diff options
author | spop <spop@138bc75d-0d04-0410-961f-82ee72b054a4> | 2010-05-07 16:15:52 +0000 |
---|---|---|
committer | spop <spop@138bc75d-0d04-0410-961f-82ee72b054a4> | 2010-05-07 16:15:52 +0000 |
commit | 3fa57e8490fea744e154c99d247822c1a44dd475 (patch) | |
tree | 7421bcfd46f6420c186e05c988f9bdf591a97bf6 /gcc/tree-ssa-loop-prefetch.c | |
parent | 3665c1ba3f3c8290fde59822a52ce6e8324d4612 (diff) | |
download | gcc-3fa57e8490fea744e154c99d247822c1a44dd475.tar.gz |
Account for loop unrolling in the insn-to-prefetch ratio heuristic.
2010-05-07 Changpeng Fang <changpeng.fang@amd.com>
* tree-ssa-loop-prefetch.c (is_loop_prefetching_profitable):
Account for loop unrolling in the insn-to-prefetch ratio heuristic.
(loop_prefetch_arrays): Pass to is_loop_prefetching_profitable
the unroll_factor.
git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@159162 138bc75d-0d04-0410-961f-82ee72b054a4
Diffstat (limited to 'gcc/tree-ssa-loop-prefetch.c')
-rw-r--r-- | gcc/tree-ssa-loop-prefetch.c | 21 |
1 files changed, 13 insertions, 8 deletions
diff --git a/gcc/tree-ssa-loop-prefetch.c b/gcc/tree-ssa-loop-prefetch.c index 6f879887c9f..38d8f233655 100644 --- a/gcc/tree-ssa-loop-prefetch.c +++ b/gcc/tree-ssa-loop-prefetch.c @@ -1532,7 +1532,7 @@ determine_loop_nest_reuse (struct loop *loop, struct mem_ref_group *refs, static bool is_loop_prefetching_profitable (unsigned ahead, HOST_WIDE_INT est_niter, unsigned ninsns, unsigned prefetch_count, - unsigned mem_ref_count) + unsigned mem_ref_count, unsigned unroll_factor) { int insn_to_mem_ratio, insn_to_prefetch_ratio; @@ -1570,13 +1570,18 @@ is_loop_prefetching_profitable (unsigned ahead, HOST_WIDE_INT est_niter, by taking the ratio between the number of prefetches and the total number of instructions. Since we are using integer arithmetic, we compute the reciprocal of this ratio. - TODO: Account for loop unrolling, which may reduce the costs of - shorter stride prefetches. Note that not accounting for loop - unrolling over-estimates the cost and hence gives more conservative - results. */ + (unroll_factor * ninsns) is used to estimate the number of instructions in + the unrolled loop. This implementation is a bit simplistic -- the number + of issued prefetch instructions is also affected by unrolling. So, + prefetch_mod and the unroll factor should be taken into account when + determining prefetch_count. Also, the number of insns of the unrolled + loop will usually be significantly smaller than the number of insns of the + original loop * unroll_factor (at least the induction variable increases + and the exit branches will get eliminated), so it might be better to use + tree_estimate_loop_size + estimated_unrolled_size. */ if (est_niter < 0) { - insn_to_prefetch_ratio = ninsns / prefetch_count; + insn_to_prefetch_ratio = (unroll_factor * ninsns) / prefetch_count; return insn_to_prefetch_ratio >= MIN_INSN_TO_PREFETCH_RATIO; } @@ -1643,8 +1648,8 @@ loop_prefetch_arrays (struct loop *loop) ahead, unroll_factor, est_niter, ninsns, mem_ref_count, prefetch_count); - if (!is_loop_prefetching_profitable (ahead, est_niter, ninsns, - prefetch_count, mem_ref_count)) + if (!is_loop_prefetching_profitable (ahead, est_niter, ninsns, prefetch_count, + mem_ref_count, unroll_factor)) goto fail; mark_nontemporal_stores (loop, refs); |