
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR      *
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE  *
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.                                     *
+ * -------------------------------------------------------------------------- */
+
+#include "Platform.h"
+#include "CudaStreamFactory.h"
+
+namespace OpenMM {
+
+/**
+ * This Platform subclass uses CUDA implementations of the OpenMM kernels to run on NVidia GPUs.
+ */
+
+class CudaPlatform : public Platform {
+public:
+    CudaPlatform();
+    std::string getName() const {
+        return "Cuda";
+    }
+    double getSpeed() const {
+        return 100;
+    }
+    bool supportsDoublePrecision() const;
+    const StreamFactory& getDefaultStreamFactory() const;
+    void contextCreated(OpenMMContextImpl& context) const;
+    void contextDestroyed(OpenMMContextImpl& context) const;
+private:
+    CudaStreamFactory defaultStreamFactory;
+};
+
+} // namespace OpenMM
+
+#endif /*OPENMM_CUDAPLATFORM_H_*/

platforms/cuda/include/CudaStreamFactory.h

+#ifndef OPENMM_CUDASTREAMFACTORY_H_
+#define OPENMM_CUDASTREAMFACTORY_H_
+
+/* -------------------------------------------------------------------------- *
+ *                                   OpenMM                                   *
+ * -------------------------------------------------------------------------- *
+ * This is part of the OpenMM molecular simulation toolkit originating from   *
+ * Simbios, the NIH National Center for Physics-Based Simulation of           *
+ * Biological Structures at Stanford, funded under the NIH Roadmap for        *
+ * Medical Research, grant U54 GM072970. See https://simtk.org.               *
+ *                                                                            *
+ * Portions copyright (c) 2008 Stanford University and the Authors.           *
+ * Authors: Peter Eastman                                                     *
+ * Contributors:                                                              *
+ *                                                                            *
+ * Permission is hereby granted, free of charge, to any person obtaining a    *
+ * copy of this software and associated documentation files (the "Software"), *
+ * to deal in the Software without restriction, including without limitation  *
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,   *
+ * and/or sell copies of the Software, and to permit persons to whom the      *
+ * Software is furnished to do so, subject to the following conditions:       *
+ *                                                                            *
+ * The above copyright notice and this permission notice shall be included in *
+ * all copies or substantial portions of the Software.                        *
+ *                                                                            *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR *
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,   *
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL    *
+ * THE AUTHORS, CONTRIBUTORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,    *
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR      *
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE  *
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.                                     *
+ * -------------------------------------------------------------------------- */
+
+#include "StreamFactory.h"
+
+namespace OpenMM {
+
+/**
+ * This StreamFactory creates all streams for CudaPlatform.
+ */
+
+class CudaStreamFactory : public StreamFactory {
+public:
+    StreamImpl* createStreamImpl(std::string name, int size, Stream::DataType type, const Platform& platform, OpenMMContextImpl& context) const;
+};
+
+} // namespace OpenMM
+
+#endif /*OPENMM_CUDASTREAMFACTORY_H_*/