2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2014,2015,2016,2017, by the GROMACS development team, led by
5 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
6 * and including many others, as listed in the AUTHORS file in the
7 * top-level source directory and at http://www.gromacs.org.
9 * GROMACS is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public License
11 * as published by the Free Software Foundation; either version 2.1
12 * of the License, or (at your option) any later version.
14 * GROMACS is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with GROMACS; if not, see
21 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
22 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
24 * If you want to redistribute modifications to GROMACS, please
25 * consider that scientific software is very special. Version
26 * control is crucial - bugs must be traceable. We will be happy to
27 * consider code for inclusion in the official distribution, but
28 * derived work must not be called official GROMACS. Details are found
29 * in the README & COPYING files - if they are missing, get the
30 * official version at http://www.gromacs.org.
32 * To help us fund GROMACS development, we humbly ask that you cite
33 * the research papers on the package. Check out http://www.gromacs.org.
35 /*! \libinternal \file
36 * \brief Declare utility routines for OpenCL
38 * \author Anca Hamuraru <anca@streamcomputing.eu>
41 #ifndef GMX_GPU_UTILS_OCLUTILS_H
42 #define GMX_GPU_UTILS_OCLUTILS_H
46 #include "gromacs/gpu_utils/gmxopencl.h"
47 #include "gromacs/utility/gmxassert.h"
49 /*! \brief OpenCL vendor IDs */
51 OCL_VENDOR_NVIDIA = 0,
58 * \brief OpenCL GPU device identificator
60 * An OpenCL device is identified by its ID.
61 * The platform ID is also included for caching reasons.
65 cl_platform_id ocl_platform_id; /**< Platform ID */
66 cl_device_id ocl_device_id; /**< Device ID */
70 * \brief OpenCL device information.
72 * The OpenCL device information is queried and set at detection and contains
73 * both information about the device/hardware returned by the runtime as well
74 * as additional data like support status.
76 struct gmx_device_info_t
78 ocl_gpu_id_t ocl_gpu_id; /**< device ID assigned at detection */
79 char device_name[256]; /**< device name */
80 char device_version[256]; /**< device version */
81 char device_vendor[256]; /**< device vendor */
82 int compute_units; /**< number of compute units */
83 int adress_bits; /**< number of adress bits the device is capable of */
84 int stat; /**< device status takes values of e_gpu_detect_res_t */
85 ocl_vendor_id_t vendor_e; /**< device vendor as defined by ocl_vendor_id_t */
89 * \brief OpenCL GPU runtime data
91 * The device runtime data is meant to hold objects associated with a GROMACS rank's
92 * (thread or process) use of a single device (multiple devices per rank is not
93 * implemented). These objects should be constructed at ther point where a device
94 * dets assigned to a rank and released at when this assignment is no longer valid
95 * (i.e. at cleanup in the current implementation).
98 struct gmx_device_runtime_data_t
100 cl_context context; /**< OpenCL context */
101 cl_program program; /**< OpenCL program */
105 /*! \brief Launches asynchronous host to device memory copy. */
106 int ocl_copy_H2D_async(cl_mem d_dest, void * h_src,
107 size_t offset, size_t bytes,
108 cl_command_queue command_queue,
109 cl_event *copy_event);
111 /*! \brief Launches asynchronous device to host memory copy. */
112 int ocl_copy_D2H_async(void * h_dest, cl_mem d_src,
113 size_t offset, size_t bytes,
114 cl_command_queue command_queue,
115 cl_event *copy_event);
117 /*! \brief Launches synchronous host to device memory copy. */
118 int ocl_copy_H2D_sync(cl_mem d_dest, void * h_src,
119 size_t offset, size_t bytes,
120 cl_command_queue command_queue);
122 /*! \brief Allocate host memory in malloc style */
123 void ocl_pmalloc(void **h_ptr, size_t nbytes);
125 /*! \brief Free host memory in malloc style */
126 void ocl_pfree(void *h_ptr);
128 /*! \brief Convert error code to diagnostic string */
129 std::string ocl_get_error_string(cl_int error);
131 /*! \brief Calls clFinish() in the stream \p s.
133 * \param[in] s stream to synchronize with
135 static inline void gpuStreamSynchronize(cl_command_queue s)
137 cl_int cl_error = clFinish(s);
138 GMX_RELEASE_ASSERT(CL_SUCCESS == cl_error,
139 ("Error caught during clFinish:" + ocl_get_error_string(cl_error)).c_str());