[llvm-commits] [Review Request][Patch]Introduce the VectorizeConfig class.

Sebastian Pop spop at codeaurora.org
Fri Apr 6 10:28:18 PDT 2012

On Fri, Apr 6, 2012 at 9:49 AM, Hal Finkel <hfinkel at anl.gov> wrote:
> LGTM. Sebastian, does this fit what you had in mind?

Looks definitely better.  Thanks ether for fixing this.

However I still have a problem with the negative flags like:

NoFMA("bb-vectorize-no-fma", cl::init(false), cl::Hidden,
  cl::desc("Don't try to vectorize the fused-multiply-add intrinsic"));

It's probably just that I'm not yet used to the way LLVM specifies flags:
I am used to define positive flags and have a systematic way to infer
a negative from the positive name.

I'll have to get used to double negations to specify flags in LLVM...

Qualcomm Innovation Center, Inc is a member of Code Aurora Forum

More information about the llvm-commits mailing list