[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Performances comparission between 5.1 and 5.2.
|
From: |
Alex fxmbsw7 Ratchev |
|
Subject: |
Re: Performances comparission between 5.1 and 5.2. |
|
Date: |
Mon, 15 Aug 2022 20:28:12 +0200 |
catastrophal numbers, ..
On Mon, Aug 15, 2022, 19:06 felix <felix@f-hauri.ch> wrote:
> Configuration Information:
> Machine: x86_64
> OS: linux-gnu
> Compiler: gcc
> Compilation CFLAGS: -g -O2
> uname output: Linux medium 5.10.0-12-amd64 #1 SMP Debian 5.10.103-1
> (2022-03-07) x86_64 GNU/Linux
> Machine Type: x86_64-pc-linux-gnu
>
> Bash Version: 5.2
> Patch Level: 0
> Release Status: rc2
>
> Description:
> Trying some script under 5.2 beta, rc1 and rc2, I was surprised by
> execution time.
> In order to compare, I've tried to establish execution time of
> elementar operations, like:
>
> - {0..999} .`: {0..999}`
> - 3x{0..9} `: {0..9}{0..9}{0..9}`
> - readUt `read -r ut idl </proc/uptime`
> - RE `[[ $cnt =~ ^[0-9]*\..*$ ]]`
> - InLneStr `: <<<"Hello world!"`
> - IncInt `int+=1`
> - AccessRt `: "$EPOCHREALTIME"`
> - PrintF `printf -v int '%s+%s' $int 1`
>
> Here is results of my comparission between different version of
> bash:
> (1st column show number of empty loop in 0.2 seconds as reference)
> $ for((i=0;i<${#bashs[@]};i++)){ ${bashs[i]} timedTest.sh
> ${args[i]:--q};}
> l/.2s {0..999} 3x{0..9} readUt RE
> InLneStr IncInt AccessRt PrintF Version BuildDir
> 66222 330.31 338.28 13.17 19.02
> 63.16 0.08 2.69 1.18 5.0.17(1)-release
> /tmp/bash/bash-5.0/bash
> 73720 366.97 379.70 13.20 17.80
> 10.09 0.10 2.46 1.19 5.1.4(1)-release /bin/bash
> 60585 409.92 419.53 13.62 19.87
> 10.87 0.09 3.00 1.38 5.1.4(1)-release
> /tmp/bash/bash-5.1.4/bash
> 62416 417.85 423.23 14.63 20.84
> 11.34 0.36 3.05 1.43 5.1.4(1)-release
> /tmp/bash/bash-5.1.4/bash_dynlib
> 62538 416.97 439.28 13.65 20.69
> 12.51 0.29 3.05 1.57 5.1.4(1)-release
> /tmp/bash/debian/bash-5.1/bash
> 60664 419.54 425.89 14.25 21.05
> 11.01 0.57 3.12 1.38 5.1.16(1)-release
> /tmp/bash/bash-5.1.16/bash
> 40808 39995.10 39995.10 18.49 35.46
> 15.65 0.34 5.81 2.40 5.2.0(1)-beta
> /tmp/bash/bash-5.2-beta/bash
> 40134 33328.35 39995.02 18.88 31.18
> 14.73 0.29 6.19 2.20 5.2.0(1)-rc1
> /tmp/bash/bash-5.2-rc1/bash
> 40350 39995.04 39995.04 19.10 32.20
> 14.66 0.59 6.15 2.26 5.2.0(1)-rc2
> /tmp/bash/bash-5.2-rc2/bash
>
> Where if everything seem slower, accessing $EPOCHREALTIME like
> `prinf -v INTEGER '%s+%s' $INTEGER 1` or
> using Regular Expression use approx 2x more time, but a sequence
> of 1000 using brace expansion require 100x more time!!
>
> Another test (first script I wrote) is something slower, but
> render approx same results:
> $ for((i=0;i<${#bashs[@]};i++)){ ${bashs[i]} looptest.sh
> ${args[i]:--q};}
> l/.02s {0..999} 3x{0..9} readUt RE
> InLneStr IncInt AccessRt PrintF Version BuildDir
> 1468 340.47 348.36 16.71 22.54
> 51.44 1.71 5.10 3.51 5.0.17(1)-release bash-5.0/bash
> 1714 366.64 380.21 15.11 20.11
> 12.23 1.33 3.99 2.35 5.1.4(1)-release /bin/bash
> 1074 414.53 423.91 15.91 22.94
> 12.78 1.72 4.77 2.54 5.1.4(1)-release
> bash-5.1.4/bash
> 1444 413.59 418.95 16.80 22.83
> 13.21 1.70 4.71 2.51 5.1.4(1)-release
> bash-5.1.4/bash_dynlib
> 1360 421.20 432.83 16.33 22.79
> 12.98 1.82 4.87 2.59 5.1.4(1)-release
> debian/bash-5.1/bash
> 1134 420.00 428.62 16.76 22.74
> 13.12 1.78 4.92 2.60 5.1.16(1)-release
> bash-5.1.16/bash
> 687 40158.21 40676.12 23.73 38.81
> 18.98 3.96 9.87 3.94 5.2.0(1)-beta
> bash-5.2-beta/bash
> 607 39559.71 40192.35 23.38 37.68
> 19.08 3.95 9.98 3.99 5.2.0(1)-rc1
> bash-5.2-rc1/bash
> 740 40531.76 40965.16 24.38 37.99
> 19.13 3.94 10.05 3.95 5.2.0(1)-rc2
> bash-5.2-rc2/bash
>
> Repeat-By:
> https://f-hauri.ch/vrac/timedTest.sh.txt
> https://f-hauri.ch/vrac/looptest.sh.txt
>
> --
> Félix Hauri - <felix@f-hauri.ch> - http://www.f-hauri.ch
>
>