This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
I was recently trying to turn on some of the debug options I put in when doing the initial power7 work to track down some performance regressions. The following patches fix problems that come up in telling the compiler to use the traditional floating point insns for scalar mode on power7 (-mno-vsx-scalar-double). It also adds reload helpers for using the VSX memory load instructions when the compiler is allowed to use VSX memory instructions for scalars. There were no regressions with these patches. Are they ok to be checked in? 2012-01-09 Michael Meissner <meissner@linux.vnet.ibm.com> * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add DF reload patterns if -mvsx-scalar-memory. * config/rs6000/vsx.md (vsx_xscvspdp): Allow xscvspdp to be generated, even -mno-vsx-scalar-double was used. (vsx_xscvdpsp_scalar): Likewise. (vsx_xscvspdp_scalar2): Likewise. -- Michael Meissner, IBM 5 Technology Place Drive, M/S 2757, Westford, MA 01886-3141, USA meissner@linux.vnet.ibm.com fax +1 (978) 399-6899
Attachment:
gcc-power7.patch307b
Description: Text document
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |