[PATCH] Devirtualize TerminatorInst

[David Majnemer via llvm-commits]

Isn't it a little bit of a misnomer to continue to call it 'getSuccessorV',
etc.?  IMO, something like getSuccessorImpl would be more fitting.

On Tue, Sep 1, 2015 at 3:41 PM, Pete Cooper via llvm-commits <
llvm-commits at lists.llvm.org> wrote:

> Hi Duncan, David
>
> I took a break from devirtualizing Value as David raised performance
> concerns I wanted to check before continuing.
>
> The original series of patches and discussion were around here:
> http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20150622/282810.html
>
> This is what was patch 0005 in that series, but updated now that we have
> additional TerminatorInst’s since then.
>
> I ran our performance suite here and there were no regressions.  It
> actually slightly improved (about 2-3%)
> MultiSource/Benchmarks/TSVC/ControlLoops-dbl/ControlLoops-dbl.
>
> Assuming this patch lands ok, i’ll continue with the same testing
> methodology for the remainder of the series.
>
> Cheers,
> Pete
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150901/ad22d618/attachment.html>