re PR rtl-optimization/88751 (Performance regression reload vs lra)
Fix PR88751 This patch implements a small improvement for the heuristic in lra which decides when it has to activate the simpler register allocation algorithm. gcc/ChangeLog: 2019-06-06 Andreas Krebbel <krebbel@linux.ibm.com> PR rtl-optimization/88751 * ira.c (ira): Use the number of the actually referenced registers when calculating the threshold. From-SVN: r271996
This commit is contained in:
parent
a96ed9f7d9
commit
891f31f9a4
|
@ -1,3 +1,9 @@
|
||||||
|
2019-06-06 Andreas Krebbel <krebbel@linux.ibm.com>
|
||||||
|
|
||||||
|
PR rtl-optimization/88751
|
||||||
|
* ira.c (ira): Use the number of the actually referenced registers
|
||||||
|
when calculating the threshold.
|
||||||
|
|
||||||
2019-06-06 Jakub Jelinek <jakub@redhat.com>
|
2019-06-06 Jakub Jelinek <jakub@redhat.com>
|
||||||
|
|
||||||
* configure: Regenerate.
|
* configure: Regenerate.
|
||||||
|
|
|
@ -5198,6 +5198,8 @@ ira (FILE *f)
|
||||||
int ira_max_point_before_emit;
|
int ira_max_point_before_emit;
|
||||||
bool saved_flag_caller_saves = flag_caller_saves;
|
bool saved_flag_caller_saves = flag_caller_saves;
|
||||||
enum ira_region saved_flag_ira_region = flag_ira_region;
|
enum ira_region saved_flag_ira_region = flag_ira_region;
|
||||||
|
unsigned int i;
|
||||||
|
int num_used_regs = 0;
|
||||||
|
|
||||||
clear_bb_flags ();
|
clear_bb_flags ();
|
||||||
|
|
||||||
|
@ -5213,12 +5215,17 @@ ira (FILE *f)
|
||||||
|
|
||||||
ira_conflicts_p = optimize > 0;
|
ira_conflicts_p = optimize > 0;
|
||||||
|
|
||||||
|
/* Determine the number of pseudos actually requiring coloring. */
|
||||||
|
for (i = FIRST_PSEUDO_REGISTER; i < DF_REG_SIZE (df); i++)
|
||||||
|
num_used_regs += !!(DF_REG_USE_COUNT (i) + DF_REG_DEF_COUNT (i));
|
||||||
|
|
||||||
/* If there are too many pseudos and/or basic blocks (e.g. 10K
|
/* If there are too many pseudos and/or basic blocks (e.g. 10K
|
||||||
pseudos and 10K blocks or 100K pseudos and 1K blocks), we will
|
pseudos and 10K blocks or 100K pseudos and 1K blocks), we will
|
||||||
use simplified and faster algorithms in LRA. */
|
use simplified and faster algorithms in LRA. */
|
||||||
lra_simple_p
|
lra_simple_p
|
||||||
= (ira_use_lra_p
|
= (ira_use_lra_p
|
||||||
&& max_reg_num () >= (1 << 26) / last_basic_block_for_fn (cfun));
|
&& num_used_regs >= (1 << 26) / last_basic_block_for_fn (cfun));
|
||||||
|
|
||||||
if (lra_simple_p)
|
if (lra_simple_p)
|
||||||
{
|
{
|
||||||
/* It permits to skip live range splitting in LRA. */
|
/* It permits to skip live range splitting in LRA. */
|
||||||
|
|
Loading…
Reference in New Issue