[llvm-commits] [llvm] r65310 - /llvm/trunk/tools/lto/LTOCodeGenerator.cpp

Tanya M. Lattner tonic at nondot.org
Mon Feb 23 09:22:41 PST 2009


No. At this stage it has to be a regression in llvm-test or dejagnu tests 
(failure, not performance).

-Tanya

On Mon, 23 Feb 2009, Duncan Sands wrote:

> On Monday 23 February 2009 08:56:07 Evan Cheng wrote:
>> Wow. How did that happen? That would explain why LTO has been
>> pessimizing code! "Fast" codegen means fast isel and disabling a lot
>> of codegen optimizations. Thanks for catching this.
>
> Maybe this should go in the release?
>
> Ciao,
>
> D.
>



More information about the llvm-commits mailing list