[llvm-commits] [Patch] Add support for v16i32 and v16i64 types to codegen

Villmow, Micah Micah.Villmow at amd.com
Thu Jul 26 12:20:40 PDT 2012


This patch adds support for v16i32 and v16i64 types to code gen. This is required because in parts of the code base in the legalizer phase, the query getSetCCResultsType is called on a v16i8, which on the AMDIL architecture needs to return a v16i32 for i8/i16/f32/i32 and v16i64 for f64/i64 types. This adds in support for this.

The attached test case shows the issue, but it doesn't appear on any in-tree backends. The patch is attached.
I have more patches coming, but trying to keep individual patches small and self contained.
Thanks,
Micah
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120726/30247c7a/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test_char16_binary_op_add_1052.ll
Type: application/octet-stream
Size: 1346 bytes
Desc: test_char16_binary_op_add_1052.ll
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120726/30247c7a/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: codegen_new_vec16_types.patch
Type: application/octet-stream
Size: 9808 bytes
Desc: codegen_new_vec16_types.patch
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120726/30247c7a/attachment-0001.obj>


More information about the llvm-commits mailing list