| commit | 04f0fd640d78ff715cb9409f03d55366ed76966e | |
| author | Rajalakshmi Srinivasaraghavan <raji@linux.vnet.ibm.com> | |
| Tue, 7 Feb 2017 05:10:26 +0000 (7 10:40 +0530) | ||
| committer | Rajalakshmi Srinivasaraghavan <raji@linux.vnet.ibm.com> | |
| Tue, 7 Feb 2017 05:10:26 +0000 (7 10:40 +0530) | ||
| tree | dfef9d58915c0a6788b1c76524949a8ed5546e5f | treesnapshot (tar.gz zip) |
| parent | dcd4cd575678c94f042c1a6b08663c648781a9eb | commitdiff |
powerpc: Improve strcmp performance for shorter strings
For strings >16B and <32B existing algorithm takes more time than default
implementation when strings are placed closed to end of page. This is due
to byte by byte access for handling page cross. This is improved by
following >32B code path where the address is adjusted to aligned memory
before doing load doubleword operation instead of loading bytes.
Tested on powerpc64 and powerpc64le.
For strings >16B and <32B existing algorithm takes more time than default
implementation when strings are placed closed to end of page. This is due
to byte by byte access for handling page cross. This is improved by
following >32B code path where the address is adjusted to aligned memory
before doing load doubleword operation instead of loading bytes.
Tested on powerpc64 and powerpc64le.
| ChangeLog | diffblobblamehistory | |
| sysdeps/powerpc/powerpc64/power8/strcmp.S | diffblobblamehistory | |
| sysdeps/powerpc/powerpc64/power9/strcmp.S | diffblobblamehistory |