git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jeff King <peff@peff.net>
To: git@vger.kernel.org
Subject: [PATCH 3/6] t/perf: add infrastructure for measuring sizes
Date: Tue, 21 Aug 2018 15:06:54 -0400	[thread overview]
Message-ID: <20180821190654.GC30764@sigill.intra.peff.net> (raw)
In-Reply-To: <20180821184140.GA24165@sigill.intra.peff.net>

The main objective of scripts in the perf framework is to
run "test_perf", which measures the time it takes to run
some operation. However, it can also be interesting to see
the change in the output size of certain operations.

This patch introduces test_size, which records a single
numeric output from the test and shows it in the aggregated
output (with pretty printing and relative size comparison).

Signed-off-by: Jeff King <peff@peff.net>
---
 t/perf/README         | 25 ++++++++++++++++++++++
 t/perf/aggregate.perl | 48 ++++++++++++++++++++++++++++++++++++++-----
 t/perf/perf-lib.sh    | 13 ++++++++++++
 3 files changed, 81 insertions(+), 5 deletions(-)

diff --git a/t/perf/README b/t/perf/README
index 21321a0f36..be12090c38 100644
--- a/t/perf/README
+++ b/t/perf/README
@@ -168,3 +168,28 @@ that
   While we have tried to make sure that it can cope with embedded
   whitespace and other special characters, it will not work with
   multi-line data.
+
+Rather than tracking the performance by run-time as `test_perf` does, you
+may also track output size by using `test_size`. The stdout of the
+function should be a single numeric value, which will be captured and
+shown in the aggregated output. For example:
+
+	test_perf 'time foo' '
+		./foo >foo.out
+	'
+
+	test_size 'output size'
+		wc -c <foo.out
+	'
+
+might produce output like:
+
+	Test                origin           HEAD
+	-------------------------------------------------------------
+	1234.1 time foo     0.37(0.79+0.02)  0.26(0.51+0.02) -29.7%
+	1234.2 output size             4.3M             3.6M -14.7%
+
+The item being measured (and its units) is up to the test; the context
+and the test title should make it clear to the user whether bigger or
+smaller numbers are better. Unlike test_perf, the test code will only be
+run once, since output sizes tend to be more deterministic than timings.
diff --git a/t/perf/aggregate.perl b/t/perf/aggregate.perl
index 3181b087ab..494907a892 100755
--- a/t/perf/aggregate.perl
+++ b/t/perf/aggregate.perl
@@ -13,10 +13,16 @@ sub get_times {
 	my $line = <$fh>;
 	return undef if not defined $line;
 	close $fh or die "cannot close $name: $!";
-	$line =~ /^(?:(\d+):)?(\d+):(\d+(?:\.\d+)?) (\d+(?:\.\d+)?) (\d+(?:\.\d+)?)$/
-		or die "bad input line: $line";
-	my $rt = ((defined $1 ? $1 : 0.0)*60+$2)*60+$3;
-	return ($rt, $4, $5);
+	# times
+	if ($line =~ /^(?:(\d+):)?(\d+):(\d+(?:\.\d+)?) (\d+(?:\.\d+)?) (\d+(?:\.\d+)?)$/) {
+		my $rt = ((defined $1 ? $1 : 0.0)*60+$2)*60+$3;
+		return ($rt, $4, $5);
+	# size
+	} elsif ($line =~ /^\d+$/) {
+		return $&;
+	} else {
+		die "bad input line: $line";
+	}
 }
 
 sub relative_change {
@@ -32,9 +38,15 @@ sub relative_change {
 
 sub format_times {
 	my ($r, $u, $s, $firstr) = @_;
+	# no value means we did not finish the test
 	if (!defined $r) {
 		return "<missing>";
 	}
+	# a single value means we have a size, not times
+	if (!defined $u) {
+		return format_size($r, $firstr);
+	}
+	# otherwise, we have real/user/system times
 	my $out = sprintf "%.2f(%.2f+%.2f)", $r, $u, $s;
 	$out .= ' ' . relative_change($r, $firstr) if defined $firstr;
 	return $out;
@@ -54,6 +66,25 @@ sub usage {
 	exit(1);
 }
 
+sub human_size {
+	my $n = shift;
+	my @units = ('', qw(K M G));
+	while ($n > 900 && @units > 1) {
+		$n /= 1000;
+		shift @units;
+	}
+	return $n unless length $units[0];
+	return sprintf '%.1f%s', $n, $units[0];
+}
+
+sub format_size {
+	my ($size, $first) = @_;
+	# match the width of a time: 0.00(0.00+0.00)
+	my $out = sprintf '%15s', human_size($size);
+	$out .= ' ' . relative_change($size, $first) if defined $first;
+	return $out;
+}
+
 my (@dirs, %dirnames, %dirabbrevs, %prefixes, @tests,
     $codespeed, $sortby, $subsection, $reponame);
 
@@ -184,7 +215,14 @@ sub print_default_results {
 		my $firstr;
 		for my $i (0..$#dirs) {
 			my $d = $dirs[$i];
-			$times{$prefixes{$d}.$t} = [get_times("$resultsdir/$prefixes{$d}$t.times")];
+			my $base = "$resultsdir/$prefixes{$d}$t";
+			$times{$prefixes{$d}.$t} = [];
+			foreach my $type (qw(times size)) {
+				if (-e "$base.$type") {
+					$times{$prefixes{$d}.$t} = [get_times("$base.$type")];
+					last;
+				}
+			}
 			my ($r,$u,$s) = @{$times{$prefixes{$d}.$t}};
 			my $w = length format_times($r,$u,$s,$firstr);
 			$colwidth[$i] = $w if $w > $colwidth[$i];
diff --git a/t/perf/perf-lib.sh b/t/perf/perf-lib.sh
index a54be09516..11d1922cf5 100644
--- a/t/perf/perf-lib.sh
+++ b/t/perf/perf-lib.sh
@@ -231,6 +231,19 @@ test_perf () {
 	test_wrapper_ test_perf_ "$@"
 }
 
+test_size_ () {
+	say >&3 "running: $2"
+	if test_eval_ "$2" 3>"$base".size; then
+		test_ok_ "$1"
+	else
+		test_failure_ "$@"
+	fi
+}
+
+test_size () {
+	test_wrapper_ test_size_ "$@"
+}
+
 # We extend test_done to print timings at the end (./run disables this
 # and does it after running everything)
 test_at_end_hook_ () {
-- 
2.19.0.rc0.398.g138a08f6f6


  parent reply	other threads:[~2018-08-21 19:06 UTC|newest]

Thread overview: 55+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-08-21 18:41 [PATCH] test-tool.h: include git-compat-util.h Jeff King
2018-08-21 19:03 ` Junio C Hamano
2018-08-21 19:06 ` [PATCH 1/6] t/perf: factor boilerplate out of test_perf Jeff King
2018-08-21 19:06 ` [PATCH 2/6] t/perf: factor out percent calculations Jeff King
2018-08-21 19:06 ` Jeff King [this message]
2018-08-22 13:40   ` [PATCH 3/6] t/perf: add infrastructure for measuring sizes Derrick Stolee
2018-08-22 15:31     ` Jeff King
2018-08-21 19:06 ` [PATCH 4/6] t/perf: add perf tests for fetches from a bitmapped server Jeff King
2018-08-21 19:07 ` [PATCH 5/6] pack-bitmap: save "have" bitmap from walk Jeff King
2018-08-21 19:47   ` Derrick Stolee
2018-08-21 19:54     ` Jeff King
2018-08-31 15:23   ` Ævar Arnfjörð Bjarmason
2018-08-31 22:55     ` Jeff King
2018-09-01  7:41       ` [PATCH 0/4] un-breaking pack-objects with bitmaps Jeff King
2018-09-01  7:44         ` [PATCH 1/4] bitmap_has_sha1_in_uninteresting(): drop BUG check Jeff King
2018-09-01  7:48         ` [PATCH 2/4] t5310: test delta reuse with bitmaps Jeff King
2018-09-01  8:03           ` Jeff King
2018-09-01 20:29             ` Ævar Arnfjörð Bjarmason
2018-09-01 22:46               ` Ben Peart
2018-09-02  5:51               ` Jeff King
2018-09-04 19:05             ` Stefan Beller
2018-09-04 19:45               ` Junio C Hamano
2018-09-04 20:02               ` Jeff King
2018-09-01  7:49         ` [PATCH 3/4] traverse_bitmap_commit_list(): don't free result Jeff King
2018-09-01  7:50         ` [PATCH 4/4] pack-bitmap: drop "loaded" flag Jeff King
2018-09-04 19:30         ` [PATCH 0/4] un-breaking pack-objects with bitmaps Stefan Beller
2018-09-04 20:03           ` Jeff King
2018-09-08  6:43         ` Ævar Arnfjörð Bjarmason
2018-09-10 16:53           ` Junio C Hamano
2018-09-10 18:48             ` Jeff King
2018-09-10 19:23               ` Junio C Hamano
2018-08-21 19:07 ` [PATCH 6/6] pack-objects: reuse on-disk deltas for thin "have" objects Jeff King
2018-08-21 19:43   ` Junio C Hamano
2018-08-21 19:50     ` Junio C Hamano
2018-08-21 20:07       ` Jeff King
2018-08-21 20:14         ` Jeff King
2018-08-21 20:52           ` Junio C Hamano
2018-08-21 21:30             ` Jeff King
2018-08-21 20:57         ` Junio C Hamano
2018-08-21 21:32           ` Jeff King
2018-08-23  0:43           ` [PATCH 0/9] trailer-parsing false positives Jeff King
2018-08-23  0:44             ` [PATCH 1/9] trailer: use size_t for string offsets Jeff King
2018-08-23  0:45             ` [PATCH 2/9] trailer: use size_t for iterating trailer list Jeff King
2018-08-23  0:46             ` [PATCH 3/9] trailer: pass process_trailer_opts to trailer_info_get() Jeff King
2018-08-23  0:48             ` [PATCH 4/9] interpret-trailers: tighten check for "---" patch boundary Jeff King
2018-08-23  0:49             ` [PATCH 5/9] interpret-trailers: allow suppressing "---" divider Jeff King
2018-08-23  0:50             ` [PATCH 6/9] pretty, ref-filter: format %(trailers) with no_divider option Jeff King
2018-08-23  0:50             ` [PATCH 7/9] sequencer: ignore "---" divider when parsing trailers Jeff King
2018-08-23  0:50             ` [PATCH 8/9] append_signoff: use size_t for string offsets Jeff King
2018-08-23  0:51             ` [PATCH 9/9] sequencer: handle ignore_footer when parsing trailers Jeff King
2018-08-23 18:30             ` [PATCH 0/9] trailer-parsing false positives Junio C Hamano
2018-08-24  7:26               ` Jeff King
2018-08-21 20:00     ` [PATCH 6/6] pack-objects: reuse on-disk deltas for thin "have" objects Jeff King
  -- strict thread matches above, loose matches on Subject: below --
2018-08-17 20:54 [PATCH 0/6] reuse on-disk deltas for fetches with bitmaps Jeff King
2018-08-17 20:56 ` [PATCH 3/6] t/perf: add infrastructure for measuring sizes Jeff King
2014-03-26  7:22 [PATCH/RFC 0/6] reuse deltas found by bitmaps Jeff King
2014-03-26  7:22 ` [PATCH 3/6] t/perf: add infrastructure for measuring sizes Jeff King

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20180821190654.GC30764@sigill.intra.peff.net \
    --to=peff@peff.net \
    --cc=git@vger.kernel.org \
    --subject='Re: [PATCH 3/6] t/perf: add infrastructure for measuring sizes' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).