[llvm] r212638 - AArch64: Better codegen for storing to __fp16.

[Jim Grosbach]

Yay crossing emails. We’re saying the same thing now, I think. Clang is definitely broken here, and has been ever since __fp16 was introduced back in 2011. We should fix it.

-j
> On Jul 10, 2014, at 12:23 PM, Tim Northover <t.p.northover at gmail.com> wrote:
> 
>> (*) As Tim and I discussed off-list yesterday, __fp16 isn’t mentioned
>> anywhere by C / C++, so there isn’t a strong standards requirement
>> here.  However, in order for the type to be useful, we should attempt
>> to have __fp16 conform to the IEEE-754 rules for the “binary16" interchange
>> type, which includes getting these conversions “right” (i.e. conforming
>> to the usual rules for conversions that apply to all FP types).
> 
> Yep. I definitely agree that if "(double)*my_fp16" means anything, it
> should be a single-step conversion. Which means we should fix Clang's
> CodeGen.
> 
> Cheers.
> 
> Tim.