Displaying 2 results from an estimated 2 matches for "d48466".
Did you mean:
d46466
2018 Jul 03
2
Question about canonicalizing cmp+select
...gt; I think you are asking if we are missing a fold (or your target is missing
> enabling another hook) to transform the sext+add into shift+or? I think the
> answer is 'yes'. We probably should add that fold. This seems like a
> similar case as the recent: https://reviews.llvm.org/D48466
>
> Note that on x86, the sext+add becomes zext+sub:
> t20: i8 = setcc t3, Constant:i16<-1>, setgt:ch
> t24: i16 = zero_extend t20
> t17: i16 = sub Constant:i16<5>, t24
>
> Would that transform help your target?
>
> On Tue, Jul 3, 2018 at 3:5...
2018 Jul 03
4
Question about canonicalizing cmp+select
Hi, Sanjay/all,
I noticed in rL331486 that some compare-select optimizations are disabled
in favor of providing canonicalized cmp+select to the backend.
I am currently working on a private backend target, and the target has a
small code size limit. With this change, some of the apps went over the
codesize limit. As an example,
C code:
b = (a > -1) ? 4 : 5;
ll code:
Before rL331486: