[llvm-commits] [llvm] r76615 - in /llvm/trunk: include/llvm/CodeGen/LiveIntervalAnalysis.h lib/CodeGen/LiveIntervalAnalysis.cpp

[Evan Cheng]

On Jul 22, 2009, at 12:49 PM, Daniel Dunbar wrote:

> On Wed, Jul 22, 2009 at 11:45 AM, David Greene<dag at cray.com> wrote:
>> On Wednesday 22 July 2009 13:00, Evan Cheng wrote:
>>
>>>> Again, I didn't see the report.  Can someone send it to me?
>>>
>>> You were explicitly cc'd on this email. Your patch 76486 slowed down
>>> compilation across the board. I'd prefer you rip it out until you  
>>> have
>>> fixed the severe performance problem.
>>
>> Hmm.  I never saw this.  I havew no idea what happened to it.
>
>> Everything should be turned off without -asm-verbose.  Do you still  
>> see
>> performance issues?
>
> This is on a standard llvm-test setup. Does llvm-test use -asm- 
> verbose?

llc defaults to -asm-verbose. Regardless, 30%-40% regression is way  
too much.

Evan

>
>>> I don't know you missed the entire thread. Your email was on  
>>> recipient
>>> list.
>>
>> I don't know either.  Which address did you send it to?
>
> I sent it to greened at obbligato.org. Note that the thread contained
> llvm-testresults, is it possible that went to a differently filtered
> mailbox?
>
> On Wed, Jul 22, 2009 at 12:41 PM, David Greene<dag at cray.com> wrote:
>>> You were explicitly cc'd on this email. Your patch 76486 slowed down
>>> compilation across the board. I'd prefer you rip it out until you  
>>> have
>>> fixed the severe performance problem.
>>
>> Is there a standard way to run compile-time tests?
>
> I think llvm-test is the current best standard we have.
>
> - Daniel
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits