Commit cb75edd4 authored by Peter Eastman's avatar Peter Eastman
Browse files

Added pthreads library for Windows. Fixed other errors and warnings under Windows.

parent dd352ee5
This diff is collapsed.
This diff is collapsed.
/*
* Module: sched.h
*
* Purpose:
* Provides an implementation of POSIX realtime extensions
* as defined in
*
* POSIX 1003.1b-1993 (POSIX.1b)
*
* --------------------------------------------------------------------------
*
* Pthreads-win32 - POSIX Threads Library for Win32
* Copyright(C) 1998 John E. Bossom
* Copyright(C) 1999,2005 Pthreads-win32 contributors
*
* Contact Email: rpj@callisto.canberra.edu.au
*
* The current list of contributors is contained
* in the file CONTRIBUTORS included with the source
* code distribution. The list can also be seen at the
* following World Wide Web location:
* http://sources.redhat.com/pthreads-win32/contributors.html
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library in the file COPYING.LIB;
* if not, write to the Free Software Foundation, Inc.,
* 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA
*/
#ifndef _SCHED_H
#define _SCHED_H
#undef PTW32_LEVEL
#if defined(_POSIX_SOURCE)
#define PTW32_LEVEL 0
/* Early POSIX */
#endif
#if defined(_POSIX_C_SOURCE) && _POSIX_C_SOURCE >= 199309
#undef PTW32_LEVEL
#define PTW32_LEVEL 1
/* Include 1b, 1c and 1d */
#endif
#if defined(INCLUDE_NP)
#undef PTW32_LEVEL
#define PTW32_LEVEL 2
/* Include Non-Portable extensions */
#endif
#define PTW32_LEVEL_MAX 3
#if !defined(PTW32_LEVEL)
#define PTW32_LEVEL PTW32_LEVEL_MAX
/* Include everything */
#endif
#if __GNUC__ && ! defined (__declspec)
# error Please upgrade your GNU compiler to one that supports __declspec.
#endif
/*
* When building the DLL code, you should define PTW32_BUILD so that
* the variables/functions are exported correctly. When using the DLL,
* do NOT define PTW32_BUILD, and then the variables/functions will
* be imported correctly.
*/
#ifndef PTW32_STATIC_LIB
# ifdef PTW32_BUILD
# define PTW32_DLLPORT __declspec (dllexport)
# else
# define PTW32_DLLPORT __declspec (dllimport)
# endif
#else
# define PTW32_DLLPORT
#endif
/*
* This is a duplicate of what is in the autoconf config.h,
* which is only used when building the pthread-win32 libraries.
*/
#ifndef PTW32_CONFIG_H
# if defined(WINCE)
# define NEED_ERRNO
# define NEED_SEM
# endif
# if defined(_UWIN) || defined(__MINGW32__)
# define HAVE_MODE_T
# endif
#endif
/*
*
*/
#if PTW32_LEVEL >= PTW32_LEVEL_MAX
#ifdef NEED_ERRNO
#include "need_errno.h"
#else
#include <errno.h>
#endif
#endif /* PTW32_LEVEL >= PTW32_LEVEL_MAX */
#if defined(__MINGW32__) || defined(_UWIN)
#if PTW32_LEVEL >= PTW32_LEVEL_MAX
/* For pid_t */
# include <sys/types.h>
/* Required by Unix 98 */
# include <time.h>
#endif /* PTW32_LEVEL >= PTW32_LEVEL_MAX */
#else
typedef int pid_t;
#endif
/* Thread scheduling policies */
enum {
SCHED_OTHER = 0,
SCHED_FIFO,
SCHED_RR,
SCHED_MIN = SCHED_OTHER,
SCHED_MAX = SCHED_RR
};
struct sched_param {
int sched_priority;
};
#ifdef __cplusplus
extern "C"
{
#endif /* __cplusplus */
PTW32_DLLPORT int __cdecl sched_yield (void);
PTW32_DLLPORT int __cdecl sched_get_priority_min (int policy);
PTW32_DLLPORT int __cdecl sched_get_priority_max (int policy);
PTW32_DLLPORT int __cdecl sched_setscheduler (pid_t pid, int policy);
PTW32_DLLPORT int __cdecl sched_getscheduler (pid_t pid);
/*
* Note that this macro returns ENOTSUP rather than
* ENOSYS as might be expected. However, returning ENOSYS
* should mean that sched_get_priority_{min,max} are
* not implemented as well as sched_rr_get_interval.
* This is not the case, since we just don't support
* round-robin scheduling. Therefore I have chosen to
* return the same value as sched_setscheduler when
* SCHED_RR is passed to it.
*/
#define sched_rr_get_interval(_pid, _interval) \
( errno = ENOTSUP, (int) -1 )
#ifdef __cplusplus
} /* End of extern "C" */
#endif /* __cplusplus */
#undef PTW32_LEVEL
#undef PTW32_LEVEL_MAX
#endif /* !_SCHED_H */
/*
* Module: semaphore.h
*
* Purpose:
* Semaphores aren't actually part of the PThreads standard.
* They are defined by the POSIX Standard:
*
* POSIX 1003.1b-1993 (POSIX.1b)
*
* --------------------------------------------------------------------------
*
* Pthreads-win32 - POSIX Threads Library for Win32
* Copyright(C) 1998 John E. Bossom
* Copyright(C) 1999,2005 Pthreads-win32 contributors
*
* Contact Email: rpj@callisto.canberra.edu.au
*
* The current list of contributors is contained
* in the file CONTRIBUTORS included with the source
* code distribution. The list can also be seen at the
* following World Wide Web location:
* http://sources.redhat.com/pthreads-win32/contributors.html
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library in the file COPYING.LIB;
* if not, write to the Free Software Foundation, Inc.,
* 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA
*/
#if !defined( SEMAPHORE_H )
#define SEMAPHORE_H
#undef PTW32_LEVEL
#if defined(_POSIX_SOURCE)
#define PTW32_LEVEL 0
/* Early POSIX */
#endif
#if defined(_POSIX_C_SOURCE) && _POSIX_C_SOURCE >= 199309
#undef PTW32_LEVEL
#define PTW32_LEVEL 1
/* Include 1b, 1c and 1d */
#endif
#if defined(INCLUDE_NP)
#undef PTW32_LEVEL
#define PTW32_LEVEL 2
/* Include Non-Portable extensions */
#endif
#define PTW32_LEVEL_MAX 3
#if !defined(PTW32_LEVEL)
#define PTW32_LEVEL PTW32_LEVEL_MAX
/* Include everything */
#endif
#if __GNUC__ && ! defined (__declspec)
# error Please upgrade your GNU compiler to one that supports __declspec.
#endif
/*
* When building the DLL code, you should define PTW32_BUILD so that
* the variables/functions are exported correctly. When using the DLL,
* do NOT define PTW32_BUILD, and then the variables/functions will
* be imported correctly.
*/
#ifndef PTW32_STATIC_LIB
# ifdef PTW32_BUILD
# define PTW32_DLLPORT __declspec (dllexport)
# else
# define PTW32_DLLPORT __declspec (dllimport)
# endif
#else
# define PTW32_DLLPORT
#endif
/*
* This is a duplicate of what is in the autoconf config.h,
* which is only used when building the pthread-win32 libraries.
*/
#ifndef PTW32_CONFIG_H
# if defined(WINCE)
# define NEED_ERRNO
# define NEED_SEM
# endif
# if defined(_UWIN) || defined(__MINGW32__)
# define HAVE_MODE_T
# endif
#endif
/*
*
*/
#if PTW32_LEVEL >= PTW32_LEVEL_MAX
#ifdef NEED_ERRNO
#include "need_errno.h"
#else
#include <errno.h>
#endif
#endif /* PTW32_LEVEL >= PTW32_LEVEL_MAX */
#define _POSIX_SEMAPHORES
#ifdef __cplusplus
extern "C"
{
#endif /* __cplusplus */
#ifndef HAVE_MODE_T
typedef unsigned int mode_t;
#endif
typedef struct sem_t_ * sem_t;
PTW32_DLLPORT int __cdecl sem_init (sem_t * sem,
int pshared,
unsigned int value);
PTW32_DLLPORT int __cdecl sem_destroy (sem_t * sem);
PTW32_DLLPORT int __cdecl sem_trywait (sem_t * sem);
PTW32_DLLPORT int __cdecl sem_wait (sem_t * sem);
PTW32_DLLPORT int __cdecl sem_timedwait (sem_t * sem,
const struct timespec * abstime);
PTW32_DLLPORT int __cdecl sem_post (sem_t * sem);
PTW32_DLLPORT int __cdecl sem_post_multiple (sem_t * sem,
int count);
PTW32_DLLPORT int __cdecl sem_open (const char * name,
int oflag,
mode_t mode,
unsigned int value);
PTW32_DLLPORT int __cdecl sem_close (sem_t * sem);
PTW32_DLLPORT int __cdecl sem_unlink (const char * name);
PTW32_DLLPORT int __cdecl sem_getvalue (sem_t * sem,
int * sval);
#ifdef __cplusplus
} /* End of extern "C" */
#endif /* __cplusplus */
#undef PTW32_LEVEL
#undef PTW32_LEVEL_MAX
#endif /* !SEMAPHORE_H */
# #
# Include OpenCL related files. # Include OpenCL related files.
# #
# INCLUDE(${CMAKE_CURRENT_SOURCE_DIR}/../FindOpenCL.cmake) # INCLUDE(${CMAKE_CURRENT_SOURCE_DIR}/../FindOpenCL.cmake)
INCLUDE_DIRECTORIES(${OPENCL_INCLUDE_DIR}) INCLUDE_DIRECTORIES(${OPENCL_INCLUDE_DIR})
FILE(GLOB OPENCL_KERNELS ${CL_SOURCE_DIR}/kernels/*.cl) FILE(GLOB OPENCL_KERNELS ${CL_SOURCE_DIR}/kernels/*.cl)
ADD_CUSTOM_COMMAND(OUTPUT ${CL_KERNELS_CPP} ${CL_KERNELS_H} ADD_CUSTOM_COMMAND(OUTPUT ${CL_KERNELS_CPP} ${CL_KERNELS_H}
COMMAND ${CMAKE_COMMAND} COMMAND ${CMAKE_COMMAND}
ARGS -D CL_SOURCE_DIR=${CL_SOURCE_DIR} -D CL_KERNELS_CPP=${CL_KERNELS_CPP} -D CL_KERNELS_H=${CL_KERNELS_H} -P ${CMAKE_CURRENT_SOURCE_DIR}/../EncodeCLFiles.cmake ARGS -D CL_SOURCE_DIR=${CL_SOURCE_DIR} -D CL_KERNELS_CPP=${CL_KERNELS_CPP} -D CL_KERNELS_H=${CL_KERNELS_H} -P ${CMAKE_CURRENT_SOURCE_DIR}/../EncodeCLFiles.cmake
DEPENDS ${OPENCL_KERNELS} DEPENDS ${OPENCL_KERNELS}
) )
SET_SOURCE_FILES_PROPERTIES(${CL_KERNELS_CPP} ${CL_KERNELS_H} PROPERTIES GENERATED TRUE) SET_SOURCE_FILES_PROPERTIES(${CL_KERNELS_CPP} ${CL_KERNELS_H} PROPERTIES GENERATED TRUE)
ADD_LIBRARY(${SHARED_TARGET} SHARED ${SOURCE_FILES} ${SOURCE_INCLUDE_FILES} ${API_ABS_INCLUDE_FILES}) ADD_LIBRARY(${SHARED_TARGET} SHARED ${SOURCE_FILES} ${SOURCE_INCLUDE_FILES} ${API_ABS_INCLUDE_FILES})
IF (UNIX AND CMAKE_BUILD_TYPE MATCHES Debug) IF (UNIX AND CMAKE_BUILD_TYPE MATCHES Debug)
SET(MAIN_OPENMM_LIB ${OPENMM_LIBRARY_NAME}_d) SET(MAIN_OPENMM_LIB ${OPENMM_LIBRARY_NAME}_d)
ELSE (UNIX AND CMAKE_BUILD_TYPE MATCHES Debug) ELSE (UNIX AND CMAKE_BUILD_TYPE MATCHES Debug)
SET(MAIN_OPENMM_LIB ${OPENMM_LIBRARY_NAME}) SET(MAIN_OPENMM_LIB ${OPENMM_LIBRARY_NAME})
ENDIF (UNIX AND CMAKE_BUILD_TYPE MATCHES Debug) ENDIF (UNIX AND CMAKE_BUILD_TYPE MATCHES Debug)
TARGET_LINK_LIBRARIES(${SHARED_TARGET} ${MAIN_OPENMM_LIB} ${OPENCL_LIBRARIES}) TARGET_LINK_LIBRARIES(${SHARED_TARGET} ${MAIN_OPENMM_LIB} ${OPENCL_LIBRARIES} ${PTHREADS_LIB})
SET_TARGET_PROPERTIES(${SHARED_TARGET} PROPERTIES COMPILE_FLAGS "-DOPENMM_BUILDING_SHARED_LIBRARY") SET_TARGET_PROPERTIES(${SHARED_TARGET} PROPERTIES COMPILE_FLAGS "-DOPENMM_BUILDING_SHARED_LIBRARY")
INSTALL_TARGETS(/lib/plugins RUNTIME_DIRECTORY /lib/plugins ${SHARED_TARGET}) INSTALL_TARGETS(/lib/plugins RUNTIME_DIRECTORY /lib/plugins ${SHARED_TARGET})
...@@ -126,10 +126,10 @@ OpenCLContext::OpenCLContext(int numParticles, int deviceIndex, OpenCLPlatform:: ...@@ -126,10 +126,10 @@ OpenCLContext::OpenCLContext(int numParticles, int deviceIndex, OpenCLPlatform::
cl::Kernel accuracyKernel(utilities, "determineNativeAccuracy"); cl::Kernel accuracyKernel(utilities, "determineNativeAccuracy");
OpenCLArray<mm_float8> values(*this, 20, "values", true); OpenCLArray<mm_float8> values(*this, 20, "values", true);
float nextValue = 1e-4; float nextValue = 1e-4f;
for (int i = 0; i < values.getSize(); ++i) { for (int i = 0; i < values.getSize(); ++i) {
values[i].s0 = nextValue; values[i].s0 = nextValue;
nextValue *= M_PI; nextValue *= (float) M_PI;
} }
values.upload(); values.upload();
accuracyKernel.setArg<cl::Buffer>(0, values.getDeviceBuffer()); accuracyKernel.setArg<cl::Buffer>(0, values.getDeviceBuffer());
......
...@@ -466,7 +466,7 @@ OpenCLIntegrationUtilities::OpenCLIntegrationUtilities(OpenCLContext& context, c ...@@ -466,7 +466,7 @@ OpenCLIntegrationUtilities::OpenCLIntegrationUtilities(OpenCLContext& context, c
int c = ccmaConstraints[index]; int c = ccmaConstraints[index];
atomsVec[i].x = atom1[c]; atomsVec[i].x = atom1[c];
atomsVec[i].y = atom2[c]; atomsVec[i].y = atom2[c];
distanceVec[i].w = distance[c]; distanceVec[i].w = (float) distance[c];
reducedMassVec[i] = (float) (0.5/(1.0/system.getParticleMass(atom1[c])+1.0/system.getParticleMass(atom2[c]))); reducedMassVec[i] = (float) (0.5/(1.0/system.getParticleMass(atom1[c])+1.0/system.getParticleMass(atom2[c])));
for (unsigned int j = 0; j < matrix[index].size(); j++) { for (unsigned int j = 0; j < matrix[index].size(); j++) {
constraintMatrixColumnVec[i+j*numCCMA] = matrix[index][j].first; constraintMatrixColumnVec[i+j*numCCMA] = matrix[index][j].first;
......
...@@ -888,10 +888,10 @@ void OpenCLCalcCMAPTorsionForceKernel::initialize(const System& system, const CM ...@@ -888,10 +888,10 @@ void OpenCLCalcCMAPTorsionForceKernel::initialize(const System& system, const CM
mapPositionsVec[i] = mm_int2(currentPosition, size); mapPositionsVec[i] = mm_int2(currentPosition, size);
currentPosition += 4*size*size; currentPosition += 4*size*size;
for (int j = 0; j < size*size; j++) { for (int j = 0; j < size*size; j++) {
coeffVec.push_back(mm_float4(c[j][0], c[j][1], c[j][2], c[j][3])); coeffVec.push_back(mm_float4((float) c[j][0], (float) c[j][1], (float) c[j][2], (float) c[j][3]));
coeffVec.push_back(mm_float4(c[j][4], c[j][5], c[j][6], c[j][7])); coeffVec.push_back(mm_float4((float) c[j][4], (float) c[j][5], (float) c[j][6], (float) c[j][7]));
coeffVec.push_back(mm_float4(c[j][8], c[j][9], c[j][10], c[j][11])); coeffVec.push_back(mm_float4((float) c[j][8], (float) c[j][9], (float) c[j][10], (float) c[j][11]));
coeffVec.push_back(mm_float4(c[j][12], c[j][13], c[j][14], c[j][15])); coeffVec.push_back(mm_float4((float) c[j][12], (float) c[j][13], (float) c[j][14], (float) c[j][15]));
} }
} }
vector<int> forceBufferCounter(system.getNumParticles(), 0); vector<int> forceBufferCounter(system.getNumParticles(), 0);
...@@ -1452,7 +1452,7 @@ double OpenCLCalcNonbondedForceKernel::execute(ContextImpl& context, bool includ ...@@ -1452,7 +1452,7 @@ double OpenCLCalcNonbondedForceKernel::execute(ContextImpl& context, bool includ
if (cosSinSums != NULL && cl.getContextIndex() == 0) { if (cosSinSums != NULL && cl.getContextIndex() == 0) {
mm_float4 boxSize = cl.getPeriodicBoxSize(); mm_float4 boxSize = cl.getPeriodicBoxSize();
mm_float4 recipBoxSize = mm_float4((float) (2*M_PI/boxSize.x), (float) (2*M_PI/boxSize.y), (float) (2*M_PI/boxSize.z), 0); mm_float4 recipBoxSize = mm_float4((float) (2*M_PI/boxSize.x), (float) (2*M_PI/boxSize.y), (float) (2*M_PI/boxSize.z), 0);
float recipCoefficient = ONE_4PI_EPS0*4*M_PI/(boxSize.x*boxSize.y*boxSize.z); float recipCoefficient = (float) (ONE_4PI_EPS0*4*M_PI/(boxSize.x*boxSize.y*boxSize.z));
ewaldSumsKernel.setArg<mm_float4>(3, recipBoxSize); ewaldSumsKernel.setArg<mm_float4>(3, recipBoxSize);
ewaldSumsKernel.setArg<cl_float>(4, recipCoefficient); ewaldSumsKernel.setArg<cl_float>(4, recipCoefficient);
cl.executeKernel(ewaldSumsKernel, cosSinSums->getSize()); cl.executeKernel(ewaldSumsKernel, cosSinSums->getSize());
...@@ -1581,7 +1581,7 @@ void OpenCLCalcCustomNonbondedForceKernel::initialize(const System& system, cons ...@@ -1581,7 +1581,7 @@ void OpenCLCalcCustomNonbondedForceKernel::initialize(const System& system, cons
string arrayName = prefix+"table"+intToString(i); string arrayName = prefix+"table"+intToString(i);
functionDefinitions.push_back(make_pair(name, arrayName)); functionDefinitions.push_back(make_pair(name, arrayName));
functions[name] = &fp; functions[name] = &fp;
tabulatedFunctionParamsVec[i] = mm_float4((float) min, (float) max, (float) ((values.size()-1)/(max-min)), values.size()-2); tabulatedFunctionParamsVec[i] = mm_float4((float) min, (float) max, (float) ((values.size()-1)/(max-min)), (float) values.size()-2);
vector<mm_float4> f = OpenCLExpressionUtilities::computeFunctionCoefficients(values, min, max); vector<mm_float4> f = OpenCLExpressionUtilities::computeFunctionCoefficients(values, min, max);
tabulatedFunctions.push_back(new OpenCLArray<mm_float4>(cl, values.size()-1, "TabulatedFunction")); tabulatedFunctions.push_back(new OpenCLArray<mm_float4>(cl, values.size()-1, "TabulatedFunction"));
tabulatedFunctions[tabulatedFunctions.size()-1]->upload(f); tabulatedFunctions[tabulatedFunctions.size()-1]->upload(f);
...@@ -1932,7 +1932,7 @@ void OpenCLCalcCustomGBForceKernel::initialize(const System& system, const Custo ...@@ -1932,7 +1932,7 @@ void OpenCLCalcCustomGBForceKernel::initialize(const System& system, const Custo
string arrayName = prefix+"table"+intToString(i); string arrayName = prefix+"table"+intToString(i);
functionDefinitions.push_back(make_pair(name, arrayName)); functionDefinitions.push_back(make_pair(name, arrayName));
functions[name] = &fp; functions[name] = &fp;
tabulatedFunctionParamsVec[i] = mm_float4((float) min, (float) max, (float) ((values.size()-1)/(max-min)), values.size()-2); tabulatedFunctionParamsVec[i] = mm_float4((float) min, (float) max, (float) ((values.size()-1)/(max-min)), (float) values.size()-2);
vector<mm_float4> f = OpenCLExpressionUtilities::computeFunctionCoefficients(values, min, max); vector<mm_float4> f = OpenCLExpressionUtilities::computeFunctionCoefficients(values, min, max);
tabulatedFunctions.push_back(new OpenCLArray<mm_float4>(cl, values.size()-1, "TabulatedFunction")); tabulatedFunctions.push_back(new OpenCLArray<mm_float4>(cl, values.size()-1, "TabulatedFunction"));
tabulatedFunctions[tabulatedFunctions.size()-1]->upload(f); tabulatedFunctions[tabulatedFunctions.size()-1]->upload(f);
...@@ -3003,7 +3003,7 @@ void OpenCLCalcCustomHbondForceKernel::initialize(const System& system, const Cu ...@@ -3003,7 +3003,7 @@ void OpenCLCalcCustomHbondForceKernel::initialize(const System& system, const Cu
string arrayName = "table"+intToString(i); string arrayName = "table"+intToString(i);
functionDefinitions.push_back(make_pair(name, arrayName)); functionDefinitions.push_back(make_pair(name, arrayName));
functions[name] = &fp; functions[name] = &fp;
tabulatedFunctionParamsVec[i] = mm_float4((float) min, (float) max, (float) ((values.size()-1)/(max-min)), values.size()-2); tabulatedFunctionParamsVec[i] = mm_float4((float) min, (float) max, (float) ((values.size()-1)/(max-min)), (float) values.size()-2);
vector<mm_float4> f = OpenCLExpressionUtilities::computeFunctionCoefficients(values, min, max); vector<mm_float4> f = OpenCLExpressionUtilities::computeFunctionCoefficients(values, min, max);
tabulatedFunctions.push_back(new OpenCLArray<mm_float4>(cl, values.size()-1, "TabulatedFunction")); tabulatedFunctions.push_back(new OpenCLArray<mm_float4>(cl, values.size()-1, "TabulatedFunction"));
tabulatedFunctions[tabulatedFunctions.size()-1]->upload(f); tabulatedFunctions[tabulatedFunctions.size()-1]->upload(f);
...@@ -3387,7 +3387,7 @@ void OpenCLIntegrateLangevinStepKernel::execute(ContextImpl& context, const Lang ...@@ -3387,7 +3387,7 @@ void OpenCLIntegrateLangevinStepKernel::execute(ContextImpl& context, const Lang
p[1] = (cl_float) fscale; p[1] = (cl_float) fscale;
p[2] = (cl_float) noisescale; p[2] = (cl_float) noisescale;
params->upload(p); params->upload(p);
integration.getStepSize()[0].y = stepSize; integration.getStepSize()[0].y = (cl_float) stepSize;
integration.getStepSize().upload(); integration.getStepSize().upload();
prevTemp = temperature; prevTemp = temperature;
prevFriction = friction; prevFriction = friction;
......
...@@ -313,7 +313,7 @@ void OpenCLNonbondedUtilities::updateNeighborListSize() { ...@@ -313,7 +313,7 @@ void OpenCLNonbondedUtilities::updateNeighborListSize() {
if (!useCutoff) if (!useCutoff)
return; return;
interactionCount->download(); interactionCount->download();
if (interactionCount->get(0) <= interactingTiles->getSize()) if (interactionCount->get(0) <= (unsigned int) interactingTiles->getSize())
return; return;
// The most recent timestep had too many interactions to fit in the arrays. Make the arrays bigger to prevent // The most recent timestep had too many interactions to fit in the arrays. Make the arrays bigger to prevent
......
...@@ -34,17 +34,17 @@ using namespace std; ...@@ -34,17 +34,17 @@ using namespace std;
*/ */
#ifdef _MSC_VER #ifdef _MSC_VER
#include <Windows.h> #include <Windows.h>
static long getTime() { static long long getTime() {
FILETIME ft; FILETIME ft;
GetSystemTimeAsFileTime(&ft); // 100-nanoseconds since 1-1-1601 GetSystemTimeAsFileTime(&ft); // 100-nanoseconds since 1-1-1601
ULARGE_INTEGER result; ULARGE_INTEGER result;
result.LowPart = ft.dwLowDateTime; result.LowPart = ft.dwLowDateTime;
result.HighPart = ft.dwHighDateTime; result.HighPart = ft.dwHighDateTime;
return result/10; return result.QuadPart/10;
} }
#else #else
#include <sys/time.h> #include <sys/time.h>
static long getTime() { static long long getTime() {
struct timeval tod; struct timeval tod;
gettimeofday(&tod, 0); gettimeofday(&tod, 0);
return 1000000*tod.tv_sec+tod.tv_usec; return 1000000*tod.tv_sec+tod.tv_usec;
...@@ -74,7 +74,7 @@ private: ...@@ -74,7 +74,7 @@ private:
class OpenCLParallelCalcForcesAndEnergyKernel::FinishComputationTask : public OpenCLContext::WorkTask { class OpenCLParallelCalcForcesAndEnergyKernel::FinishComputationTask : public OpenCLContext::WorkTask {
public: public:
FinishComputationTask(ContextImpl& context, OpenCLContext& cl, OpenCLCalcForcesAndEnergyKernel& kernel, FinishComputationTask(ContextImpl& context, OpenCLContext& cl, OpenCLCalcForcesAndEnergyKernel& kernel,
bool includeForce, bool includeEnergy, double& energy, long& completionTime) : context(context), cl(cl), kernel(kernel), bool includeForce, bool includeEnergy, double& energy, long long& completionTime) : context(context), cl(cl), kernel(kernel),
includeForce(includeForce), includeEnergy(includeEnergy), energy(energy), completionTime(completionTime) { includeForce(includeForce), includeEnergy(includeEnergy), energy(energy), completionTime(completionTime) {
} }
void execute() { void execute() {
...@@ -91,7 +91,7 @@ private: ...@@ -91,7 +91,7 @@ private:
OpenCLCalcForcesAndEnergyKernel& kernel; OpenCLCalcForcesAndEnergyKernel& kernel;
bool includeForce, includeEnergy; bool includeForce, includeEnergy;
double& energy; double& energy;
long& completionTime; long long& completionTime;
}; };
OpenCLParallelCalcForcesAndEnergyKernel::OpenCLParallelCalcForcesAndEnergyKernel(string name, const Platform& platform, OpenCLPlatform::PlatformData& data) : OpenCLParallelCalcForcesAndEnergyKernel::OpenCLParallelCalcForcesAndEnergyKernel(string name, const Platform& platform, OpenCLPlatform::PlatformData& data) :
......
...@@ -76,7 +76,7 @@ private: ...@@ -76,7 +76,7 @@ private:
class FinishComputationTask; class FinishComputationTask;
OpenCLPlatform::PlatformData& data; OpenCLPlatform::PlatformData& data;
std::vector<Kernel> kernels; std::vector<Kernel> kernels;
std::vector<long> completionTimes; std::vector<long long> completionTimes;
std::vector<int> contextTiles; std::vector<int> contextTiles;
}; };
......
...@@ -87,7 +87,7 @@ public: ...@@ -87,7 +87,7 @@ public:
sortKernelSize = rangeKernelSize/2; sortKernelSize = rangeKernelSize/2;
if (rangeKernelSize > length) if (rangeKernelSize > length)
rangeKernelSize = length; rangeKernelSize = length;
int maxLocalBuffer = (context.getDevice().getInfo<CL_DEVICE_LOCAL_MEM_SIZE>()/sizeof(TYPE))/2; int maxLocalBuffer = (int) ((context.getDevice().getInfo<CL_DEVICE_LOCAL_MEM_SIZE>()/sizeof(TYPE))/2);
if (sortKernelSize > maxLocalBuffer) if (sortKernelSize > maxLocalBuffer)
sortKernelSize = maxLocalBuffer; sortKernelSize = maxLocalBuffer;
int targetBucketSize = sortKernelSize/2; int targetBucketSize = sortKernelSize/2;
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment