[llvm-dev] RFC: [GlobalISel] Representing fp types in LLT

[Jay Foad via llvm-dev]

On Tue, 8 Dec 2020 at 19:04, Justin Bogner via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
>
> For a while now there have been discussions here and there about
> representing floating point types explicitly in LLT. Having this type
> information explicitly makes RegBankSelect easier to implement, avoids
> inefficiencies in disambiguating which operation is needed to lower
> various operations, and is necessary for correctness in light of variant
> floating point types like bfloat.
>
> I believe there is general consensus that we need this, so this RFC is
> about how we should go about modeling these types.
>
> At the dev meeting, we discussed two approaches IIRC:
>
> 1. Explicitly encode the number of bits in the exponent and mantissa for
>    flexibility
> 2. Encode only the types we need (ie, model the same fp types as
>    llvm::Type does)
>
> The idea with (1) was that this way we could easily handle IEEE-754
> binary formats and variants like bfloat16 and tf32 without needing to
> burden ISel with knowledge of specifics. The major drawback of this is
> that it would struggle to represent types that don't fit the general
> mold, like ppc-fp128 (which is two doubles) or the IEEE 754-2008 decimal
> formats.
>
> I think (2) is the safer way forward

I strongly agree!

Jay.