[PATCH][LoopVectorizer] Enable fsub reductions

James Molloy james.molloy at arm.com
Fri Aug 8 05:50:39 PDT 2014


Thanks Arnold,

 

Committed in r215200.

 

Cheers,

 

James

 

From: Arnold Schwaighofer [mailto:aschwaighofer at apple.com] 
Sent: 08 August 2014 01:21
To: James Molloy
Cc: llvm-commits
Subject: Re: [PATCH][LoopVectorizer] Enable fsub reductions

 

 

On Aug 7, 2014, at 1:14 PM, James Molloy <James.Molloy at arm.com> wrote:

 

Hi Arnold,

 

The attached patch enables floating-point subtraction reductions in
fast-math mode.

 

I'm not sure why it wasn't enabled before - the logic seems sound and I
can't find any FP math pitfalls.

 

I can't see a reason either. LGTM.

 





This, coupled with making the loop vectorizer run before SLP, should give us
about 10% in sphinx3 in SPEC.

 

Very nice!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140808/cdc4ba38/attachment.html>


More information about the llvm-commits mailing list