[llvm-commits] [PATCH] Integer max/min intrinsics

[Nadav Rotem]

The code in this patch looks good.

The problem is that we can't test this patch without the second half of the patch that canonicalizes select/cmp into min/max intrinsics. 

Nadav

On Dec 17, 2012, at 11:58 AM, "Redmond, Paul" <paul.redmond at intel.com> wrote:

> Hi,
> 
> The attached patch adds the integer max/min intrinsics discussed here: http://lists.cs.uiuc.edu/pipermail/llvmdev/2012-December/056785.html
> 
> This patch includes:
> 
> - new smin/smax/umin/umax intrinsics
> - generic lowering to cmp-select during selection dag building
> - language ref documentation
> - basic lit tests
> 
> This patch does not include:
> 
> - optimizations operating on the new intrinsics (e.g., no constant folding)
> - compare-select to max/min transformations
> 
> The intention is to add support for optimization/transformation in subsequent patches. Also, floating point max/min intrinsics will be addressed separately.
> 
> 
> paul
> <int_maxmin.diff>_______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits