[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
bench: minor improvements
From: |
Akim Demaille |
Subject: |
bench: minor improvements |
Date: |
Sat, 25 Apr 2020 11:37:54 +0200 |
I am quite disappointed with the benchmarks, they are way to sensitive. Even
the order in which you run them has a clear impact. I was hoping Google
Benchmarks would help on this regard, but it does not seem to be the case.
commit 30ce43b2b5384c6cb96a100942d996b40ac95599
Author: Akim Demaille <address@hidden>
Date: Sat Apr 25 08:51:52 2020 +0200
bench: minor improvements
* etc/bench.pl.in: Don't force parse.error=detailed
Use a simpler way to display the pseudo %bison directive.
(&bench_with_gbenchmark): Give details about the compiler.
diff --git a/etc/bench.pl.in b/etc/bench.pl.in
index 6ba0be0c..1498dd6f 100755
--- a/etc/bench.pl.in
+++ b/etc/bench.pl.in
@@ -252,7 +252,6 @@ sub generate_grammar_triangular ($$@)
my $out = new IO::File ">$base.y"
or die;
print $out <<EOF;
-%define parse.error detailed
%{
#include <stdio.h>
#include <stdlib.h>
@@ -388,7 +387,6 @@ sub generate_grammar_calc ($$@)
%}
%define api.value.type union
-%define parse.error detailed
$directives
%{
@@ -812,7 +810,8 @@ sub compile ($)
my ($base) = @_;
my $compiler = compiler ($base);
- my $my_bison = `sed -ne '/%bison "\\(.*\\)"/{s//\\1/;p;q;}' $base.y`;
+ my $my_bison = `sed -ne '/[/][/] %bison "\\(.*\\)"/{s//\\1/;p;q;}' $base.y`;
+ chop $my_bison;
run ((length $my_bison ? $my_bison : $bison) . " $base.y -o $base.c");
if ($gbench)
{
@@ -902,8 +901,13 @@ directive list.
sub bench_with_gbenchmark ($@)
{
my ($grammar, @directive) = @_;
+ my $compiler = compiler ("y0");
my $readme = new IO::File ">README.md";
+ print $readme <<EOF;
+compiler: $compiler $cflags
+EOF
+
my @obj = ();
for my $i (0 .. $#directive)
{
@@ -923,7 +927,6 @@ sub bench_with_gbenchmark ($@)
BENCHMARK_MAIN();
EOF
- my $compiler = compiler ("y0");
run "$compiler -o main $cflags main.c @obj -lbenchmark";
run "./main | tee -a README.md";
}
@@ -1121,7 +1124,7 @@ sub parse_dir ()
elsif ($token[0] eq '%b')
{
shift @token;
- @res = ("/*\n%bison \"$token[0]\"\\\n*/");
+ @res = ("// %bison \"$token[0]\"");
}
else
{
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- bench: minor improvements,
Akim Demaille <=