[PATCH] D13427: RFC: faster isa<IntrinsicInst> (bugged tests?)

Daniel Berlin via llvm-commits llvm-commits at lists.llvm.org
Mon Oct 5 18:44:31 PDT 2015


+1.
Please don't get discouraged here.
We should be fixing the underlying inconsistency in how comparisons are done.


On Mon, Oct 5, 2015 at 6:27 PM, Sean Silva via llvm-commits
<llvm-commits at lists.llvm.org> wrote:
> The inconsistency is clearly troubling and is worth fixing. This review may
> have gotten off to a bad start by focusing on the performance aspect. Any
> time correctness and performance get mixed into the same discussion this
> sort of thing tends to happen.
>
> (also, I don't think anybody is really advocating the use of string
> comparisons)
>
> @vedant: do you know if it is important for compatibility for us to use
> strings in the tests Fiona pointed out? Can we just switch those tests to
> use a real intrinsic?
>
> -- Sean Silva
>
> On Mon, Oct 5, 2015 at 5:40 PM, escha <escha at apple.com> wrote:
>>
>> escha abandoned this revision.
>> escha added a comment.
>>
>> Since people don't seem to be particularly interested in fixing
>> broken+inconsistent behavior and believe performing string comparisons is a
>> reasonable and performant way of checking class membership, I'm not going to
>> bother with this.
>>
>>
>> Repository:
>>   rL LLVM
>>
>> http://reviews.llvm.org/D13427
>>
>>
>>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>


More information about the llvm-commits mailing list