diff options
Diffstat (limited to 'bm/bm_history.pl')
| -rw-r--r-- | bm/bm_history.pl | 90 |
1 files changed, 56 insertions, 34 deletions
diff --git a/bm/bm_history.pl b/bm/bm_history.pl index 5ebf7e1..5f2ea4a 100644 --- a/bm/bm_history.pl +++ b/bm/bm_history.pl @@ -19,6 +19,9 @@ my $bm_repo = "bm_repo"; my $sample_rate = int($total_commits / 5) || 1; my %results; +my $rebase_count = 0; +my %last_base_hashes; + sub get_size { my $dir = shift; return "0 KB" unless -d $dir; @@ -35,71 +38,89 @@ sub count_inodes { return $count || 0; } +sub track_rebases { + return unless -f ".urn/index"; + open(my $fh, '<', ".urn/index") or return; + while (<$fh>) { + chomp; + my @cols = split(/\t/); + next unless @cols >= 6; + my ($b_hash, $path) = ($cols[2], $cols[5]); + if (exists $last_base_hashes{$path} && $last_base_hashes{$path} ne $b_hash) { + $rebase_count++; + } + $last_base_hashes{$path} = $b_hash; + } + close($fh); +} + sub run_history_benchmark { my ($tool_name) = @_; print ">>> Starting History Benchmark: $tool_name\n"; remove_tree($bm_repo) if -d $bm_repo; - system("perl $seed_bin $files $depth > /dev/null 2>&1") == 0 - or die "FATAL: seed.pl failed.\n"; + system("perl $seed_bin $files $depth > /dev/null 2>&1"); - chdir($bm_repo) or die "FATAL: Could not enter $bm_repo: $!\n"; + chdir($bm_repo) or die $!; my $init_cmd = ($tool_name eq "URN") ? "perl $urn_bin init" : "git init"; + my $add_cmd = ($tool_name eq "URN") ? "perl $urn_bin add ." : "git add ."; + system("$init_cmd > /dev/null 2>&1"); + system("$add_cmd > /dev/null 2>&1"); + system(($tool_name eq "URN" ? "perl $urn_bin" : "git") . " commit -m 'initial' > /dev/null 2>&1"); + + track_rebases() if $tool_name eq "URN"; my @file_list; - find(sub { - push @file_list, $File::Find::name if -f $_ && $File::Find::name !~ m{\.(git|urn)}; - }, "."); - - my $count = scalar(@file_list); - die "FATAL: No files found in $bm_repo after seeding!" if $count == 0; - - # Use explicit add + commit to ensure untracked files are caught - my $initial_cmd = ($tool_name eq "URN") ? "perl $urn_bin add . && perl $urn_bin commit -m 'initial'" - : "git add . && git commit -m 'initial'"; - system("$initial_cmd > /dev/null 2>&1"); + find(sub { push @file_list, $File::Find::name if -f $_ && $File::Find::name !~ /\.(git|urn)/ }, "."); for my $i (1 .. $total_commits) { my $to_modify = int($files * 0.02) || 1; for (1 .. $to_modify) { my $target = $file_list[rand @file_list]; if (open(my $fh, '>>', $target)) { - print $fh "Commit $i: Mod for $tool_name test.\n"; + print $fh "Churn $i\n"; close($fh); } } - my $msg = "commit_$i"; - my $run_cmd = ($tool_name eq "URN") ? "perl $urn_bin add . && perl $urn_bin commit -m '$msg'" - : "git add . && git commit -m '$msg'"; - + my $cmd = ($tool_name eq "URN") + ? "perl $urn_bin add . && perl $urn_bin commit -m 'c$i'" + : "git add . && git commit -m 'c$i'"; + if ($i % $sample_rate == 0 || $i == $total_commits) { - # Run add and commit together inside the time wrapper - my $raw = `(/usr/bin/time -l sh -c "$run_cmd" > /dev/null) 2>&1`; + # Capture hardware metrics via /usr/bin/time -l + my $stats = `/usr/bin/time -l sh -c "$cmd" 2>&1 > /dev/null`; - my ($real) = $raw =~ /(\d+\.\d+)\s+real/; - my ($rss) = $raw =~ /(\d+)\s+maximum resident set size/; - my ($maj) = $raw =~ /(\d+)\s+page faults caused by physical I\/O/; - my ($min) = $raw =~ /(\d+)\s+page reclaims by virtual memory/; + my ($real, $rss, $maj, $min) = (0, 0, 0, 0); + if ($stats =~ /(\d+\.\d+)\s+real/) { $real = $1; } + if ($stats =~ /(\d+)\s+maximum resident set size/) { $rss = sprintf("%.2f MB", $1 / 1024 / 1024); } + if ($stats =~ /(\d+)\s+page reclaims/) { $min = $1; } + if ($stats =~ /(\d+)\s+page faults/) { $maj = $1; } + + if ($tool_name eq "URN") { track_rebases(); } my $meta = ($tool_name eq "URN") ? ".urn" : ".git"; $results{$i}{$tool_name} = { - real => $real // "0.00", - rss => $rss ? sprintf("%.2f MB", $rss / 1024 / 1024) : "0.00 MB", - faults => sprintf("Maj:%d/Min:%d", $maj // 0, $min // 0), + real => $real . "s", + rss => $rss, + faults => "Maj:$maj / Min:$min", inodes => count_inodes($meta), size => get_size($meta), }; print " [Commit $i] $tool_name sampled.\n"; } else { - system("$run_cmd > /dev/null 2>&1"); + system("$cmd > /dev/null 2>&1"); + if ($tool_name eq "URN") { track_rebases(); } } } chdir($base_dir); + remove_tree($bm_repo); } +$rebase_count = 0; +%last_base_hashes = (); run_history_benchmark("URN"); run_history_benchmark("GIT"); @@ -116,15 +137,16 @@ foreach my $i (sort { $a <=> $b } keys %results) { print $res "-------------------------------------------------------------\n"; printf $res "%-15s | %-20s | %-20s\n", "METRIC", "URN", "GIT"; print $res "----------------+----------------------+---------------------\n"; - printf $res "%-15s | %20s | %20s\n", "Time", $u->{real}."s", $g->{real}."s"; + printf $res "%-15s | %20s | %20s\n", "Time", $u->{real}, $g->{real}; printf $res "%-15s | %20s | %20s\n", "Max RSS", $u->{rss}, $g->{rss}; - printf $res "%-15s | %20s | %20s\n", "Page Faults", $u->{faults}, $g->{faults}; + printf $res "%-15s | %20s | %20s\n", "Page faults", $u->{faults}, $g->{faults}; printf $res "%-15s | %20s | %20s\n", "Inodes", $u->{inodes}, $g->{inodes}; - printf $res "%-15s | %20s | %20s\n", "Repo Size", $u->{size}, $g->{size}; + printf $res "%-15s | %20s | %20s\n", "Repo size", $u->{size}, $g->{size}; print $res "-------------------------------------------------------------\n\n"; } +print $res "TOTAL URN REBASES: $rebase_count\n"; close($res); -print "\nRun complete! Opening results...\n"; -system($ENV{PAGER} || 'less', $out_file); +my $pager = $ENV{PAGER} || 'less'; +exec $pager, $out_file; |
