Converted AMOEBA to common platform (#3120)

* Began converting AMOEBA to common platform

* Beginning of OpenCL platform for AMOEBA

* Converted AmoebaVdwForce to common platform

* Cleaned up reference AMOEBA tests

* Began converting AmoebaMultipoleForce to common platform

* Continue converting AmoebaMultipoleForce to common platform

* Bug fixes

* Bug fix

* Continue converting AmoebaMultipoleForce to common platform

* Converting AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce to common platform

* Converting AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce to common platform

* Creating OpenCL version of AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce

* Creating OpenCL version of AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce

* Creating OpenCL version of AmoebaMultipoleForce and AmoebaGeneralizedKirkwoodForce

* Converted arrays from real3 to real

* Bug fix to OpenCL AmoebaGeneralizedKirkwoodForce

* Fixes for AMD GPUs

* Began converting HippoNonbondedForce to common platform

* Continuing to convert HippoNonbondedForce to common platform

* Continuing to convert HippoNonbondedForce to common platform

* Working on unifying PME kernels

* Fixed error on devices without 64 bit atomics

* Unified PME kernels

* Converted HippoNonbondedForce to common platform

* Creating OpenCL implementation of HippoNonbondedForce

* Continuing OpenCL implementation of HippoNonbondedForce

* Mostly finished OpenCL implementation of HippoNonbondedForce

* Eliminated three component vector types in host code

* Fix errors on CPU OpenCL

* Skip double precision tests for AMOEBA on OpenCL

* Bug fixes

* Bug fixes

* Fixed compilation error

plugins/amoeba/platforms/cuda/src/kernels/hippoFixedField.cu → plugins/amoeba/platforms/common/src/kernels/hippoFixedField.cc

@@ -32,7 +32,7 @@ real bn3 = (5*bn2+alsq2n*exp2a)*invR2;
 // Calculate the field at particle 1 due to multipoles at particle 2
 
 real fdamp3, fdamp5, fdamp7;
-computeDirectFieldDampingFactors(alpha2, r, fdamp3, fdamp5, fdamp7);
+computeDirectFieldDampingFactors(alpha2, r, &fdamp3, &fdamp5, &fdamp7);
 #ifndef COMPUTING_EXCEPTIONS
 real scale = 1;
 #endif
@@ -58,7 +58,7 @@ tempField1 = -delta*factor2 - dipole2*rr3j + qDotDelta2*2*rr5j;
 
 // Calculate the field at particle 2 due to multipoles at particle 1
 
-computeDirectFieldDampingFactors(alpha1, r, fdamp3, fdamp5, fdamp7);
+computeDirectFieldDampingFactors(alpha1, r, &fdamp3, &fdamp5, &fdamp7);
 #ifdef USE_EWALD
 real rr3i = bn1 - (1-scale*fdamp3)*invR3;
 real rr5i = bn2 - (1-scale*fdamp5)*3*invR5;

plugins/amoeba/platforms/cuda/src/kernels/hippoMutualField.cu → plugins/amoeba/platforms/common/src/kernels/hippoMutualField.cc

 real fdamp3, fdamp5;
-computeMutualFieldDampingFactors(alpha1, alpha2, r, fdamp3, fdamp5);
+computeMutualFieldDampingFactors(alpha1, alpha2, r, &fdamp3, &fdamp5);
 #ifdef COMPUTING_EXCEPTIONS
 fdamp3 *= scale;
 fdamp5 *= scale;

plugins/amoeba/platforms/cuda/src/kernels/sphericalMultipoles.cu → plugins/amoeba/platforms/common/src/kernels/sphericalMultipoles.cc

-__device__ void buildQIRotationMatrix(real3 deltaR, real rInv, real (&rotationMatrix)[3][3]) {
+DEVICE void buildQIRotationMatrix(real3 deltaR, real rInv, real rotationMatrix[][3]) {
     real3 vectorZ = deltaR*rInv;
     real3 vectorX = vectorZ;
     if (deltaR.y != 0 || deltaR.z != 0)
@@ -23,14 +23,13 @@ __device__ void buildQIRotationMatrix(real3 deltaR, real rInv, real (&rotationMa
     rotationMatrix[2][2] = vectorY.y;
 }
 
-__device__ real3 rotateDipole(real3& dipole, const real (&rotationMatrix)[3][3]) {
+DEVICE real3 rotateDipole(real3 dipole, const real rotationMatrix[][3]) {
     return make_real3(rotationMatrix[0][0]*dipole.x + rotationMatrix[0][1]*dipole.y + rotationMatrix[0][2]*dipole.z,
                       rotationMatrix[1][0]*dipole.x + rotationMatrix[1][1]*dipole.y + rotationMatrix[1][2]*dipole.z,
                       rotationMatrix[2][0]*dipole.x + rotationMatrix[2][1]*dipole.y + rotationMatrix[2][2]*dipole.z);
 }
 
-
-__device__ void rotateQuadupoles(const real (&rotationMatrix)[3][3], const real* quad1, const real* quad2, real* rotated1, real* rotated2) {
+DEVICE void rotateQuadrupoles(const real rotationMatrix[][3], const real* quad1, LOCAL_ARG const real* quad2, real* rotated1, real* rotated2) {
     real sqrtThree = SQRT((real) 3);
     real element;
     element = 0.5f*(3.0f*rotationMatrix[0][0]*rotationMatrix[0][0] - 1.0f);

plugins/amoeba/platforms/cuda/CMakeLists.txt

@@ -12,7 +12,7 @@
 
 # The source is organized into subdirectories, but we handle them all from
 # this CMakeLists file rather than letting CMake visit them as SUBDIRS.
-SET(OPENMM_SOURCE_SUBDIRS .)
+SET(OPENMM_SOURCE_SUBDIRS . ../common)
 
 
 # Collect up information about the version of the OpenMM library we're building
@@ -61,11 +61,17 @@ FOREACH(subdir ${OPENMM_SOURCE_SUBDIRS})
     INCLUDE_DIRECTORIES(BEFORE ${CMAKE_CURRENT_SOURCE_DIR}/${subdir}/include)
 ENDFOREACH(subdir)
 
+SET(COMMON_KERNELS_CPP ${CMAKE_CURRENT_BINARY_DIR}/../common/src/CommonAmoebaKernelSources.cpp)
+SET(SOURCE_FILES ${SOURCE_FILES} ${COMMON_KERNELS_CPP})
+
 INCLUDE_DIRECTORIES(BEFORE ${CMAKE_CURRENT_SOURCE_DIR}/src)
+INCLUDE_DIRECTORIES(BEFORE ${CMAKE_CURRENT_SOURCE_DIR}/../common/src)
 INCLUDE_DIRECTORIES(BEFORE ${CMAKE_SOURCE_DIR}/platforms/cuda/include)
 INCLUDE_DIRECTORIES(BEFORE ${CMAKE_SOURCE_DIR}/platforms/cuda/src)
 INCLUDE_DIRECTORIES(BEFORE ${CMAKE_BINARY_DIR}/platforms/cuda/src)
 INCLUDE_DIRECTORIES(BEFORE ${CMAKE_SOURCE_DIR}/platforms/common/include)
+INCLUDE_DIRECTORIES(BEFORE ${CMAKE_BINARY_DIR}/platforms/common/src)
+INCLUDE_DIRECTORIES(BEFORE ${CMAKE_CURRENT_BINARY_DIR}/../common/src)
 
 # Set variables needed for encoding kernel sources into a C++ class
 
@@ -80,6 +86,8 @@ INCLUDE_DIRECTORIES(BEFORE ${CMAKE_CURRENT_BINARY_DIR}/src)
 
 INCLUDE_DIRECTORIES(${CUDA_TOOLKIT_INCLUDE})
 
+SET_SOURCE_FILES_PROPERTIES(${COMMON_KERNELS_CPP} PROPERTIES GENERATED TRUE)
+
 FILE(GLOB CUDA_KERNELS ${KERNEL_SOURCE_DIR}/kernels/*.cu)
 ADD_CUSTOM_COMMAND(OUTPUT ${KERNELS_CPP} ${KERNELS_H}
     COMMAND ${CMAKE_COMMAND}
@@ -92,6 +100,7 @@ SET_SOURCE_FILES_PROPERTIES(${KERNELS_CPP} ${KERNELS_H} PROPERTIES GENERATED TRU
 
 IF (OPENMM_BUILD_SHARED_LIB)
     ADD_LIBRARY(${SHARED_TARGET} SHARED ${SOURCE_FILES} ${SOURCE_INCLUDE_FILES} ${API_ABS_INCLUDE_FILES})
+    ADD_DEPENDENCIES(${SHARED_TARGET} AmoebaCommonKernels)
 
     TARGET_LINK_LIBRARIES(${SHARED_TARGET} ${OPENMM_LIBRARY_NAME} ${PTHREADS_LIB})
     TARGET_LINK_LIBRARIES(${SHARED_TARGET} ${OPENMM_LIBRARY_NAME}CUDA)