This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[PATCH] Compare predictor values with these defined in predict.def.
- From: Martin Liška <mliska at suse dot cz>
- To: GCC Patches <gcc-patches at gcc dot gnu dot org>
- Cc: Jan Hubicka <hubicka at ucw dot cz>
- Date: Tue, 30 May 2017 09:13:00 +0200
- Subject: [PATCH] Compare predictor values with these defined in predict.def.
- Authentication-results: sourceware.org; auth=none
Hello.
It's quite handy to compare numbers given by analyze_brprob.py script and
these that live in predict.def. With the patch numbers are displayed:
HEURISTICS BRANCHES (REL) BR. HITRATE HITRATE COVERAGE COVERAGE (REL) predict.def (REL)
unconditional jump 3 0.0% 100.00% 100.00% / 100.00% 2459650 2.46M 0.0%
fp_opcode (on trees) 5 0.0% 100.00% 100.00% / 100.00% 714059 714.06K 0.0% 90% 10.0%
Fortran loop preheader 9 0.0% 100.00% 100.00% / 100.00% 31044476 31.04M 0.0% 99% 1.0%
and one can use:
-w, --write-def-file Modify predict.def file in order to set new numbers
in order to propagate numbers to predict.def file.
I'm going to install the patch.
Martin
>From 45336a649ef735f62f1d669b3228bbf1b194ac4f Mon Sep 17 00:00:00 2001
From: marxin <mliska@suse.cz>
Date: Mon, 29 May 2017 15:19:24 +0200
Subject: [PATCH] Compare predictor values with these defined in predict.def.
contrib/ChangeLog:
2017-05-30 Martin Liska <mliska@suse.cz>
* analyze_brprob.py: Add new argument to parse and modify
predict.def file.
* analyze_brprob_spec.py: Likewise.
---
contrib/analyze_brprob.py | 80 ++++++++++++++++++++++++++++++++++++------
contrib/analyze_brprob_spec.py | 4 +++
2 files changed, 74 insertions(+), 10 deletions(-)
diff --git a/contrib/analyze_brprob.py b/contrib/analyze_brprob.py
index b4dbbc4ac15..8f83b5a0221 100755
--- a/contrib/analyze_brprob.py
+++ b/contrib/analyze_brprob.py
@@ -90,6 +90,48 @@ def median(values):
values.sort()
return values[int(len(values) / 2)]
+class PredictDefFile:
+ def __init__(self, path):
+ self.path = path
+ self.predictors = {}
+
+ def parse_and_modify(self, heuristics, write_def_file):
+ lines = [x.rstrip() for x in open(self.path).readlines()]
+
+ p = None
+ modified_lines = []
+ for l in lines:
+ if l.startswith('DEF_PREDICTOR'):
+ m = re.match('.*"(.*)".*', l)
+ p = m.group(1)
+ elif l == '':
+ p = None
+
+ if p != None:
+ heuristic = [x for x in heuristics if x.name == p]
+ heuristic = heuristic[0] if len(heuristic) == 1 else None
+
+ m = re.match('.*HITRATE \(([^)]*)\).*', l)
+ if (m != None):
+ self.predictors[p] = int(m.group(1))
+
+ # modify the line
+ if heuristic != None:
+ new_line = (l[:m.start(1)]
+ + str(round(heuristic.get_hitrate()))
+ + l[m.end(1):])
+ l = new_line
+ p = None
+ elif 'PROB_VERY_LIKELY' in l:
+ self.predictors[p] = 100
+ modified_lines.append(l)
+
+ # save the file
+ if write_def_file:
+ with open(self.path, 'w+') as f:
+ for l in modified_lines:
+ f.write(l + '\n')
+
class Summary:
def __init__(self, name):
self.name = name
@@ -113,7 +155,11 @@ class Summary:
v /= 1000.0
return "%.1f%s" % (v, 'Y')
- def print(self, branches_max, count_max):
+ def print(self, branches_max, count_max, predict_def):
+ predicted_as = None
+ if predict_def != None and self.name in predict_def.predictors:
+ predicted_as = predict_def.predictors[self.name]
+
print('%-40s %8i %5.1f%% %11.2f%% %7.2f%% / %6.2f%% %14i %8s %5.1f%%' %
(self.name, self.branches,
percentage(self.branches, branches_max),
@@ -121,7 +167,12 @@ class Summary:
self.get_hitrate(),
percentage(self.fits, self.count),
self.count, self.count_formatted(),
- percentage(self.count, count_max)))
+ percentage(self.count, count_max)), end = '')
+
+ if predicted_as != None:
+ print('%12i%% %5.1f%%' % (predicted_as,
+ self.get_hitrate() - predicted_as), end = '')
+ print()
class Profile:
def __init__(self, filename):
@@ -156,7 +207,7 @@ class Profile:
def count_max(self):
return max([v.count for k, v in self.heuristics.items()])
- def print_group(self, sorting, group_name, heuristics):
+ def print_group(self, sorting, group_name, heuristics, predict_def):
count_max = self.count_max()
branches_max = self.branches_max()
@@ -170,11 +221,12 @@ class Profile:
elif sorting == 'name':
sorter = lambda x: x.name.lower()
- print('%-40s %8s %6s %12s %18s %14s %8s %6s' %
+ print('%-40s %8s %6s %12s %18s %14s %8s %6s %12s %6s' %
('HEURISTICS', 'BRANCHES', '(REL)',
- 'BR. HITRATE', 'HITRATE', 'COVERAGE', 'COVERAGE', '(REL)'))
+ 'BR. HITRATE', 'HITRATE', 'COVERAGE', 'COVERAGE', '(REL)',
+ 'predict.def', '(REL)'))
for h in sorted(heuristics, key = sorter):
- h.print(branches_max, count_max)
+ h.print(branches_max, count_max, predict_def)
def dump(self, sorting):
heuristics = self.heuristics.values()
@@ -182,14 +234,19 @@ class Profile:
print('No heuristics available')
return
+ predict_def = None
+ if args.def_file != None:
+ predict_def = PredictDefFile(args.def_file)
+ predict_def.parse_and_modify(heuristics, args.write_def_file)
+
special = list(filter(lambda x: x.name in counter_aggregates,
heuristics))
normal = list(filter(lambda x: x.name not in counter_aggregates,
heuristics))
- self.print_group(sorting, 'HEURISTICS', normal)
+ self.print_group(sorting, 'HEURISTICS', normal, predict_def)
print()
- self.print_group(sorting, 'HEURISTIC AGGREGATES', special)
+ self.print_group(sorting, 'HEURISTIC AGGREGATES', special, predict_def)
if len(self.niter_vector) > 0:
print ('\nLoop count: %d' % len(self.niter_vector)),
@@ -206,13 +263,16 @@ parser.add_argument('dump_file', metavar = 'dump_file',
parser.add_argument('-s', '--sorting', dest = 'sorting',
choices = ['branches', 'branch-hitrate', 'hitrate', 'coverage', 'name'],
default = 'branches')
+parser.add_argument('-d', '--def-file', help = 'path to predict.def')
+parser.add_argument('-w', '--write-def-file', action = 'store_true',
+ help = 'Modify predict.def file in order to set new numbers')
args = parser.parse_args()
-profile = Profile(sys.argv[1])
+profile = Profile(args.dump_file)
r = re.compile(' (.*) heuristics( of edge [0-9]*->[0-9]*)?( \\(.*\\))?: (.*)%.*exec ([0-9]*) hit ([0-9]*)')
loop_niter_str = ';; profile-based iteration count: '
-for l in open(args.dump_file).readlines():
+for l in open(args.dump_file):
m = r.match(l)
if m != None and m.group(3) == None:
name = m.group(1)
diff --git a/contrib/analyze_brprob_spec.py b/contrib/analyze_brprob_spec.py
index 908db3960dc..e621853ba4e 100755
--- a/contrib/analyze_brprob_spec.py
+++ b/contrib/analyze_brprob_spec.py
@@ -30,6 +30,7 @@ parser.add_argument('location', metavar = 'dump_file',
parser.add_argument('-s', '--sorting', dest = 'sorting',
choices = ['branches', 'branch-hitrate', 'hitrate', 'coverage', 'name'],
default = 'branches')
+parser.add_argument('-d', '--def-file', help = 'path to predict.def')
args = parser.parse_args()
@@ -56,6 +57,9 @@ for b in sorted(benchmarks):
sys.stdout.flush()
p = [os.path.join(os.path.dirname(script_location), 'analyze_brprob.py'),
temp.name, '--sorting', args.sorting]
+ if args.def_file != None:
+ p += ['-d', args.def_file]
+
p = subprocess.check_call(p)
sys.stdout.flush()
--
2.12.2