* [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly
@ 2014-10-10 15:40 Evgeny Stupachenko
2014-10-13 8:09 ` Richard Biener
0 siblings, 1 reply; 23+ messages in thread
From: Evgeny Stupachenko @ 2014-10-10 15:40 UTC (permalink / raw)
To: Uros Bizjak, GCC Patches
Hi,
The patch increase PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with
high branch cost.
Bootstrap and make check are in progress.
The patch boosts (up to 2,5 times improve) several benchmarks compiled
with "-Ofast" on Silvermont
Spec2000:
+5% gain on 173.applu
+1% gain on 255.vortex
Is it ok for trunk when pass bootstrap and make check?
Thanks,
Evgeny
2014-10-10 Evgeny Stupachenko <evstupac@gmail.com>
* config/i386/i386.c (ix86_option_override_internal): Increase
PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with high branch cost.
* config/i386/i386.h (TARGET_HIGH_BRANCH_COST): New.
* config/i386/x86-tune.def (X86_TUNE_HIGH_BRANCH_COST): Indicates
CPUs with high branch cost.
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index 6337aa5..5ac10eb 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -4081,6 +4081,14 @@ ix86_option_override_internal (bool main_args_p,
opts->x_param_values,
opts_set->x_param_values);
+ /* Extend full peel max insns parameter for CPUs with high branch cost. */
+ if (TARGET_HIGH_BRANCH_COST)
+ maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS,
+ 120,
+ opts->x_param_values,
+ opts_set->x_param_values);
+
+
/* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */
if (opts->x_flag_prefetch_loop_arrays < 0
&& HAVE_prefetch
diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h
index 2c64162..da0c57b 100644
--- a/gcc/config/i386/i386.h
+++ b/gcc/config/i386/i386.h
@@ -415,6 +415,7 @@ extern unsigned char ix86_tune_features[X86_TUNE_LAST];
#define TARGET_INTER_UNIT_CONVERSIONS \
ix86_tune_features[X86_TUNE_INTER_UNIT_CONVERSIONS]
#define TARGET_FOUR_JUMP_LIMIT ix86_tune_features[X86_TUNE_FOUR_JUMP_LIMIT]
+#define TARGET_HIGH_BRANCH_COST
ix86_tune_features[X86_TUNE_HIGH_BRANCH_COST]
#define TARGET_SCHEDULE ix86_tune_features[X86_TUNE_SCHEDULE]
#define TARGET_USE_BT ix86_tune_features[X86_TUNE_USE_BT]
#define TARGET_USE_INCDEC ix86_tune_features[X86_TUNE_USE_INCDEC]
diff --git a/gcc/config/i386/x86-tune.def b/gcc/config/i386/x86-tune.def
index b6b210e..04d8bf8 100644
--- a/gcc/config/i386/x86-tune.def
+++ b/gcc/config/i386/x86-tune.def
@@ -208,6 +208,11 @@ DEF_TUNE (X86_TUNE_FOUR_JUMP_LIMIT, "four_jump_limit",
m_PPRO | m_P4_NOCONA | m_BONNELL | m_SILVERMONT | m_INTEL |
m_ATHLON_K8 | m_AMDFAM10)
+/* X86_TUNE_HIGH_BRANCH_COST: Some CPUs have higher branch cost. This could be
+ used to tune unroll, if-cvt, inline... heuristics. */
+DEF_TUNE (X86_TUNE_HIGH_BRANCH_COST, "high_branch_cost",
+ m_BONNELL | m_SILVERMONT | m_INTEL)
+
/*****************************************************************************/
/* Integer instruction selection tuning */
/*****************************************************************************/
^ permalink raw reply [flat|nested] 23+ messages in thread* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-10 15:40 [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly Evgeny Stupachenko @ 2014-10-13 8:09 ` Richard Biener 2014-10-13 11:12 ` Evgeny Stupachenko 2014-10-13 12:42 ` Jan Hubicka 0 siblings, 2 replies; 23+ messages in thread From: Richard Biener @ 2014-10-13 8:09 UTC (permalink / raw) To: Evgeny Stupachenko; +Cc: Uros Bizjak, GCC Patches On Fri, Oct 10, 2014 at 5:40 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > Hi, > > The patch increase PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with > high branch cost. > Bootstrap and make check are in progress. > The patch boosts (up to 2,5 times improve) several benchmarks compiled > with "-Ofast" on Silvermont > Spec2000: > +5% gain on 173.applu > +1% gain on 255.vortex > > Is it ok for trunk when pass bootstrap and make check? This is only a 20% increase - from 100 to 120. I would instead suggest to explore doing this change unconditionally if it helps that much. Richard. > Thanks, > Evgeny > > 2014-10-10 Evgeny Stupachenko <evstupac@gmail.com> > * config/i386/i386.c (ix86_option_override_internal): Increase > PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with high branch cost. > * config/i386/i386.h (TARGET_HIGH_BRANCH_COST): New. > * config/i386/x86-tune.def (X86_TUNE_HIGH_BRANCH_COST): Indicates > CPUs with high branch cost. > > diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c > index 6337aa5..5ac10eb 100644 > --- a/gcc/config/i386/i386.c > +++ b/gcc/config/i386/i386.c > @@ -4081,6 +4081,14 @@ ix86_option_override_internal (bool main_args_p, > opts->x_param_values, > opts_set->x_param_values); > > + /* Extend full peel max insns parameter for CPUs with high branch cost. */ > + if (TARGET_HIGH_BRANCH_COST) > + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, > + 120, > + opts->x_param_values, > + opts_set->x_param_values); > + > + > /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ > if (opts->x_flag_prefetch_loop_arrays < 0 > && HAVE_prefetch > diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h > index 2c64162..da0c57b 100644 > --- a/gcc/config/i386/i386.h > +++ b/gcc/config/i386/i386.h > @@ -415,6 +415,7 @@ extern unsigned char ix86_tune_features[X86_TUNE_LAST]; > #define TARGET_INTER_UNIT_CONVERSIONS \ > ix86_tune_features[X86_TUNE_INTER_UNIT_CONVERSIONS] > #define TARGET_FOUR_JUMP_LIMIT ix86_tune_features[X86_TUNE_FOUR_JUMP_LIMIT] > +#define TARGET_HIGH_BRANCH_COST > ix86_tune_features[X86_TUNE_HIGH_BRANCH_COST] > #define TARGET_SCHEDULE ix86_tune_features[X86_TUNE_SCHEDULE] > #define TARGET_USE_BT ix86_tune_features[X86_TUNE_USE_BT] > #define TARGET_USE_INCDEC ix86_tune_features[X86_TUNE_USE_INCDEC] > diff --git a/gcc/config/i386/x86-tune.def b/gcc/config/i386/x86-tune.def > index b6b210e..04d8bf8 100644 > --- a/gcc/config/i386/x86-tune.def > +++ b/gcc/config/i386/x86-tune.def > @@ -208,6 +208,11 @@ DEF_TUNE (X86_TUNE_FOUR_JUMP_LIMIT, "four_jump_limit", > m_PPRO | m_P4_NOCONA | m_BONNELL | m_SILVERMONT | m_INTEL | > m_ATHLON_K8 | m_AMDFAM10) > > +/* X86_TUNE_HIGH_BRANCH_COST: Some CPUs have higher branch cost. This could be > + used to tune unroll, if-cvt, inline... heuristics. */ > +DEF_TUNE (X86_TUNE_HIGH_BRANCH_COST, "high_branch_cost", > + m_BONNELL | m_SILVERMONT | m_INTEL) > + > /*****************************************************************************/ > /* Integer instruction selection tuning */ > /*****************************************************************************/ ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-13 8:09 ` Richard Biener @ 2014-10-13 11:12 ` Evgeny Stupachenko 2014-10-13 12:42 ` Jan Hubicka 1 sibling, 0 replies; 23+ messages in thread From: Evgeny Stupachenko @ 2014-10-13 11:12 UTC (permalink / raw) To: Richard Biener; +Cc: Uros Bizjak, GCC Patches I need to collect data from Haswell, but the patch should not help it's performance much, just increase code size. On Mon, Oct 13, 2014 at 12:01 PM, Richard Biener <richard.guenther@gmail.com> wrote: > On Fri, Oct 10, 2014 at 5:40 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: >> Hi, >> >> The patch increase PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with >> high branch cost. >> Bootstrap and make check are in progress. >> The patch boosts (up to 2,5 times improve) several benchmarks compiled >> with "-Ofast" on Silvermont >> Spec2000: >> +5% gain on 173.applu >> +1% gain on 255.vortex >> >> Is it ok for trunk when pass bootstrap and make check? > > This is only a 20% increase - from 100 to 120. I would instead suggest > to explore doing this change unconditionally if it helps that much. > > Richard. > >> Thanks, >> Evgeny >> >> 2014-10-10 Evgeny Stupachenko <evstupac@gmail.com> >> * config/i386/i386.c (ix86_option_override_internal): Increase >> PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with high branch cost. >> * config/i386/i386.h (TARGET_HIGH_BRANCH_COST): New. >> * config/i386/x86-tune.def (X86_TUNE_HIGH_BRANCH_COST): Indicates >> CPUs with high branch cost. >> >> diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c >> index 6337aa5..5ac10eb 100644 >> --- a/gcc/config/i386/i386.c >> +++ b/gcc/config/i386/i386.c >> @@ -4081,6 +4081,14 @@ ix86_option_override_internal (bool main_args_p, >> opts->x_param_values, >> opts_set->x_param_values); >> >> + /* Extend full peel max insns parameter for CPUs with high branch cost. */ >> + if (TARGET_HIGH_BRANCH_COST) >> + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, >> + 120, >> + opts->x_param_values, >> + opts_set->x_param_values); >> + >> + >> /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ >> if (opts->x_flag_prefetch_loop_arrays < 0 >> && HAVE_prefetch >> diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h >> index 2c64162..da0c57b 100644 >> --- a/gcc/config/i386/i386.h >> +++ b/gcc/config/i386/i386.h >> @@ -415,6 +415,7 @@ extern unsigned char ix86_tune_features[X86_TUNE_LAST]; >> #define TARGET_INTER_UNIT_CONVERSIONS \ >> ix86_tune_features[X86_TUNE_INTER_UNIT_CONVERSIONS] >> #define TARGET_FOUR_JUMP_LIMIT ix86_tune_features[X86_TUNE_FOUR_JUMP_LIMIT] >> +#define TARGET_HIGH_BRANCH_COST >> ix86_tune_features[X86_TUNE_HIGH_BRANCH_COST] >> #define TARGET_SCHEDULE ix86_tune_features[X86_TUNE_SCHEDULE] >> #define TARGET_USE_BT ix86_tune_features[X86_TUNE_USE_BT] >> #define TARGET_USE_INCDEC ix86_tune_features[X86_TUNE_USE_INCDEC] >> diff --git a/gcc/config/i386/x86-tune.def b/gcc/config/i386/x86-tune.def >> index b6b210e..04d8bf8 100644 >> --- a/gcc/config/i386/x86-tune.def >> +++ b/gcc/config/i386/x86-tune.def >> @@ -208,6 +208,11 @@ DEF_TUNE (X86_TUNE_FOUR_JUMP_LIMIT, "four_jump_limit", >> m_PPRO | m_P4_NOCONA | m_BONNELL | m_SILVERMONT | m_INTEL | >> m_ATHLON_K8 | m_AMDFAM10) >> >> +/* X86_TUNE_HIGH_BRANCH_COST: Some CPUs have higher branch cost. This could be >> + used to tune unroll, if-cvt, inline... heuristics. */ >> +DEF_TUNE (X86_TUNE_HIGH_BRANCH_COST, "high_branch_cost", >> + m_BONNELL | m_SILVERMONT | m_INTEL) >> + >> /*****************************************************************************/ >> /* Integer instruction selection tuning */ >> /*****************************************************************************/ ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-13 8:09 ` Richard Biener 2014-10-13 11:12 ` Evgeny Stupachenko @ 2014-10-13 12:42 ` Jan Hubicka 2014-10-27 8:23 ` Evgeny Stupachenko 2014-10-31 13:03 ` Eric Botcazou 1 sibling, 2 replies; 23+ messages in thread From: Jan Hubicka @ 2014-10-13 12:42 UTC (permalink / raw) To: Richard Biener; +Cc: Evgeny Stupachenko, Uros Bizjak, GCC Patches > On Fri, Oct 10, 2014 at 5:40 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > > Hi, > > > > The patch increase PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with > > high branch cost. > > Bootstrap and make check are in progress. > > The patch boosts (up to 2,5 times improve) several benchmarks compiled > > with "-Ofast" on Silvermont > > Spec2000: > > +5% gain on 173.applu > > +1% gain on 255.vortex > > > > Is it ok for trunk when pass bootstrap and make check? > > This is only a 20% increase - from 100 to 120. I would instead suggest > to explore doing this change unconditionally if it helps that much. Agreed, I think the value of 100 was set decade ago by Zdenek and me completely artifically. I do not recall any serious tuning of this flag. Note that I plan to update https://gcc.gnu.org/ml/gcc-patches/2013-11/msg02270.html to current tree so PARAM_MAX_COMPLETELY_PEELED_INSNS will be used at gimple level rather than tree changing its meaning somewhat. Perhaps I could try to find time this or next week to update the patch so we do not need to do the tuning twice. Honza > > Richard. > > > Thanks, > > Evgeny > > > > 2014-10-10 Evgeny Stupachenko <evstupac@gmail.com> > > * config/i386/i386.c (ix86_option_override_internal): Increase > > PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with high branch cost. > > * config/i386/i386.h (TARGET_HIGH_BRANCH_COST): New. > > * config/i386/x86-tune.def (X86_TUNE_HIGH_BRANCH_COST): Indicates > > CPUs with high branch cost. > > > > diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c > > index 6337aa5..5ac10eb 100644 > > --- a/gcc/config/i386/i386.c > > +++ b/gcc/config/i386/i386.c > > @@ -4081,6 +4081,14 @@ ix86_option_override_internal (bool main_args_p, > > opts->x_param_values, > > opts_set->x_param_values); > > > > + /* Extend full peel max insns parameter for CPUs with high branch cost. */ > > + if (TARGET_HIGH_BRANCH_COST) > > + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, > > + 120, > > + opts->x_param_values, > > + opts_set->x_param_values); > > + > > + > > /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ > > if (opts->x_flag_prefetch_loop_arrays < 0 > > && HAVE_prefetch > > diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h > > index 2c64162..da0c57b 100644 > > --- a/gcc/config/i386/i386.h > > +++ b/gcc/config/i386/i386.h > > @@ -415,6 +415,7 @@ extern unsigned char ix86_tune_features[X86_TUNE_LAST]; > > #define TARGET_INTER_UNIT_CONVERSIONS \ > > ix86_tune_features[X86_TUNE_INTER_UNIT_CONVERSIONS] > > #define TARGET_FOUR_JUMP_LIMIT ix86_tune_features[X86_TUNE_FOUR_JUMP_LIMIT] > > +#define TARGET_HIGH_BRANCH_COST > > ix86_tune_features[X86_TUNE_HIGH_BRANCH_COST] > > #define TARGET_SCHEDULE ix86_tune_features[X86_TUNE_SCHEDULE] > > #define TARGET_USE_BT ix86_tune_features[X86_TUNE_USE_BT] > > #define TARGET_USE_INCDEC ix86_tune_features[X86_TUNE_USE_INCDEC] > > diff --git a/gcc/config/i386/x86-tune.def b/gcc/config/i386/x86-tune.def > > index b6b210e..04d8bf8 100644 > > --- a/gcc/config/i386/x86-tune.def > > +++ b/gcc/config/i386/x86-tune.def > > @@ -208,6 +208,11 @@ DEF_TUNE (X86_TUNE_FOUR_JUMP_LIMIT, "four_jump_limit", > > m_PPRO | m_P4_NOCONA | m_BONNELL | m_SILVERMONT | m_INTEL | > > m_ATHLON_K8 | m_AMDFAM10) > > > > +/* X86_TUNE_HIGH_BRANCH_COST: Some CPUs have higher branch cost. This could be > > + used to tune unroll, if-cvt, inline... heuristics. */ > > +DEF_TUNE (X86_TUNE_HIGH_BRANCH_COST, "high_branch_cost", > > + m_BONNELL | m_SILVERMONT | m_INTEL) > > + > > /*****************************************************************************/ > > /* Integer instruction selection tuning */ > > /*****************************************************************************/ ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-13 12:42 ` Jan Hubicka @ 2014-10-27 8:23 ` Evgeny Stupachenko 2014-10-28 12:22 ` Evgeny Stupachenko 2014-10-31 13:03 ` Eric Botcazou 1 sibling, 1 reply; 23+ messages in thread From: Evgeny Stupachenko @ 2014-10-27 8:23 UTC (permalink / raw) To: Jan Hubicka; +Cc: Richard Biener, Uros Bizjak, GCC Patches The results are the same for Silvermont. There are no significant changes on Haswell. So I agree with Richard, let's enable this x86 wide. Bootstrap/ passed. Make check in progress. Is it ok? 2014-10-25 Evgeny Stupachenko <evstupac@gmail.com> * config/i386/i386.c (ix86_option_override_internal): Increase PARAM_MAX_COMPLETELY_PEELED_INSNS. diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c index 6337aa5..5ac10eb 100644 --- a/gcc/config/i386/i386.c +++ b/gcc/config/i386/i386.c @@ -4081,6 +4081,12 @@ ix86_option_override_internal (bool main_args_p, opts->x_param_values, opts_set->x_param_values); + /* Extend full peel max insns parameter for x86. */ + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, + 120, + opts->x_param_values, + opts_set->x_param_values); + /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ if (opts->x_flag_prefetch_loop_arrays < 0 && HAVE_prefetch On Mon, Oct 13, 2014 at 4:23 PM, Jan Hubicka <hubicka@ucw.cz> wrote: >> On Fri, Oct 10, 2014 at 5:40 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: >> > Hi, >> > >> > The patch increase PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with >> > high branch cost. >> > Bootstrap and make check are in progress. >> > The patch boosts (up to 2,5 times improve) several benchmarks compiled >> > with "-Ofast" on Silvermont >> > Spec2000: >> > +5% gain on 173.applu >> > +1% gain on 255.vortex >> > >> > Is it ok for trunk when pass bootstrap and make check? >> >> This is only a 20% increase - from 100 to 120. I would instead suggest >> to explore doing this change unconditionally if it helps that much. > > Agreed, I think the value of 100 was set decade ago by Zdenek and me completely > artifically. I do not recall any serious tuning of this flag. > > Note that I plan to update > https://gcc.gnu.org/ml/gcc-patches/2013-11/msg02270.html to current tree so > PARAM_MAX_COMPLETELY_PEELED_INSNS will be used at gimple level rather than tree > changing its meaning somewhat. > > Perhaps I could try to find time this or next week to update the patch so we do > not need to do the tuning twice. > > Honza > >> >> Richard. >> >> > Thanks, >> > Evgeny >> > >> > 2014-10-10 Evgeny Stupachenko <evstupac@gmail.com> >> > * config/i386/i386.c (ix86_option_override_internal): Increase >> > PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with high branch cost. >> > * config/i386/i386.h (TARGET_HIGH_BRANCH_COST): New. >> > * config/i386/x86-tune.def (X86_TUNE_HIGH_BRANCH_COST): Indicates >> > CPUs with high branch cost. >> > >> > diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c >> > index 6337aa5..5ac10eb 100644 >> > --- a/gcc/config/i386/i386.c >> > +++ b/gcc/config/i386/i386.c >> > @@ -4081,6 +4081,14 @@ ix86_option_override_internal (bool main_args_p, >> > opts->x_param_values, >> > opts_set->x_param_values); >> > >> > + /* Extend full peel max insns parameter for CPUs with high branch cost. */ >> > + if (TARGET_HIGH_BRANCH_COST) >> > + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, >> > + 120, >> > + opts->x_param_values, >> > + opts_set->x_param_values); >> > + >> > + >> > /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ >> > if (opts->x_flag_prefetch_loop_arrays < 0 >> > && HAVE_prefetch >> > diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h >> > index 2c64162..da0c57b 100644 >> > --- a/gcc/config/i386/i386.h >> > +++ b/gcc/config/i386/i386.h >> > @@ -415,6 +415,7 @@ extern unsigned char ix86_tune_features[X86_TUNE_LAST]; >> > #define TARGET_INTER_UNIT_CONVERSIONS \ >> > ix86_tune_features[X86_TUNE_INTER_UNIT_CONVERSIONS] >> > #define TARGET_FOUR_JUMP_LIMIT ix86_tune_features[X86_TUNE_FOUR_JUMP_LIMIT] >> > +#define TARGET_HIGH_BRANCH_COST >> > ix86_tune_features[X86_TUNE_HIGH_BRANCH_COST] >> > #define TARGET_SCHEDULE ix86_tune_features[X86_TUNE_SCHEDULE] >> > #define TARGET_USE_BT ix86_tune_features[X86_TUNE_USE_BT] >> > #define TARGET_USE_INCDEC ix86_tune_features[X86_TUNE_USE_INCDEC] >> > diff --git a/gcc/config/i386/x86-tune.def b/gcc/config/i386/x86-tune.def >> > index b6b210e..04d8bf8 100644 >> > --- a/gcc/config/i386/x86-tune.def >> > +++ b/gcc/config/i386/x86-tune.def >> > @@ -208,6 +208,11 @@ DEF_TUNE (X86_TUNE_FOUR_JUMP_LIMIT, "four_jump_limit", >> > m_PPRO | m_P4_NOCONA | m_BONNELL | m_SILVERMONT | m_INTEL | >> > m_ATHLON_K8 | m_AMDFAM10) >> > >> > +/* X86_TUNE_HIGH_BRANCH_COST: Some CPUs have higher branch cost. This could be >> > + used to tune unroll, if-cvt, inline... heuristics. */ >> > +DEF_TUNE (X86_TUNE_HIGH_BRANCH_COST, "high_branch_cost", >> > + m_BONNELL | m_SILVERMONT | m_INTEL) >> > + >> > /*****************************************************************************/ >> > /* Integer instruction selection tuning */ >> > /*****************************************************************************/ On Mon, Oct 13, 2014 at 3:23 PM, Jan Hubicka <hubicka@ucw.cz> wrote: >> On Fri, Oct 10, 2014 at 5:40 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: >> > Hi, >> > >> > The patch increase PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with >> > high branch cost. >> > Bootstrap and make check are in progress. >> > The patch boosts (up to 2,5 times improve) several benchmarks compiled >> > with "-Ofast" on Silvermont >> > Spec2000: >> > +5% gain on 173.applu >> > +1% gain on 255.vortex >> > >> > Is it ok for trunk when pass bootstrap and make check? >> >> This is only a 20% increase - from 100 to 120. I would instead suggest >> to explore doing this change unconditionally if it helps that much. > > Agreed, I think the value of 100 was set decade ago by Zdenek and me completely > artifically. I do not recall any serious tuning of this flag. > > Note that I plan to update > https://gcc.gnu.org/ml/gcc-patches/2013-11/msg02270.html to current tree so > PARAM_MAX_COMPLETELY_PEELED_INSNS will be used at gimple level rather than tree > changing its meaning somewhat. > > Perhaps I could try to find time this or next week to update the patch so we do > not need to do the tuning twice. > > Honza > >> >> Richard. >> >> > Thanks, >> > Evgeny >> > >> > 2014-10-10 Evgeny Stupachenko <evstupac@gmail.com> >> > * config/i386/i386.c (ix86_option_override_internal): Increase >> > PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with high branch cost. >> > * config/i386/i386.h (TARGET_HIGH_BRANCH_COST): New. >> > * config/i386/x86-tune.def (X86_TUNE_HIGH_BRANCH_COST): Indicates >> > CPUs with high branch cost. >> > >> > diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c >> > index 6337aa5..5ac10eb 100644 >> > --- a/gcc/config/i386/i386.c >> > +++ b/gcc/config/i386/i386.c >> > @@ -4081,6 +4081,14 @@ ix86_option_override_internal (bool main_args_p, >> > opts->x_param_values, >> > opts_set->x_param_values); >> > >> > + /* Extend full peel max insns parameter for CPUs with high branch cost. */ >> > + if (TARGET_HIGH_BRANCH_COST) >> > + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, >> > + 120, >> > + opts->x_param_values, >> > + opts_set->x_param_values); >> > + >> > + >> > /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ >> > if (opts->x_flag_prefetch_loop_arrays < 0 >> > && HAVE_prefetch >> > diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h >> > index 2c64162..da0c57b 100644 >> > --- a/gcc/config/i386/i386.h >> > +++ b/gcc/config/i386/i386.h >> > @@ -415,6 +415,7 @@ extern unsigned char ix86_tune_features[X86_TUNE_LAST]; >> > #define TARGET_INTER_UNIT_CONVERSIONS \ >> > ix86_tune_features[X86_TUNE_INTER_UNIT_CONVERSIONS] >> > #define TARGET_FOUR_JUMP_LIMIT ix86_tune_features[X86_TUNE_FOUR_JUMP_LIMIT] >> > +#define TARGET_HIGH_BRANCH_COST >> > ix86_tune_features[X86_TUNE_HIGH_BRANCH_COST] >> > #define TARGET_SCHEDULE ix86_tune_features[X86_TUNE_SCHEDULE] >> > #define TARGET_USE_BT ix86_tune_features[X86_TUNE_USE_BT] >> > #define TARGET_USE_INCDEC ix86_tune_features[X86_TUNE_USE_INCDEC] >> > diff --git a/gcc/config/i386/x86-tune.def b/gcc/config/i386/x86-tune.def >> > index b6b210e..04d8bf8 100644 >> > --- a/gcc/config/i386/x86-tune.def >> > +++ b/gcc/config/i386/x86-tune.def >> > @@ -208,6 +208,11 @@ DEF_TUNE (X86_TUNE_FOUR_JUMP_LIMIT, "four_jump_limit", >> > m_PPRO | m_P4_NOCONA | m_BONNELL | m_SILVERMONT | m_INTEL | >> > m_ATHLON_K8 | m_AMDFAM10) >> > >> > +/* X86_TUNE_HIGH_BRANCH_COST: Some CPUs have higher branch cost. This could be >> > + used to tune unroll, if-cvt, inline... heuristics. */ >> > +DEF_TUNE (X86_TUNE_HIGH_BRANCH_COST, "high_branch_cost", >> > + m_BONNELL | m_SILVERMONT | m_INTEL) >> > + >> > /*****************************************************************************/ >> > /* Integer instruction selection tuning */ >> > /*****************************************************************************/ ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-27 8:23 ` Evgeny Stupachenko @ 2014-10-28 12:22 ` Evgeny Stupachenko 2014-10-30 8:32 ` Uros Bizjak 0 siblings, 1 reply; 23+ messages in thread From: Evgeny Stupachenko @ 2014-10-28 12:22 UTC (permalink / raw) To: Jan Hubicka; +Cc: Richard Biener, Uros Bizjak, GCC Patches make check for gcc passed On Mon, Oct 27, 2014 at 11:10 AM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > The results are the same for Silvermont. > There are no significant changes on Haswell. > So I agree with Richard, let's enable this x86 wide. > > Bootstrap/ passed. > Make check in progress. > Is it ok? > > 2014-10-25 Evgeny Stupachenko <evstupac@gmail.com> > * config/i386/i386.c (ix86_option_override_internal): Increase > PARAM_MAX_COMPLETELY_PEELED_INSNS. > > diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c > index 6337aa5..5ac10eb 100644 > --- a/gcc/config/i386/i386.c > +++ b/gcc/config/i386/i386.c > @@ -4081,6 +4081,12 @@ ix86_option_override_internal (bool main_args_p, > opts->x_param_values, > opts_set->x_param_values); > > + /* Extend full peel max insns parameter for x86. */ > + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, > + 120, > + opts->x_param_values, > + opts_set->x_param_values); > + > /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ > if (opts->x_flag_prefetch_loop_arrays < 0 > && HAVE_prefetch > > On Mon, Oct 13, 2014 at 4:23 PM, Jan Hubicka <hubicka@ucw.cz> wrote: >>> On Fri, Oct 10, 2014 at 5:40 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: >>> > Hi, >>> > >>> > The patch increase PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with >>> > high branch cost. >>> > Bootstrap and make check are in progress. >>> > The patch boosts (up to 2,5 times improve) several benchmarks compiled >>> > with "-Ofast" on Silvermont >>> > Spec2000: >>> > +5% gain on 173.applu >>> > +1% gain on 255.vortex >>> > >>> > Is it ok for trunk when pass bootstrap and make check? >>> >>> This is only a 20% increase - from 100 to 120. I would instead suggest >>> to explore doing this change unconditionally if it helps that much. >> >> Agreed, I think the value of 100 was set decade ago by Zdenek and me completely >> artifically. I do not recall any serious tuning of this flag. >> >> Note that I plan to update >> https://gcc.gnu.org/ml/gcc-patches/2013-11/msg02270.html to current tree so >> PARAM_MAX_COMPLETELY_PEELED_INSNS will be used at gimple level rather than tree >> changing its meaning somewhat. >> >> Perhaps I could try to find time this or next week to update the patch so we do >> not need to do the tuning twice. >> >> Honza >> >>> >>> Richard. >>> >>> > Thanks, >>> > Evgeny >>> > >>> > 2014-10-10 Evgeny Stupachenko <evstupac@gmail.com> >>> > * config/i386/i386.c (ix86_option_override_internal): Increase >>> > PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with high branch cost. >>> > * config/i386/i386.h (TARGET_HIGH_BRANCH_COST): New. >>> > * config/i386/x86-tune.def (X86_TUNE_HIGH_BRANCH_COST): Indicates >>> > CPUs with high branch cost. >>> > >>> > diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c >>> > index 6337aa5..5ac10eb 100644 >>> > --- a/gcc/config/i386/i386.c >>> > +++ b/gcc/config/i386/i386.c >>> > @@ -4081,6 +4081,14 @@ ix86_option_override_internal (bool main_args_p, >>> > opts->x_param_values, >>> > opts_set->x_param_values); >>> > >>> > + /* Extend full peel max insns parameter for CPUs with high branch cost. */ >>> > + if (TARGET_HIGH_BRANCH_COST) >>> > + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, >>> > + 120, >>> > + opts->x_param_values, >>> > + opts_set->x_param_values); >>> > + >>> > + >>> > /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ >>> > if (opts->x_flag_prefetch_loop_arrays < 0 >>> > && HAVE_prefetch >>> > diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h >>> > index 2c64162..da0c57b 100644 >>> > --- a/gcc/config/i386/i386.h >>> > +++ b/gcc/config/i386/i386.h >>> > @@ -415,6 +415,7 @@ extern unsigned char ix86_tune_features[X86_TUNE_LAST]; >>> > #define TARGET_INTER_UNIT_CONVERSIONS \ >>> > ix86_tune_features[X86_TUNE_INTER_UNIT_CONVERSIONS] >>> > #define TARGET_FOUR_JUMP_LIMIT ix86_tune_features[X86_TUNE_FOUR_JUMP_LIMIT] >>> > +#define TARGET_HIGH_BRANCH_COST >>> > ix86_tune_features[X86_TUNE_HIGH_BRANCH_COST] >>> > #define TARGET_SCHEDULE ix86_tune_features[X86_TUNE_SCHEDULE] >>> > #define TARGET_USE_BT ix86_tune_features[X86_TUNE_USE_BT] >>> > #define TARGET_USE_INCDEC ix86_tune_features[X86_TUNE_USE_INCDEC] >>> > diff --git a/gcc/config/i386/x86-tune.def b/gcc/config/i386/x86-tune.def >>> > index b6b210e..04d8bf8 100644 >>> > --- a/gcc/config/i386/x86-tune.def >>> > +++ b/gcc/config/i386/x86-tune.def >>> > @@ -208,6 +208,11 @@ DEF_TUNE (X86_TUNE_FOUR_JUMP_LIMIT, "four_jump_limit", >>> > m_PPRO | m_P4_NOCONA | m_BONNELL | m_SILVERMONT | m_INTEL | >>> > m_ATHLON_K8 | m_AMDFAM10) >>> > >>> > +/* X86_TUNE_HIGH_BRANCH_COST: Some CPUs have higher branch cost. This could be >>> > + used to tune unroll, if-cvt, inline... heuristics. */ >>> > +DEF_TUNE (X86_TUNE_HIGH_BRANCH_COST, "high_branch_cost", >>> > + m_BONNELL | m_SILVERMONT | m_INTEL) >>> > + >>> > /*****************************************************************************/ >>> > /* Integer instruction selection tuning */ >>> > /*****************************************************************************/ > > > On Mon, Oct 13, 2014 at 3:23 PM, Jan Hubicka <hubicka@ucw.cz> wrote: >>> On Fri, Oct 10, 2014 at 5:40 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: >>> > Hi, >>> > >>> > The patch increase PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with >>> > high branch cost. >>> > Bootstrap and make check are in progress. >>> > The patch boosts (up to 2,5 times improve) several benchmarks compiled >>> > with "-Ofast" on Silvermont >>> > Spec2000: >>> > +5% gain on 173.applu >>> > +1% gain on 255.vortex >>> > >>> > Is it ok for trunk when pass bootstrap and make check? >>> >>> This is only a 20% increase - from 100 to 120. I would instead suggest >>> to explore doing this change unconditionally if it helps that much. >> >> Agreed, I think the value of 100 was set decade ago by Zdenek and me completely >> artifically. I do not recall any serious tuning of this flag. >> >> Note that I plan to update >> https://gcc.gnu.org/ml/gcc-patches/2013-11/msg02270.html to current tree so >> PARAM_MAX_COMPLETELY_PEELED_INSNS will be used at gimple level rather than tree >> changing its meaning somewhat. >> >> Perhaps I could try to find time this or next week to update the patch so we do >> not need to do the tuning twice. >> >> Honza >> >>> >>> Richard. >>> >>> > Thanks, >>> > Evgeny >>> > >>> > 2014-10-10 Evgeny Stupachenko <evstupac@gmail.com> >>> > * config/i386/i386.c (ix86_option_override_internal): Increase >>> > PARAM_MAX_COMPLETELY_PEELED_INSNS for CPUs with high branch cost. >>> > * config/i386/i386.h (TARGET_HIGH_BRANCH_COST): New. >>> > * config/i386/x86-tune.def (X86_TUNE_HIGH_BRANCH_COST): Indicates >>> > CPUs with high branch cost. >>> > >>> > diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c >>> > index 6337aa5..5ac10eb 100644 >>> > --- a/gcc/config/i386/i386.c >>> > +++ b/gcc/config/i386/i386.c >>> > @@ -4081,6 +4081,14 @@ ix86_option_override_internal (bool main_args_p, >>> > opts->x_param_values, >>> > opts_set->x_param_values); >>> > >>> > + /* Extend full peel max insns parameter for CPUs with high branch cost. */ >>> > + if (TARGET_HIGH_BRANCH_COST) >>> > + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, >>> > + 120, >>> > + opts->x_param_values, >>> > + opts_set->x_param_values); >>> > + >>> > + >>> > /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ >>> > if (opts->x_flag_prefetch_loop_arrays < 0 >>> > && HAVE_prefetch >>> > diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h >>> > index 2c64162..da0c57b 100644 >>> > --- a/gcc/config/i386/i386.h >>> > +++ b/gcc/config/i386/i386.h >>> > @@ -415,6 +415,7 @@ extern unsigned char ix86_tune_features[X86_TUNE_LAST]; >>> > #define TARGET_INTER_UNIT_CONVERSIONS \ >>> > ix86_tune_features[X86_TUNE_INTER_UNIT_CONVERSIONS] >>> > #define TARGET_FOUR_JUMP_LIMIT ix86_tune_features[X86_TUNE_FOUR_JUMP_LIMIT] >>> > +#define TARGET_HIGH_BRANCH_COST >>> > ix86_tune_features[X86_TUNE_HIGH_BRANCH_COST] >>> > #define TARGET_SCHEDULE ix86_tune_features[X86_TUNE_SCHEDULE] >>> > #define TARGET_USE_BT ix86_tune_features[X86_TUNE_USE_BT] >>> > #define TARGET_USE_INCDEC ix86_tune_features[X86_TUNE_USE_INCDEC] >>> > diff --git a/gcc/config/i386/x86-tune.def b/gcc/config/i386/x86-tune.def >>> > index b6b210e..04d8bf8 100644 >>> > --- a/gcc/config/i386/x86-tune.def >>> > +++ b/gcc/config/i386/x86-tune.def >>> > @@ -208,6 +208,11 @@ DEF_TUNE (X86_TUNE_FOUR_JUMP_LIMIT, "four_jump_limit", >>> > m_PPRO | m_P4_NOCONA | m_BONNELL | m_SILVERMONT | m_INTEL | >>> > m_ATHLON_K8 | m_AMDFAM10) >>> > >>> > +/* X86_TUNE_HIGH_BRANCH_COST: Some CPUs have higher branch cost. This could be >>> > + used to tune unroll, if-cvt, inline... heuristics. */ >>> > +DEF_TUNE (X86_TUNE_HIGH_BRANCH_COST, "high_branch_cost", >>> > + m_BONNELL | m_SILVERMONT | m_INTEL) >>> > + >>> > /*****************************************************************************/ >>> > /* Integer instruction selection tuning */ >>> > /*****************************************************************************/ ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-28 12:22 ` Evgeny Stupachenko @ 2014-10-30 8:32 ` Uros Bizjak 2014-10-30 17:54 ` Jan Hubicka 0 siblings, 1 reply; 23+ messages in thread From: Uros Bizjak @ 2014-10-30 8:32 UTC (permalink / raw) To: Evgeny Stupachenko; +Cc: Jan Hubicka, Richard Biener, GCC Patches On Tue, Oct 28, 2014 at 1:07 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > make check for gcc passed > > On Mon, Oct 27, 2014 at 11:10 AM, Evgeny Stupachenko <evstupac@gmail.com> wrote: >> The results are the same for Silvermont. >> There are no significant changes on Haswell. >> So I agree with Richard, let's enable this x86 wide. >> >> Bootstrap/ passed. >> Make check in progress. >> Is it ok? >> >> 2014-10-25 Evgeny Stupachenko <evstupac@gmail.com> >> * config/i386/i386.c (ix86_option_override_internal): Increase >> PARAM_MAX_COMPLETELY_PEELED_INSNS. Let's wait for Honza's approval ... Uros. ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-30 8:32 ` Uros Bizjak @ 2014-10-30 17:54 ` Jan Hubicka 2014-10-30 18:48 ` Evgeny Stupachenko 0 siblings, 1 reply; 23+ messages in thread From: Jan Hubicka @ 2014-10-30 17:54 UTC (permalink / raw) To: Uros Bizjak; +Cc: Evgeny Stupachenko, Jan Hubicka, Richard Biener, GCC Patches > On Tue, Oct 28, 2014 at 1:07 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > > make check for gcc passed > > > > On Mon, Oct 27, 2014 at 11:10 AM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > >> The results are the same for Silvermont. > >> There are no significant changes on Haswell. > >> So I agree with Richard, let's enable this x86 wide. > >> > >> Bootstrap/ passed. > >> Make check in progress. > >> Is it ok? > >> > >> 2014-10-25 Evgeny Stupachenko <evstupac@gmail.com> > >> * config/i386/i386.c (ix86_option_override_internal): Increase > >> PARAM_MAX_COMPLETELY_PEELED_INSNS. > > Let's wait for Honza's approval ... Looking through the emails, it is not clear to me if you re-tested that this still makes the intended speedup with the tree-level loop peeling? (comitted 2014-10-14). If it still works as intended, I do not think we have any reason to not change the default in params.def given that even ARM folks are calling for peeling by default. Honza > > Uros. ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-30 17:54 ` Jan Hubicka @ 2014-10-30 18:48 ` Evgeny Stupachenko 0 siblings, 0 replies; 23+ messages in thread From: Evgeny Stupachenko @ 2014-10-30 18:48 UTC (permalink / raw) To: Jan Hubicka; +Cc: Uros Bizjak, Richard Biener, GCC Patches Yes the speed up is the same. However I'm testing only x86 performance. Potentially we can somehow hurt ARM or others performance. GCC already has the tuning enabled for rs6000,s390, spu. Evgeny On Thu, Oct 30, 2014 at 8:27 PM, Jan Hubicka <hubicka@ucw.cz> wrote: >> On Tue, Oct 28, 2014 at 1:07 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: >> > make check for gcc passed >> > >> > On Mon, Oct 27, 2014 at 11:10 AM, Evgeny Stupachenko <evstupac@gmail.com> wrote: >> >> The results are the same for Silvermont. >> >> There are no significant changes on Haswell. >> >> So I agree with Richard, let's enable this x86 wide. >> >> >> >> Bootstrap/ passed. >> >> Make check in progress. >> >> Is it ok? >> >> >> >> 2014-10-25 Evgeny Stupachenko <evstupac@gmail.com> >> >> * config/i386/i386.c (ix86_option_override_internal): Increase >> >> PARAM_MAX_COMPLETELY_PEELED_INSNS. >> >> Let's wait for Honza's approval ... > > Looking through the emails, it is not clear to me if you re-tested that this still > makes the intended speedup with the tree-level loop peeling? (comitted 2014-10-14). > If it still works as intended, I do not think we have any reason to not change the > default in params.def given that even ARM folks are calling for peeling by default. > > Honza >> >> Uros. ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-13 12:42 ` Jan Hubicka 2014-10-27 8:23 ` Evgeny Stupachenko @ 2014-10-31 13:03 ` Eric Botcazou 2014-10-31 17:07 ` Evgeny Stupachenko 1 sibling, 1 reply; 23+ messages in thread From: Eric Botcazou @ 2014-10-31 13:03 UTC (permalink / raw) To: Jan Hubicka; +Cc: gcc-patches, Richard Biener, Evgeny Stupachenko, Uros Bizjak > Agreed, I think the value of 100 was set decade ago by Zdenek and me > completely artifically. I do not recall any serious tuning of this flag. Are you talking bout PARAM_MAX_COMPLETELY_PEELED_INSNS here? If so, see: https://gcc.gnu.org/ml/gcc-patches/2012-11/msg01193.html We have experienced performance regressions because of this arbitrary change and bumped it back to 200 unconditionally. -- Eric Botcazou ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-31 13:03 ` Eric Botcazou @ 2014-10-31 17:07 ` Evgeny Stupachenko 2014-11-07 15:37 ` Evgeny Stupachenko 0 siblings, 1 reply; 23+ messages in thread From: Evgeny Stupachenko @ 2014-10-31 17:07 UTC (permalink / raw) To: Eric Botcazou; +Cc: Jan Hubicka, GCC Patches, Richard Biener, Uros Bizjak I've measured spec2000, spec2006 as well and EEMBC for Silvermont in addition. 100->120 change gives gain for Silvermont, the results on Haswell are flat. On Fri, Oct 31, 2014 at 3:14 PM, Eric Botcazou <ebotcazou@adacore.com> wrote: >> Agreed, I think the value of 100 was set decade ago by Zdenek and me >> completely artifically. I do not recall any serious tuning of this flag. > > Are you talking bout PARAM_MAX_COMPLETELY_PEELED_INSNS here? If so, see: > https://gcc.gnu.org/ml/gcc-patches/2012-11/msg01193.html > > We have experienced performance regressions because of this arbitrary change > and bumped it back to 200 unconditionally. > > -- > Eric Botcazou ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-10-31 17:07 ` Evgeny Stupachenko @ 2014-11-07 15:37 ` Evgeny Stupachenko 2014-11-11 15:31 ` Evgeny Stupachenko 0 siblings, 1 reply; 23+ messages in thread From: Evgeny Stupachenko @ 2014-11-07 15:37 UTC (permalink / raw) To: Eric Botcazou; +Cc: Jan Hubicka, GCC Patches, Richard Biener, Uros Bizjak So are there any objections to enable this (PARAM_MAX_COMPLETELY_PEELED_INSNS increase from 100 to 120) for x86? On Fri, Oct 31, 2014 at 7:52 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > I've measured spec2000, spec2006 as well and EEMBC for Silvermont in addition. > 100->120 change gives gain for Silvermont, the results on Haswell are flat. > > On Fri, Oct 31, 2014 at 3:14 PM, Eric Botcazou <ebotcazou@adacore.com> wrote: >>> Agreed, I think the value of 100 was set decade ago by Zdenek and me >>> completely artifically. I do not recall any serious tuning of this flag. >> >> Are you talking bout PARAM_MAX_COMPLETELY_PEELED_INSNS here? If so, see: >> https://gcc.gnu.org/ml/gcc-patches/2012-11/msg01193.html >> >> We have experienced performance regressions because of this arbitrary change >> and bumped it back to 200 unconditionally. >> >> -- >> Eric Botcazou ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-07 15:37 ` Evgeny Stupachenko @ 2014-11-11 15:31 ` Evgeny Stupachenko 2014-11-11 16:39 ` Eric Botcazou 0 siblings, 1 reply; 23+ messages in thread From: Evgeny Stupachenko @ 2014-11-11 15:31 UTC (permalink / raw) To: Eric Botcazou; +Cc: Jan Hubicka, GCC Patches, Richard Biener, Uros Bizjak 150 and 200 make Silvermont performance better on 173.applu (+8%) and 183.equake (+3%); Haswell spec2006 performance stays almost unchanged. Higher value of 300 leave the performance of mentioned tests unchanged, but add some regressions on other benchmarks. So I like 200 as well as 120 and 150, but can confirm performance gains only for x86. On Fri, Nov 7, 2014 at 6:37 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > So are there any objections to enable this > (PARAM_MAX_COMPLETELY_PEELED_INSNS increase from 100 to 120) for x86? > > On Fri, Oct 31, 2014 at 7:52 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: >> I've measured spec2000, spec2006 as well and EEMBC for Silvermont in addition. >> 100->120 change gives gain for Silvermont, the results on Haswell are flat. >> >> On Fri, Oct 31, 2014 at 3:14 PM, Eric Botcazou <ebotcazou@adacore.com> wrote: >>>> Agreed, I think the value of 100 was set decade ago by Zdenek and me >>>> completely artifically. I do not recall any serious tuning of this flag. >>> >>> Are you talking bout PARAM_MAX_COMPLETELY_PEELED_INSNS here? If so, see: >>> https://gcc.gnu.org/ml/gcc-patches/2012-11/msg01193.html >>> >>> We have experienced performance regressions because of this arbitrary change >>> and bumped it back to 200 unconditionally. >>> >>> -- >>> Eric Botcazou ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-11 15:31 ` Evgeny Stupachenko @ 2014-11-11 16:39 ` Eric Botcazou 2014-11-11 23:58 ` Jan Hubicka 0 siblings, 1 reply; 23+ messages in thread From: Eric Botcazou @ 2014-11-11 16:39 UTC (permalink / raw) To: Evgeny Stupachenko; +Cc: gcc-patches, Jan Hubicka, Richard Biener, Uros Bizjak > 150 and 200 make Silvermont performance better on 173.applu (+8%) and > 183.equake (+3%); Haswell spec2006 performance stays almost unchanged. > Higher value of 300 leave the performance of mentioned tests > unchanged, but add some regressions on other benchmarks. > > So I like 200 as well as 120 and 150, but can confirm performance > gains only for x86. IMO it's either 150 or 200. We chose 200 for our 4.9-based compiler because this gave the performance boost without affecting the code size (on x86-64) and because this was previously 400, but it's your call. -- Eric Botcazou ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-11 16:39 ` Eric Botcazou @ 2014-11-11 23:58 ` Jan Hubicka 2014-11-12 14:05 ` Evgeny Stupachenko 0 siblings, 1 reply; 23+ messages in thread From: Jan Hubicka @ 2014-11-11 23:58 UTC (permalink / raw) To: Eric Botcazou Cc: Evgeny Stupachenko, gcc-patches, Jan Hubicka, Richard Biener, Uros Bizjak > > 150 and 200 make Silvermont performance better on 173.applu (+8%) and > > 183.equake (+3%); Haswell spec2006 performance stays almost unchanged. > > Higher value of 300 leave the performance of mentioned tests > > unchanged, but add some regressions on other benchmarks. > > > > So I like 200 as well as 120 and 150, but can confirm performance > > gains only for x86. > > IMO it's either 150 or 200. We chose 200 for our 4.9-based compiler because > this gave the performance boost without affecting the code size (on x86-64) > and because this was previously 400, but it's your call. Both 150 or 200 globally work for me if there is not too much of code size bloat (did not see code size mentioned here). What I did before decreasing the bounds was strenghtening the loop iteraton count bounds and adding logic the predicts constant propagation enabled by unrolling. For this reason 400 became too large as we did a lot more complete unrolling than before. Also 400 in older compilers is not really 400 in newer. Because I saw performance to drop only with values bellow 50, I went for 100. It would be very interesting to actually analyze what happends for those two benchmarks (that should not be too hard with perf). Honza ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-11 23:58 ` Jan Hubicka @ 2014-11-12 14:05 ` Evgeny Stupachenko 2014-11-21 11:01 ` Evgeny Stupachenko 0 siblings, 1 reply; 23+ messages in thread From: Evgeny Stupachenko @ 2014-11-12 14:05 UTC (permalink / raw) To: Jan Hubicka; +Cc: Eric Botcazou, GCC Patches, Richard Biener, Uros Bizjak Code size for spec2000 is almost unchanged (many benchmarks have the same binaries). For those that are changed we have the following numbers (200 vs 100, both dynamic build -Ofast -funroll-loops -flto): 183.equake +10% 164.gzip, 173.applu +3,5% 187.facerec, 191.fma3d +2,5% 200.sixstrack +2% 177.mesa, 178.galgel +1% On Wed, Nov 12, 2014 at 2:51 AM, Jan Hubicka <hubicka@ucw.cz> wrote: >> > 150 and 200 make Silvermont performance better on 173.applu (+8%) and >> > 183.equake (+3%); Haswell spec2006 performance stays almost unchanged. >> > Higher value of 300 leave the performance of mentioned tests >> > unchanged, but add some regressions on other benchmarks. >> > >> > So I like 200 as well as 120 and 150, but can confirm performance >> > gains only for x86. >> >> IMO it's either 150 or 200. We chose 200 for our 4.9-based compiler because >> this gave the performance boost without affecting the code size (on x86-64) >> and because this was previously 400, but it's your call. > > Both 150 or 200 globally work for me if there is not too much of code size > bloat (did not see code size mentioned here). > > What I did before decreasing the bounds was strenghtening the loop iteraton > count bounds and adding logic the predicts constant propagation enabled by > unrolling. For this reason 400 became too large as we did a lot more complete > unrolling than before. Also 400 in older compilers is not really 400 in newer. > > Because I saw performance to drop only with values bellow 50, I went for 100. > It would be very interesting to actually analyze what happends for those two > benchmarks (that should not be too hard with perf). > > Honza ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-12 14:05 ` Evgeny Stupachenko @ 2014-11-21 11:01 ` Evgeny Stupachenko 2014-11-21 11:21 ` Uros Bizjak 0 siblings, 1 reply; 23+ messages in thread From: Evgeny Stupachenko @ 2014-11-21 11:01 UTC (permalink / raw) To: Jan Hubicka; +Cc: Eric Botcazou, GCC Patches, Richard Biener, Uros Bizjak PING. "200" currently looks optimal for x86. Let's commit the following: 2014-11-21 Evgeny Stupachenko <evstupac@gmail.com> * config/i386/i386.c (ix86_option_override_internal): Increase PARAM_MAX_COMPLETELY_PEELED_INSNS. diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c index 6337aa5..5ac10eb 100644 --- a/gcc/config/i386/i386.c +++ b/gcc/config/i386/i386.c @@ -4081,6 +4081,12 @@ ix86_option_override_internal (bool main_args_p, opts->x_param_values, opts_set->x_param_values); + /* Extend full peel max insns parameter for x86. */ + maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, + 200, + opts->x_param_values, + opts_set->x_param_values); + /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ if (opts->x_flag_prefetch_loop_arrays < 0 && HAVE_prefetch On Wed, Nov 12, 2014 at 5:02 PM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > Code size for spec2000 is almost unchanged (many benchmarks have the > same binaries). > For those that are changed we have the following numbers (200 vs 100, > both dynamic build -Ofast -funroll-loops -flto): > 183.equake +10% > 164.gzip, 173.applu +3,5% > 187.facerec, 191.fma3d +2,5% > 200.sixstrack +2% > 177.mesa, 178.galgel +1% > > > On Wed, Nov 12, 2014 at 2:51 AM, Jan Hubicka <hubicka@ucw.cz> wrote: >>> > 150 and 200 make Silvermont performance better on 173.applu (+8%) and >>> > 183.equake (+3%); Haswell spec2006 performance stays almost unchanged. >>> > Higher value of 300 leave the performance of mentioned tests >>> > unchanged, but add some regressions on other benchmarks. >>> > >>> > So I like 200 as well as 120 and 150, but can confirm performance >>> > gains only for x86. >>> >>> IMO it's either 150 or 200. We chose 200 for our 4.9-based compiler because >>> this gave the performance boost without affecting the code size (on x86-64) >>> and because this was previously 400, but it's your call. >> >> Both 150 or 200 globally work for me if there is not too much of code size >> bloat (did not see code size mentioned here). >> >> What I did before decreasing the bounds was strenghtening the loop iteraton >> count bounds and adding logic the predicts constant propagation enabled by >> unrolling. For this reason 400 became too large as we did a lot more complete >> unrolling than before. Also 400 in older compilers is not really 400 in newer. >> >> Because I saw performance to drop only with values bellow 50, I went for 100. >> It would be very interesting to actually analyze what happends for those two >> benchmarks (that should not be too hard with perf). >> >> Honza ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-21 11:01 ` Evgeny Stupachenko @ 2014-11-21 11:21 ` Uros Bizjak 2014-11-22 10:17 ` Eric Botcazou 0 siblings, 1 reply; 23+ messages in thread From: Uros Bizjak @ 2014-11-21 11:21 UTC (permalink / raw) To: Evgeny Stupachenko Cc: Jan Hubicka, Eric Botcazou, GCC Patches, Richard Biener On Fri, Nov 21, 2014 at 11:46 AM, Evgeny Stupachenko <evstupac@gmail.com> wrote: > PING. > "200" currently looks optimal for x86. > Let's commit the following: > > 2014-11-21 Evgeny Stupachenko <evstupac@gmail.com> > * config/i386/i386.c (ix86_option_override_internal): Increase > PARAM_MAX_COMPLETELY_PEELED_INSNS. OK. Looks like a good performance vs. codesize tradeoff. Uros. ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-21 11:21 ` Uros Bizjak @ 2014-11-22 10:17 ` Eric Botcazou 2014-11-22 11:24 ` Uros Bizjak 0 siblings, 1 reply; 23+ messages in thread From: Eric Botcazou @ 2014-11-22 10:17 UTC (permalink / raw) To: Uros Bizjak; +Cc: gcc-patches, Evgeny Stupachenko, Jan Hubicka, Richard Biener > OK. Looks like a good performance vs. codesize tradeoff. Yes, but IMO this should be done in the generic code, unrolling small loops is profitable on most architectures. -- Eric Botcazou ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-22 10:17 ` Eric Botcazou @ 2014-11-22 11:24 ` Uros Bizjak 2014-11-22 12:24 ` Eric Botcazou 0 siblings, 1 reply; 23+ messages in thread From: Uros Bizjak @ 2014-11-22 11:24 UTC (permalink / raw) To: Eric Botcazou Cc: gcc-patches, Evgeny Stupachenko, Jan Hubicka, Richard Biener On Sat, Nov 22, 2014 at 10:49 AM, Eric Botcazou <ebotcazou@adacore.com> wrote: >> OK. Looks like a good performance vs. codesize tradeoff. > > Yes, but IMO this should be done in the generic code, unrolling small loops is > profitable on most architectures. Yeah, but after a couple of pings for a generic change, we went the target way. Uros. ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-22 11:24 ` Uros Bizjak @ 2014-11-22 12:24 ` Eric Botcazou 2014-11-22 19:31 ` Richard Biener 0 siblings, 1 reply; 23+ messages in thread From: Eric Botcazou @ 2014-11-22 12:24 UTC (permalink / raw) To: Uros Bizjak; +Cc: gcc-patches, Evgeny Stupachenko, Jan Hubicka, Richard Biener > Yeah, but after a couple of pings for a generic change, we went the target > way. That's a bit of a shame, the 400 -> 100 change was very likely tested only on x86-64 and nevetheless applied to the generic code, so the fix repairing the damages should also be applied to the generic code. -- Eric Botcazou ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-22 12:24 ` Eric Botcazou @ 2014-11-22 19:31 ` Richard Biener 2014-11-22 22:38 ` Uros Bizjak 0 siblings, 1 reply; 23+ messages in thread From: Richard Biener @ 2014-11-22 19:31 UTC (permalink / raw) To: Eric Botcazou, Uros Bizjak; +Cc: gcc-patches, Evgeny Stupachenko, Jan Hubicka On November 22, 2014 12:24:22 PM CET, Eric Botcazou <ebotcazou@adacore.com> wrote: >> Yeah, but after a couple of pings for a generic change, we went the >target >> way. > >That's a bit of a shame, the 400 -> 100 change was very likely tested >only on >x86-64 and nevetheless applied to the generic code, so the fix >repairing the >damages should also be applied to the generic code. A patch to bump the generic limit is OK. Targets that dont want it can reduce it in target specific code. Richard. ^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly 2014-11-22 19:31 ` Richard Biener @ 2014-11-22 22:38 ` Uros Bizjak 0 siblings, 0 replies; 23+ messages in thread From: Uros Bizjak @ 2014-11-22 22:38 UTC (permalink / raw) To: Richard Biener Cc: Eric Botcazou, gcc-patches, Evgeny Stupachenko, Jan Hubicka [-- Attachment #1: Type: text/plain, Size: 904 bytes --] On Sat, Nov 22, 2014 at 7:38 PM, Richard Biener <richard.guenther@gmail.com> wrote: > On November 22, 2014 12:24:22 PM CET, Eric Botcazou <ebotcazou@adacore.com> wrote: >>> Yeah, but after a couple of pings for a generic change, we went the >>target >>> way. >> >>That's a bit of a shame, the 400 -> 100 change was very likely tested >>only on >>x86-64 and nevetheless applied to the generic code, so the fix >>repairing the >>damages should also be applied to the generic code. > > A patch to bump the generic limit is OK. > > Targets that dont want it can reduce it in target specific code. I have committed the attached patch: 2014-11-22 Uros Bizjak <ubizjak@gmail.com> * params.def (PARAM_MAX_COMPLETELY_PEELED_INSNS): Increase to 200. * config/i386/i386.c (ix86_option_override_internal): Do not increase PARAM_MAX_COMPLETELY_PEELED_INSNS. Bootstrapped on x86_64-linux-gnu. Uros. [-- Attachment #2: p.diff.txt --] [-- Type: text/plain, Size: 1167 bytes --] Index: params.def =================================================================== --- params.def (revision 217961) +++ params.def (working copy) @@ -303,7 +303,7 @@ DEFPARAM(PARAM_MAX_PEEL_BRANCHES, DEFPARAM(PARAM_MAX_COMPLETELY_PEELED_INSNS, "max-completely-peeled-insns", "The maximum number of insns of a completely peeled loop", - 100, 0, 0) + 200, 0, 0) /* The maximum number of peelings of a single loop that is peeled completely. */ DEFPARAM(PARAM_MAX_COMPLETELY_PEEL_TIMES, "max-completely-peel-times", Index: config/i386/i386.c =================================================================== --- config/i386/i386.c (revision 217961) +++ config/i386/i386.c (working copy) @@ -4142,12 +4142,6 @@ ix86_option_override_internal (bool main_args_p, opts->x_param_values, opts_set->x_param_values); - /* Increase full peel max insns parameter for x86. */ - maybe_set_param_value (PARAM_MAX_COMPLETELY_PEELED_INSNS, - 200, - opts->x_param_values, - opts_set->x_param_values); - /* Enable sw prefetching at -O3 for CPUS that prefetching is helpful. */ if (opts->x_flag_prefetch_loop_arrays < 0 && HAVE_prefetch ^ permalink raw reply [flat|nested] 23+ messages in thread
end of thread, other threads:[~2014-11-22 19:03 UTC | newest] Thread overview: 23+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2014-10-10 15:40 [PATCH x86] Increase PARAM_MAX_COMPLETELY_PEELED_INSNS when branch is costly Evgeny Stupachenko 2014-10-13 8:09 ` Richard Biener 2014-10-13 11:12 ` Evgeny Stupachenko 2014-10-13 12:42 ` Jan Hubicka 2014-10-27 8:23 ` Evgeny Stupachenko 2014-10-28 12:22 ` Evgeny Stupachenko 2014-10-30 8:32 ` Uros Bizjak 2014-10-30 17:54 ` Jan Hubicka 2014-10-30 18:48 ` Evgeny Stupachenko 2014-10-31 13:03 ` Eric Botcazou 2014-10-31 17:07 ` Evgeny Stupachenko 2014-11-07 15:37 ` Evgeny Stupachenko 2014-11-11 15:31 ` Evgeny Stupachenko 2014-11-11 16:39 ` Eric Botcazou 2014-11-11 23:58 ` Jan Hubicka 2014-11-12 14:05 ` Evgeny Stupachenko 2014-11-21 11:01 ` Evgeny Stupachenko 2014-11-21 11:21 ` Uros Bizjak 2014-11-22 10:17 ` Eric Botcazou 2014-11-22 11:24 ` Uros Bizjak 2014-11-22 12:24 ` Eric Botcazou 2014-11-22 19:31 ` Richard Biener 2014-11-22 22:38 ` Uros Bizjak
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).