diff options
| author | Hans-Peter Nilsson <hp@axis.se> | 1998-08-30 21:21:22 +0200 |
|---|---|---|
| committer | Jeff Law <law@gcc.gnu.org> | 1998-08-30 13:21:22 -0600 |
| commit | 9862dea9245ee7727f7c48ac542fa3972808f01c (patch) | |
| tree | 16c6d9614f1f8e05978455ca1935f5e5ec64150d /gcc | |
| parent | e916423b59f100866c969fc5df9099d19ee32e8c (diff) | |
| download | gcc-9862dea9245ee7727f7c48ac542fa3972808f01c.zip gcc-9862dea9245ee7727f7c48ac542fa3972808f01c.tar.gz gcc-9862dea9245ee7727f7c48ac542fa3972808f01c.tar.bz2 | |
expr.c (expand_expr): Change ">" to ">=" making MOVE_RATIO use consistent.
* expr.c (expand_expr): Change ">" to ">=" making MOVE_RATIO use
consistent.
* tm.texi (Costs): Say MOVE_RATIO is number of mem-mem move
*sequences* *below* which scalar moves will be used.
From-SVN: r22107
Diffstat (limited to 'gcc')
| -rw-r--r-- | gcc/ChangeLog | 7 | ||||
| -rw-r--r-- | gcc/expr.c | 2 | ||||
| -rw-r--r-- | gcc/tm.texi | 6 |
3 files changed, 13 insertions, 2 deletions
diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 159983f..7af6216 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,10 @@ +Sun Aug 30 20:19:43 1998 Hans-Peter Nilsson <hp@axis.se> + + * expr.c (expand_expr): Change ">" to ">=" making MOVE_RATIO use + consistent. + * tm.texi (Costs): Say MOVE_RATIO is number of mem-mem move + *sequences* *below* which scalar moves will be used. + Sun Aug 30 17:18:43 1998 Jeffrey A Law (law@cygnus.com) * collect2.c (mktemp): Delete unused declaration. @@ -5889,7 +5889,7 @@ expand_expr (exp, target, tmode, modifier) && (move_by_pieces_ninsns (TREE_INT_CST_LOW (TYPE_SIZE (type))/BITS_PER_UNIT, TYPE_ALIGN (type) / BITS_PER_UNIT) - > MOVE_RATIO) + >= MOVE_RATIO) && ! mostly_zeros_p (exp)))) || (modifier == EXPAND_INITIALIZER && TREE_CONSTANT (exp))) { diff --git a/gcc/tm.texi b/gcc/tm.texi index bf9aa4a..2f72176 100644 --- a/gcc/tm.texi +++ b/gcc/tm.texi @@ -4727,10 +4727,14 @@ than good.) @findex MOVE_RATIO @item MOVE_RATIO -The number of scalar move insns which should be generated instead of a +The threshold of number of scalar memory-to-memory move insns, @emph{below} +which a sequence of insns should be generated instead of a string move insn or a library call. Increasing the value will always make code faster, but eventually incurs high cost in increased code size. +Note that on machines with no memory-to-memory move insns, this macro denotes +the corresponding number of memory-to-memory @emph{sequences}. + If you don't define this, a reasonable default is used. @findex NO_FUNCTION_CSE |