From: Frantisek Hanzlik <franta@hanzlici.cz>
To: Eddy Setyawan <eddymilis@teguhabadi.com>
Cc: linux-msdos@vger.kernel.org
Subject: Re: Dosemu Slower than CMD
Date: Fri, 30 May 2008 05:53:56 +0200 [thread overview]
Message-ID: <483F7A54.5050403@hanzlici.cz> (raw)
In-Reply-To: <34cfb2e90805291019r1f2db5dew81fac5e6f6184f76@mail.gmail.com>
Hello,
I download Your app (mitra.tar.bz2: sl.exe etc) and configure and run it
as You reccomended in mail from 16-Mar-2008, with config.sys param
"buffers=10" (leaved at dosemu default value - as I/O is cached at Linux
side, I not see any benefit from increasing number of buffers.
And available DOS memory was 630kB for "buffers=1", and only 578 kB for
"buffers=100". Difference in app speed wasn't perceivable.
At PC with 5 years old P4 2.4GHz (Northwood) CPU/1GB RAM/320GB SATA HDD
and dosemu 1.4.0-svn_1844 it take about 6 sec.
At PC with E8200 Core2 Duo 2.66 GHz (Wolfdale) CPU/8GB RAM/750GB SATA HDD
and dosemu 1.4.0.1-4.svn1858 it take about 3 sec.
On both PCs run Fedora 9.
Then I run it with some dosemu debug options (+9#DTEMx - note: this
produce ~ 200MB log file and dosemu must be patched for it, as it log
circular log with limited size - see "src/include/dosemu_debug.h" and
"src/base/misc/utilities.c").
After some logfile analysis I got these results about called functions:
How many
times
called What
======= =========
196624 : int16_0x11-kbd-Check_Enhanced_Keystroke
623 : int21_2ac0-GetDate
623 : int21_2cc0-GetTime
70707 : int21_3f00-ReadFromFile
20035 bx=0x000a (20027x read 12.Byte, 8x read 1024.Byte)
5577 bx=0x000c read 339. bytes
5726 bx=0x000d (5577x read 12.Byte, 149x read 1024.Byte)
36252 bx=0x0009 read 180. Byte
1870 bx=0x0015 read 144. Byte
1247 bx=0x0016 read 12. Byte
70707 : int21_4200-SeekFromStartFile
26850 : int21_5c00-RecordLock
26851 : int21_5c01-RecordUnlock
53701 : int2f_110a-network_lock/unlock
70707 : int2f_1108-network_read-from-file
Maybe this program section is slow because it has many read attempts,
each of them also do seek in file and lock/unlock functions. And because
it is on Linux native partition, these I/O go over INT2F network redirector
too. Tried You run this app under Windows from remote share?
Eddy Setyawan wrote:
> hello Frantisek Hanzlik,
>
> On Sun, Mar 16, 2008 at 3:41 AM, Frantisek Hanzlik<franta@hanzlici.cz> wrote:
>> - When is section of Your program slow, it does any CPU-intensive task,
>> or rather I/O intensive?
>
> it run CPU to 100% task, i've ask the programmer who code it, and he
> say that process is do so many task but in CMD it run fast even i look
> at task manager CPU usage jump to 100% but only less than 1sec using
> dosemu it takes more than 10secs
>
>> - Your DOS files are on native DOS partition, or on Linux filesystem?
>
> it is Linux filesystem... i need run on it for sharing each user run it
>
>> - application is slow always, or only sometimes?
>
> only sometimes
Only sometimes? It's possible somehow simulate this?
>
>> - Is slow only when there runs multiple those clipper apps, or is
>
> I try run only 1 apps it run slow
>
>> slow when run only one too?
>> - You run DOSEMU under console or X Window?
>
> run in console bcause my plan is run it remotely with ssh
>
>> - Did You run any diagnostic tools, e.g. "top", "sar",...?
>
> i don't know what is sar, but using "top" show that dosemu process eat
> CPU to 100%
At these short interval it maybe OK.
>
>> Franta Hanzlik
>>
>> Eddy Setyawan wrote:
>>> Hello DosEmu Mailing list,
>>>
>>> I am new here... i have a problem i hope guys here can assist me solve it
>>>
>>> i try to run application made using clipper when run in w*nd*ws cmd
>>> command, and do some process it is a little bit slower user can feel
>>> it have a delay but not more that 1 secs, but when i run it in dosemu,
>>> it take more than 5 secs i don't know where this problem cause,
>>> because the application is made by other person
>>> any suggestion? or perhaps do i need to attach the application so you
>>> guys can try on ?
>>>
>>> thanks
>>>
>>> NB:
>>> w*nd*ws is run on Pentium 3 733 machine
>>> dosemu (v. 1.4.0) is run on Linux Fedora 8 on AMD 64 3000+ machine
>>>
>>> regards,
>>> Eddy Setyawan
>>> --
Franta Hanzlik
next prev parent reply other threads:[~2008-05-30 3:53 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-03-15 16:47 Dosemu Slower than CMD Eddy Setyawan
2008-03-15 19:54 ` Reinhard Karcher
2008-03-16 2:22 ` Eddy Setyawan
2008-03-16 7:23 ` Rafał Cygnarowski
2008-03-16 10:41 ` Frantisek Hanzlik
2008-03-16 13:35 ` Eddy Setyawan
2008-03-16 13:43 ` Eddy Setyawan
2008-04-01 18:33 ` Mike McCarty
2008-05-29 17:19 ` Eddy Setyawan
2008-05-30 3:53 ` Frantisek Hanzlik [this message]
2008-05-30 8:27 ` Frantisek Hanzlik
2008-05-30 8:40 ` Eddy Setyawan
2008-05-30 8:56 ` Frantisek Hanzlik
[not found] ` <34cfb2e90805300203t1f904938q60405bbbe3e035a2@mail.gmail.com>
2008-05-31 8:15 ` Frantisek Hanzlik
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=483F7A54.5050403@hanzlici.cz \
--to=franta@hanzlici.cz \
--cc=eddymilis@teguhabadi.com \
--cc=linux-msdos@vger.kernel.org \
/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
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox