http://developer.amd.com/tools/hc/AMDAPPSDK/assets/AMD_Accelerated_Parallel_Processing_OpenCL_Progra...
On documentation page A-4
A.8.1 cl_amd_fp64
Before using double data types, double-precision floating point operators, and/or
double-precision floating point routines in OpenCL™ C kernels, include the
#pragma OPENCL EXTENSION cl_amd_fp64 : enable directive. See Table A.1
for a list of supported routines.
However when you go to see page A-15 there is a section:
A.10 Supported Functions for cl_amd_fp64 / cl_khr_fp64
AMD OpenCL is now cl_khr_fp64-compliant on devices compliant with OpenCL
1.1 and greater (every GPU later than 7xx, and all CPUs). Thus, cl_amd_fp64
is now a synonym for cl_khr_fp64 on all supported devices.
But the Table A.1 only lists devices which support devices which support cl_amd_fp64 only.
Also, what is 7xx? We know that 5870 also supports fp64 ? and not all 7xxx GPUs support it either. What about mobile APUs?
Documentation Table A.1 shows Trinity APUs do not support fp64, yet OpenCL is showing support for it in clinfo?
http://images.anandtech.com/reviews/mobile/Trinity-CLinfo.zip
(also mentioned at thread http://devgurus.amd.com/message/1283794 moreover, DP support does not seem to be uniformly implemented on Windows and Linux platforms)
I think these sections require some clarification and corrections!