summaryrefslogtreecommitdiff
path: root/lib/Benchmark.t
diff options
context:
space:
mode:
authorDavid Mitchell <davem@iabyn.com>2011-06-10 14:07:28 +0100
committerDavid Mitchell <davem@iabyn.com>2011-06-10 14:09:27 +0100
commitbb6c6e4b8d10f2e460a7fe48e677d3d998a7f77d (patch)
tree57f9a419860f6c409f117314cc2d7cbabd376f45 /lib/Benchmark.t
parentee682a851e75ef16ba37952dd6a8f29c835ef554 (diff)
downloadperl-bb6c6e4b8d10f2e460a7fe48e677d3d998a7f77d.tar.gz
Benchmart.t: improved diagnostics
use diag() to show absolutely all variables on failing the notorious test 15. Maybe this will help nail it one day...
Diffstat (limited to 'lib/Benchmark.t')
-rw-r--r--lib/Benchmark.t34
1 files changed, 22 insertions, 12 deletions
diff --git a/lib/Benchmark.t b/lib/Benchmark.t
index e54d8294c0..fdf2d039a8 100644
--- a/lib/Benchmark.t
+++ b/lib/Benchmark.t
@@ -63,10 +63,11 @@ isnt ($baz, 0, "benchmarked code was run");
my $in_threesecs = $threesecs->iters;
print "# in_threesecs=$in_threesecs iterations\n";
ok ($in_threesecs > 0, "iters returned positive iterations");
-my $cpu = $threesecs->[1] + $threesecs->[2]; # user + sys
-cmp_ok($cpu, '>=', 3.0, "3s cpu is at least 3s");
-$in_threesecs *= (3/$cpu); # adjust because may not have run for exactly 3s
-print "# in_threesecs=$in_threesecs adjusted iterations\n";
+my $cpu3 = $threesecs->[1] + $threesecs->[2]; # user + sys
+cmp_ok($cpu3, '>=', 3.0, "3s cpu3 is at least 3s");
+my $in_threesecs_adj = $in_threesecs;
+$in_threesecs_adj *= (3/$cpu3); # adjust because may not have run for exactly 3s
+print "# in_threesecs_adj=$in_threesecs_adj adjusted iterations\n";
my $estimate = int (100 * $in_threesecs / 3) / 100;
print "# from the 3 second run estimate $estimate iterations in 1 second...\n";
@@ -77,16 +78,25 @@ isnt ($baz, 0, "benchmarked code was run");
my $in_onesec = $onesec->iters;
print "# in_onesec=$in_onesec iterations\n";
ok ($in_onesec > 0, "iters returned positive iterations");
-$cpu = $onesec->[1] + $onesec->[2]; # user + sys
-cmp_ok($cpu, '>=', 1.0, "1s cpu is at least 1s");
-$in_onesec *= (1/$cpu); # adjust because may not have run for exactly 1s
-print "# in_onesec=$in_onesec adjusted iterations\n";
+my $cpu1 = $onesec->[1] + $onesec->[2]; # user + sys
+cmp_ok($cpu1, '>=', 1.0, "is cpu1 is at least 1s");
+my $in_onesec_adj = $in_onesec;
+$in_onesec_adj *= (1/$cpu1); # adjust because may not have run for exactly 1s
+print "# in_onesec_adj=$in_onesec_adj adjusted iterations\n";
{
- my $difference = $in_onesec - $estimate;
- my $actual = abs ($difference / $in_onesec);
- cmp_ok($actual, '<=', $delta, "is $in_onesec within $delta of estimate ($estimate)")
- or diag("# $in_onesec is between " . ($delta / 2) . " and $delta of estimate. Not that safe.");
+ my $difference = $in_onesec_adj - $estimate;
+ my $actual = abs ($difference / $in_onesec_adj);
+ cmp_ok($actual, '<=', $delta, "is $in_onesec_adj within $delta of estimate ($estimate)")
+ or do {
+ diag(" in_threesecs = $in_threesecs");
+ diag(" in_threesecs_adj = $in_threesecs_adj");
+ diag(" cpu3 = $cpu3");
+ diag(" estimate = $estimate");
+ diag(" in_onesec = $in_onesec");
+ diag(" in_onesec_adj = $in_onesec_adj");
+ diag(" cpu1 = $cpu1");
+ };
}
# I found that the eval'ed version was 3 times faster than the coderef.