[llvm-commits] [PATCH] PowerPC Vector Floating-Point Rounding

[Hal Finkel]

----- Original Message -----
> From: "Adhemerval Zanella" <azanella at linux.vnet.ibm.com>
> To: llvm-commits at cs.uiuc.edu
> Sent: Friday, October 5, 2012 1:03:02 PM
> Subject: [llvm-commits]  [PATCH] PowerPC Vector Floating-Point Rounding
> 
> Hi all,
> 
> This patch add support for fptosi/fptoui/sitofp/uitofp operation
> using v4i32 to/from v4f342 types. I added all four transformations
> on the testcases.

Two quick comments:

 1. These instructions are not being explicitly marked as Legal in PPCISelLowering. I suppose this is the default, but we should explicitly mark them as legal for clarity.

 2. This comments seems oddly worded:
> +// Defines with the UIM field set to 0 to being used on floating-point
> +// to integer (fp_to_sint/fp_to_uint) and on integer to floating-point
> +// (sint_to_fp/uint_to_fp) conversions.

How's this:

// Defines with the UIM field set to 0 for floating-point
// to integer (fp_to_sint/fp_to_uint) conversions and integer
// to floating-point (sint_to_fp/uint_to_fp) conversions.

Otherwise, LGTM.

Thanks again,
Hal

> 
> Comments/tips/advices?
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
> 

-- 
Hal Finkel
Postdoctoral Appointee
Leadership Computing Facility
Argonne National Laboratory