projects / gcc.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: e4c6a07) | patch
PR 44576: miss rate computation improvement for prefetching loop arrays.
authorChangpeng Fang <changpeng.fang@amd.com>
Fri, 2 Jul 2010 16:34:29 +0000 (16:34 +0000)
committerSebastian Pop <spop@gcc.gnu.org>
Fri, 2 Jul 2010 16:34:29 +0000 (16:34 +0000)
commita245c04bccad9eea35cf21def68ae812fd29a953
tree126ae0acb0967dcd6c470c42705dd86addbd1cc9tree
parente4c6a07ab580140cc8780ff45a3b6e8f552d3df5commit | diff
PR 44576: miss rate computation improvement for prefetching loop arrays.

2010-07-02  Changpeng Fang  <changpeng.fang@amd.com>

PR middle-end/44576
* tree-ssa-loop-prefetch.c (compute_miss_rate): Return 1000 (out
of 1000) for miss rate if the address diference is greater than or
equal to the cache line size (the two reference will never hit the
same cache line).

From-SVN: r161727
gcc/ChangeLog diff | blob | history
gcc/tree-ssa-loop-prefetch.c diff | blob | history