This set of patches adds support for _Float16 (IEEE 754 16-bit floating
point) and for __bfloat16 (Google brain 16-bit floating point format)
to the PowerPC.
The PowerPC does not have native 16-bit floating point support other
than conversions and support within the MMA unit.
Power9 has both scalar and vector conversions between _Float16 and
32-bit floating point.
Power10 has only vector conversions between __bfloat16 and 32-bit
floating point.
There are 10 patches in this patch set.
I have committed all of the patches in my backlog (dense math registers, other
-mcpu=future instructions, random bug fixes, support for _Float16 and
__bfloat16, and optimizations for vector logical operations on power10/power11)
into the IBM vendor branch:
vendors/ibm/gcc-17-future
--
Michael Meissner, IBM
PO Box 98, Ayer, Massachusetts, USA, 01432
email: [email protected]
