I am a developer for Intel's OpenCL backend driver, beignet.
I find the suffix 'h' or 'H' for constant half float like:
0.1h or 0.2H
is not supported, even when I enable the OPENCL EXTENSION cl_khr_fp16 pragma.
I check the code of LiteralSupport.cpp under Lex, find that the NumericLiteralParser
seems never check H or h suffix for float, no matter what options for Lex.
The OpenCL 1.2 says:
The OpenCL compiler accepts an h and H suffix on floating point literals, indicating the literal is typed as a half.
So is this a bug?