langs-performance/results/analyze

#!/usr/bin/perl
use strict;
use warnings;

use Data::Dumper;

sub analyze_run_data {
	my $test_name = shift;
	my $data = shift;

	my $i;
	my $deviation;
	my $cnt;
	my $averages = {};
	my $k;
	my $avg;

	$cnt = scalar @{$data};

	for ($i = 0; $i < $cnt; ++$i) {
		if ($data->[$i]->{'swaps'} != 0) {
			# Did we run out of memory?
			warn(sprintf(
				"%s: run_try=%d: swaps is non-zero: %d\n",
				$test_name, $i + 1, $data->[$i]->{'swaps'}
			));
		}
		if ($data->[$i]->{'sys_t'} + $data->[$i]->{'user_t'} > $data->[$i]->{'real_t'}) {
			warn(sprintf(
				"%s: run_try=%d: CPU time (%.1f) is bigger than wall-clock time (%.1f); multi-threaded?\n",
				$test_name, $i + 1,
				$data->[$i]->{'sys_t'} + $data->[$i]->{'user_t'},
				$data->[$i]->{'real_t'}
			));
		}
	}

	foreach $k (qw/ real_t sys_t user_t cpu_t /) {
		$averages->{$k} = 0;
	}

	for ($i = 0; $i < $cnt; ++$i) {
		$data->[$i]->{'cpu_t'} = $data->[$i]->{'sys_t'} + $data->[$i]->{'user_t'};

		$data->[$i]->{'per_10_lines'} = {};
		foreach $k (qw/ real_t sys_t user_t cpu_t /) {
			$data->[$i]->{'per_10_lines'}->{$k} = $data->[$i]->{$k} / $data->[$i]->{'nlines'} * 10;

			# XXX: Average values for 10 lines of output.
			# XXX: So that we can compare with the previous test method.
			$averages->{$k} += $data->[$i]->{'per_10_lines'}->{$k};
		}
	}

	foreach $k (qw/ real_t sys_t user_t cpu_t /) {
		$averages->{$k} = $averages->{$k} / $cnt;
	}

	#print Dumper($data)."\n";
	#print Dumper($averages)."\n";

	for ($i = 0; $i < $cnt; ++$i) {
		foreach $k (qw/ real_t cpu_t /) {
			$avg = $averages->{$k};
			$deviation = (($avg - $data->[$i]->{'per_10_lines'}->{$k}) / $avg) * 100;
			if (abs($deviation) > 5) { # percents
				warn(sprintf(
					"%s: run_try=%d: Measurement deviations are too high for '%s': %.1f\n",
					$test_name, $i + 1, $k, $deviation
				));
			}
		}
	}

	return $averages;
}

sub main {
	my $line;
	my $data = {};
	my $averages;
	my $k;
	my $test_data;
	my $test_name;

	while ($line = <STDIN>) {
		chomp($line);

		if ($line =~ /^\s*#/) {
			next;
		}

		if ($line !~ /^real_TIME:(\d+\.\d+)sec user_CPU:(\d+\.\d+)sec sys_CPU:(\d+\.\d+)sec max_RSS:(\d+)kb swaps:(\d+) ctx_sw:(\d+)\+(\d+) nlines:(\d+) run_try:(\d+)  header:'([^']+)' version:'([^']+)' src_file:(\S+)$/) {
			die("Unable to parse line: $line\n");
		}

		my $row = {}; # new reference every time
		my $run_try;
		my $test_name;
		$row->{'real_t'} = $1;
		$row->{'user_t'} = $2;
		$row->{'sys_t'} = $3;
		$row->{'max_RSS'} = $4;
		$row->{'swaps'} = $5;
		$row->{'ctx_sw_unvol'} = $6;
		$row->{'ctx_sw_vol'} = $7;
		$row->{'nlines'} = $8;
		$run_try = $9 - 1; # XXX: fix index
		#
		$test_name = $10;
		$row->{'version'} = $11;
		$row->{'src_file'} = $12;

		# trim white-space
		$row->{'version'} =~ s/^\s*//;
		$row->{'version'} =~ s/\s*$//;

		if (!exists($data->{$test_name})) {
			$data->{$test_name} = {
				'meta' => {},
				'run_data' => [],
			};
			foreach $k (qw/ version src_file /) {
				$data->{$test_name}->{'meta'}->{$k} = $row->{$k};
			}
		}

		foreach $k (qw/ version src_file /) {
			if ($data->{$test_name}->{'meta'}->{$k} ne $row->{$k}) {
				die("Sanity check failed for meta: $line");
			}
		}

		if (exists($data->{$test_name}->{'run_data'}->[$run_try])) {
			die("Duplicate run#: $line");
		}
		delete($row->{'version'});
		delete($row->{'src_file'});
		$data->{$test_name}->{'run_data'}->[$run_try] = $row;
		foreach $k (qw/ version src_file /) {
			delete($data->{$test_name}->{'run_data'}->[$run_try]->{$k});
		}
	}

	while (($test_name, $test_data) = each %{$data}) {
		$averages = analyze_run_data($test_name, $test_data->{'run_data'});
		printf(
			"%-30s: user_t=%.3f sys_t=%.3f cpu_t=%.3f\n",
			$test_name,
			$averages->{'user_t'},
			$averages->{'sys_t'},
			$averages->{'cpu_t'}
		);
	}
}

main();