Begin porting string performance tests to benchtests
commitc1f75dc386d533806d29b7e94118363a7b50eed8
authorSiddhesh Poyarekar <siddhesh@redhat.com>
Tue, 16 Apr 2013 12:07:24 +0000 (16 17:37 +0530)
committerSiddhesh Poyarekar <siddhesh@redhat.com>
Tue, 11 Jun 2013 09:38:13 +0000 (11 15:08 +0530)
tree38d917f7a99bf569704a09e56a0dd5b7b4475be5
parent50fd745b4dec07e8e213cf2703b5cabcfa128225
Begin porting string performance tests to benchtests

This is the initial support for string function performance tests,
along with copying tests for memcpy and memcpy-ifunc as proof of
concept.  The string function benchmarks perform operations at
different alignments and for different sizes and compare performance
between plain operations and the optimized string operations.  Due to
this their output is incompatible with the function benchmarks where
we're interested in fastest time, throughput, etc.

In future, the correctness checks in the benchmark tests can be
removed.  Same goes for the performance measurements in the
string/test-*.
ChangeLog
benchtests/Makefile
benchtests/README
benchtests/bench-memcpy-ifunc.c [new file with mode: 0644]
benchtests/bench-memcpy.c [new file with mode: 0644]
benchtests/bench-string.h [new file with mode: 0644]