Provided by: opencl-1.2-man-doc_1.0~svn22836-1.1_all
atom_cmpxchg - 64-bit atom_cmpxchg functions. long atom_cmpxchg(volatile global(3clc) long *p, long cmp, long val); long atom_cmpxchg(volatile global(3clc) long *p, long cmp, long val); ulong atom_cmpxchg(volatile local(3clc) ulong *p, ulong cmp, ulong val); ulong atom_cmpxchg(volatile local(3clc) ulong *p, ulong cmp, ulong val);
Read the 64-bit value (referred to as old) stored at location pointed by p. Compute (old == cmp) ? val : old and store result at location pointed by p. The function returns old. An application that wants to use this extension will need to include the #pragma OPENCL EXTENSION cl_khr_int64_base_atomics : enable in the OpenCL program source. This function is performed in one atomic transaction. These transactions are atomic for the device executing these atomic functions. There is no guarantee of atomicity if the atomic operations to the same memory location are being performed by kernels executing on multiple devices. Atomic operations on 64-bit integers and 32-bit integers (and float) are also atomic with regard to each other.
atomicFunctions(3clc), atomic_cmpxchg(3clc), cl_khr_int64_base_atomics(3clc)
The Khronos Group
Copyright © 2007-2011 The Khronos Group Inc. Permission is hereby granted, free of charge, to any person obtaining a copy of this software and/or associated documentation files (the "Materials"), to deal in the Materials without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Materials, and to permit persons to whom the Materials are furnished to do so, subject to the condition that this copyright notice and permission notice shall be included in all copies or substantial portions of the Materials.
1. OpenCL Specification page 279, section 6.12.11 - Atomic Functions for 32-bit integers