[llvm-commits] [PATCH] Break 256-bit vector int add/sub/mul into two 128-bit operations
Bruno Cardoso Lopes
bruno.cardoso at gmail.com
Tue Aug 23 00:12:18 PDT 2011
Hi Craig,
Look great, just some minor stuff:
1) Place some asserts into LowerADD and LowerSUB before calling Lower256IntOp.
2) Remove the ISD::MUL,ADD,SUB assert from Lower256IntOp.
3) Rename Lower256IntOp to Lower256IntArith
After fixing this please commit!
On Mon, Aug 22, 2011 at 11:48 PM, Craig Topper <craig.topper at gmail.com> wrote:
> This avoids costly scalarization. Unfortunately, there is no 128-bit
> byte multiply operation in X86 so I didn't break up MUL of v32i8.
> Fixes PR10711.
>
> --
> ~Craig
>
--
Bruno Cardoso Lopes
http://www.brunocardoso.cc
More information about the llvm-commits
mailing list