[cfe-commits] RFC: Half floating point support

Anton Korobeynikov anton at korobeynikov.info
Wed Oct 5 11:12:10 PDT 2011


Hi John,

> Is there a standard we should be following here?
I think the only "standard" is ARM docs.

> Without one, my inclination is to treat it as a normal floating-point type, without this extra promotion stuff, and then just have IR generation know that half operations should actually be done in float.
I thought about this way, but it looks too big and fragile - almost
every Emit* method should know about this special case, so there will
be a lot of changes almost for "nothing". AST building of promotions
looks much cleaner, because of very few necessary changes.

-- 
With best regards, Anton Korobeynikov
Faculty of Mathematics and Mechanics, Saint Petersburg State University



More information about the cfe-commits mailing list