2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 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 #ifndef GMX_GPU_UTILS_CUDA_KERNEL_UTILS_CUH
36 #define GMX_GPU_UTILS_CUDA_KERNEL_UTILS_CUH
39 * \brief CUDA device util functions (callable from GPU kernels).
41 * \author Szilard Pall <pall.szilard@gmail.com>
44 #include "gromacs/gpu_utils/cuda_arch_utils.cuh"
46 /*! Load directly or using __ldg() when supported. */
48 __device__ __forceinline__ T LDG(const T* ptr)
50 #if GMX_PTX_ARCH >= 350
51 /* CC >=3.5 supports constant loads through texture or L1 */
54 /* Device does not have LDG support, fall back to direct load. */
59 /*! \brief Fetch the value by \p index from the texture object or reference.
60 * Fetching from the object is the preferred behaviour on CC >= 3.0.
62 * \tparam[in] T Raw data type
63 * \param[in] texObj Table texture object
64 * \param[in] texRef Table texture reference
65 * \param[in] index Non-negative element index
66 * \returns The value from the table at \p index
69 static __forceinline__ __device__
70 T fetchFromTexture(const cudaTextureObject_t texObj,
71 const struct texture<T, 1, cudaReadModeElementType> texRef,
75 assert(!c_disableCudaTextures);
77 #if GMX_PTX_ARCH >= 300 // Preferring texture objects on any new arch
78 GMX_UNUSED_VALUE(texRef);
79 result = tex1Dfetch<T>(texObj, index);
81 GMX_UNUSED_VALUE(texObj);
82 result = tex1Dfetch(texRef, index);
87 /*! \brief Fetch the value by \p index from the parameter lookup table.
89 * Depending on what is supported, it fetches parameters either
90 * using direct load, texture objects, or texture references.
92 * \tparam[in] T Raw data type
93 * \param[in] d_ptr Device pointer to the raw table memory
94 * \param[in] texObj Table texture object
95 * \param[in] texRef Table texture reference
96 * \param[in] index Non-negative element index
97 * \returns The value from the table at \p index
100 static __forceinline__ __device__
101 T fetchFromParamLookupTable(const T *d_ptr,
102 const cudaTextureObject_t texObj,
103 const struct texture<T, 1, cudaReadModeElementType> texRef,
108 #if DISABLE_CUDA_TEXTURES
109 GMX_UNUSED_VALUE(texObj);
110 GMX_UNUSED_VALUE(texRef);
111 result = LDG(d_ptr + index);
113 GMX_UNUSED_VALUE(d_ptr);
114 result = fetchFromTexture<T>(texObj, texRef, index);
120 #endif /* GMX_GPU_UTILS_CUDA_KERNEL_UTILS_CUH */