[PATCH] D21560: Relax the clearance calculating for breaking partial register dependency.

[Dehao Chen via llvm-commits]

In terms of code size, for speccpu2006int, the text size change is listed
below:

xalan 0.00%
astar 0.10%
bzip2 0.00%
gcc 0.00%
gobmk 0.02%
h265 0.03%
hmmer 0.12%
libquantum 0.00%
mcf 0.00%
omnetpp 0.04%
perlbench 0.00%
sjeng 0.00%
geomean 0.03%

Dehao

On Mon, Jun 27, 2016 at 7:50 PM, Dehao Chen <danielcdh at gmail.com> wrote:

> It's sandybridge. Each benchmark is repeated 3 times and get the medium.
> But by looking at the raw data, the run-to-run variation are pretty large
> (2%~5% range across 3 iterations). Thus I would suspect there might be some
> noise in the result.
>
> Dehao
>
> On Mon, Jun 27, 2016 at 7:19 PM, Xinliang David Li <davidxl at google.com>
> wrote:
>
>> What platform, Ivybridge ?
>>
>> David
>>
>> On Mon, Jun 27, 2016 at 6:28 PM, Dehao Chen <danielcdh at gmail.com> wrote:
>>
>>> danielcdh added a comment.
>>>
>>> I just tested the perf impact of this patch for SPECCPU INT2006
>>> benchmarks:
>>>
>>> 400.perlbench   2.21%
>>> 401.bzip2       -0.46%
>>> 403.gcc 1.59%
>>> 429.mcf 1.79%
>>> 445.gobmk       1.52%
>>> 456.hmmer       1.63%
>>> 458.sjeng       5.62%
>>> 462.libquantum  1.54%
>>> 464.h264ref     -1.28%
>>> 471.omnetpp     0.00%
>>> 473.astar       2.35%
>>> 483.xalancbmk   -0.27%
>>> overall 1.29%
>>>
>>> About the -Oz issue, I agree that there should be no xor inserted at
>>> -Oz. But that's a different issue and should be fixed in a separate patch.
>>>
>>>
>>> http://reviews.llvm.org/D21560
>>>
>>>
>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160627/3beb0085/attachment.html>