SH optimized software floating point routines
Naveen H. S
Naveen.S@kpitcummins.com
Fri Jul 16 10:04:00 GMT 2010
Hi,
>> you can free to propose a complete and regtested patch for SH
>> assembly soft fp against trunk.
Please find attached the ported soft float patch "sh_softfloat.patch".
The original patch was posted at the following link by Joern RENNECKE.
http://gcc.gnu.org/ml/gcc-patches/2006-09/msg00614.html
The following modifications have been done in the patch.
sched-deps.c :- Hunk was not applied due to modifications in the
current gcc source.
t-sh :- divsf3, extendsfdf2 and truncdfsf2 routines are not included
as they resulted in some regressions.
divsf3 - c-c++-common/torture/complex-sign-mixed-div.c
extendsfdf2 - gcc.c-torture/execute/conversion.c
gcc.dg/torture/fp-int-convert-float.c, gcc.dg/pr28796-2.c
gcc.dg/torture/type-generic-1.c
sh.md :- cbranchsf4, cbranchdf4, cstoresf4, cstoredf4 instruction
patterns are not included as they are already present in current source.
Modifying the routines referring patch resulted in build failure.
Branch and other related instruction patterns bgt, blt, ble, bge, bleu,
bunle, bunordered, bunlt, bunge, bungt, seq, sge, sgeu, sne etc are not
modified as they should not be present in current gcc source.
After the above modifications, the regressions are reduced. However,
there are some regressions in SH3 and related (m4-nofpu,
m4a-nofpu and m2a-single) targets.
gcc.c-torture/execute/20060420-1.c - O1 and O2 execution failures
gcc.dg/fold-overflow-1.c scan-assembler-times 2139095040
We are working on fixing the above failures and also extendsfdf2, divsf3
and truncdfsf2 routines. Please review the patch and let us know if
there should be any modifications in it.
Please let us know if there are any inputs in resolving the failures and
routines.
>> It would be better to try soft-fp on SH
Please find attached the patch ""sh_softfp.patch" which implements basic
support of soft-fp for SH target. There were no regressions found with
the patch. Please let us know if there should be any further improvements
required for complete soft-fp support.
Thanks for the co-operation and support.
Regards,
Naveen.H.S
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sh_softfloat.patch
Type: application/octet-stream
Size: 270966 bytes
Desc: sh_softfloat.patch
URL: <https://gcc.gnu.org/pipermail/gcc/attachments/20100716/b602c5ad/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sh_softfp.patch
Type: application/octet-stream
Size: 5477 bytes
Desc: sh_softfp.patch
URL: <https://gcc.gnu.org/pipermail/gcc/attachments/20100716/b602c5ad/attachment-0001.obj>
More information about the Gcc
mailing list