Re: [PATCH-1v2, rs6000] Enable SImode in FP registers on P7 [PR88558]

["Kewen.Lin" <linkw@linux.ibm.com>]

Hi,

on 2023/9/25 09:57, HAO CHEN GUI wrote:
> Hi Kewen,
> 
> 在 2023/9/18 15:34, Kewen.Lin 写道:
>> hanks for checking!  So for P7, this patch looks neutral, but for P8 and
>> later, it may cause some few differences in code gen.  I'm curious that how
>> many total object files and different object files were checked and found
>> on P8?  
> P8 with -O2, following object files are different.
> 507.cactuBSSN_r datestamp.o
> 511.povray_r colutils.o
> 521.wrf_r module_cu_kfeta.fppized.o
> 526.blender_r particle_edit.o
> 526.blender_r glutil.o
> 526.blender_r displist.o
> 526.blender_r CCGSubSurf.o
> 
> P8 with -O3, following object files are different.
> 502.gcc_r ifcvt.o
> 502.gcc_r rtlanal.o
> 548.exchange2_r exchange2.fppized.o
> 507.cactuBSSN_r datestamp.o
> 511.povray_r colutils.o
> 521.wrf_r module_bc.fppized.o
> 521.wrf_r module_cu_kfeta.fppized.o
> 526.blender_r particle_edit.o
> 526.blender_r displist.o
> 526.blender_r CCGSubSurf.o
> 526.blender_r sketch.o
> 

OK, it's concluded that the percentage of the total number of affected object
files is quite small ...

> 
> 
> 
>> https://gcc.gnu.org/pipermail/gcc-patches/2023-February/612821.html
>> I also wonder if it's easy to reduce some of them further as small test cases.
>>
>> Since xxlor is better than fmr at least on Power10, could you also evaluate
>> the affected bmks on P10 (even P8/P9) to ensure no performance degradation?
> There is no performance recession on P10/P9/P8. The detail data is listed on
> internal issue.

... and no runtime performance impact as evaluated, so this patch looks good to
me and thanks for further testing.

Please wait for a week or so to see if Segher and David have comments.  Thanks!

BR,
Kewen

> 
> Thanks
> Gui Haochen