Return-Path: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=subject:to:cc:references:from:message-id:date:user-agent :mime-version:in-reply-to:content-language:content-transfer-encoding; bh=UJgnnL0In/W0g1lO+tzVe29cBGVbDK6odjX4N8JbDos=; b=q6klcZsrGH9+LllLgvaH16q+dG+VvtYz4VmU/Q2Q5gvDLjRPMUyUb5oZpgvc3786q3 Rpf2TfXw/87d1/dAu7IlzrtWnqj99WywAIVHUgtY4yxXSAtjbyBsbBLBRjThOZtVSthC Zq50YQWFyU1gtySHF2kn1FlL1da2gIXRVEbNdbn/Voie2PDosY/dtKnD8dmB4W8J7Mw/ +TepgVK3vBH80WtEct4OmCLHRs1YrFqKUxWu+GwCnTZKRO4DPOzxfcd4+dyzs/8AxjO+ yG6woRjnkV8Oxgwy4CHM0OsbJBJGoBusLbsD+Wqjf+LkbeZuJoefrRNcvngVlxKpLDiQ Zxeg== Subject: Re: [PATCH 2/2] CodeSamples/formal/herd/absperf-reduce: Enforce compare as number References: <7dea1e63-95bf-8d5b-dca2-47e741a5ed48@gmail.com> <8c68598d-847a-81c8-4612-66f957627601@gmail.com> <20200528005015.GC2869@paulmck-ThinkPad-P72> From: Akira Yokosawa Message-ID: <9d294bb4-ea3f-34d7-a8b6-0825e033b15f@gmail.com> Date: Thu, 28 May 2020 10:05:06 +0900 MIME-Version: 1.0 In-Reply-To: <20200528005015.GC2869@paulmck-ThinkPad-P72> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 7bit To: "Paul E. McKenney" Cc: perfbook@vger.kernel.org, Akira Yokosawa List-ID: On Wed, 27 May 2020 17:50:15 -0700, Paul E. McKenney wrote: > On Thu, May 28, 2020 at 07:44:11AM +0900, Akira Yokosawa wrote: >> >From 213bf072d15dfc3dce36b51a8bc1d35bd78e6609 Mon Sep 17 00:00:00 2001 >> From: Akira Yokosawa >> Date: Thu, 28 May 2020 07:25:04 +0900 >> Subject: [PATCH 2/2] CodeSamples/formal/herd/absperf-reduce: Enforce compare as number >> >> When there are differences in the number of digits in runtime, >> absperf-reduce.sh miscalculates min and max values and shows strange >> results such as: >> >> C-SB+l-o-o-u+l-o-o-u+l-o-o-u-XE.litmus 13.255 11.81 9.64 >> (avr) (min) (max) >> >> Fix this by adding 0 on both sides of comparison and enforce conversion >> to numbers. After this change: >> >> C-SB+l-o-o-u+l-o-o-u+l-o-o-u-XE.litmus 13.255 9.23 15.77 >> >> Signed-off-by: Akira Yokosawa > > Applied and pushed both, thank you! > > Did any of these runtime errors make their way into the tables in > the text? I don't think so. Tables 17.4 and E.5 present averages only. I spotted them while I did some absperf runs to compare herd7 7.52 and herd7 HEAD during discussions with Andrii. Thanks, Akira > > Thanx, Paul > >> --- >> CodeSamples/formal/herd/absperf-reduce.sh | 4 ++-- >> 1 file changed, 2 insertions(+), 2 deletions(-) >> >> diff --git a/CodeSamples/formal/herd/absperf-reduce.sh b/CodeSamples/formal/herd/absperf-reduce.sh >> index 0c338b8b..9cf52182 100644 >> --- a/CodeSamples/formal/herd/absperf-reduce.sh >> +++ b/CodeSamples/formal/herd/absperf-reduce.sh >> @@ -16,9 +16,9 @@ awk ' >> gsub(/user .*$/, "", curtesttime); >> testtime_n[curtest]++; >> testtime_sum[curtest] += curtesttime; >> - if (testtime_max[curtest] == "" || curtesttime > testtime_max[curtest]) >> + if (testtime_max[curtest] == "" || curtesttime + 0 > testtime_max[curtest] + 0) >> testtime_max[curtest] = curtesttime; >> - if (testtime_min[curtest] == "" || curtesttime < testtime_min[curtest]) >> + if (testtime_min[curtest] == "" || curtesttime + 0 < testtime_min[curtest] + 0) >> testtime_min[curtest] = curtesttime; >> } >> } >> -- >> 2.17.1 >> >>