[llvm-commits] PATCH: remove VICmp and VFCmp.

[Eli Friedman]

On Thu, Jul 9, 2009 at 10:34 AM, David Greene<dag at cray.com> wrote:
> On Thursday 09 July 2009 11:37, Chris Lattner wrote:
>
>> > By the way, I'm pretty sure that gcc lays out vectors of x86 long
>> > double the same as an array.
>>
>> I think vectors should always be stored to memory with the same layout
>> as the equivalent array.
>
> So then how will we represent vector masks in the future, which are
> bit-packed vectors of i1?

What does it mean in terms of the LLVM IR semantics that they're
bit-packed?  Do you actually care about the memory representation of
<8 x i1>, or is it just that you want to be able to convert from an <8
x i1> to an i8?

-Eli