[patch, PR44297] prefetch improvements to fix 465.tonto from non-constant step prefetching
Fang, Changpeng
Changpeng.Fang@amd.com
Mon Jun 7 22:27:00 GMT 2010
Hi,
Attached is the patch to fix 465/tonto regression (> 9%) caused (or exposed) by non-constant step prefetching.
The regressions mainly comes from loops which have array references with both constant (prefetch_mod = 8)
and non-constant (prefetch_mod=1) steps. The loops are unrolled 8 times, and 8 non-constant step prefetches
are inserted into the unrolled loops.
In this patch, we re-compute the prefetch count using prefetch_mod and unroll_factor after the unroll_factor
is determined.
The patch passed bootstrapping and gcc regression tests on amd-linux64 system.
Is it OK for the trunk?
Thanks,
Changpeng
More information about the Gcc-patches
mailing list