On Tue, 20 Aug 2024, Tamar Christina wrote:
> Hi,
>
> As you know I've been working on removing the code that demotes GIMPLE
> COND_EXPR to GENERIC during vect_recog_bool_pattern.
>
> To restate why, The issue we currently have today is that the mask (boolean
> argument of a COND_EXPR) is not al
Hi,
As you know I've been working on removing the code that demotes GIMPLE
COND_EXPR to GENERIC during vect_recog_bool_pattern.
To restate why, The issue we currently have today is that the mask (boolean
argument of a COND_EXPR) is not always available during pattern matching.
This is a problem
