On Wed, 22 Apr 2015, Alan Lawrence wrote: > [Resending with correct in-reply-to header] > > This adds basic support for moving __fp16 values around, passing and > returning, and operating on them by promoting to 32-bit floats. Also a few > scalar testcases.
I'd think it would be desirable to share tests between ARM and AArch64 as far as possible (where applicable to both - so not the tests for the alternative format, and some of the gcc.target/arm/fp16-* tests using scan-assembler might need adapting to work for AArch64). To the extent that the ARM implementation follows an old specification and AArch64 is following newer ACLE, of course, this might require ARM to be updated to follow newer ACLE before tests can be shared. (For example, the older specification implemented for ARM includes double rounding when converting from double to fp16, but ACLE specifies single rounding.) Longer term, I'd hope the aim would be for semantics to follow TS 18661-3 (DTS ballot recently passed), using the standard name _Float16 (and FLT_EVAL_METHOD == 32 as the nearest equivalent of the promotions to float). -- Joseph S. Myers jos...@codesourcery.com
