----- Begin Included Message -----
>From [email protected] (Kay Kasemir) Thu Jun 6 15:44:44 1996
X-Sender: [email protected]
X-Mailer: Windows Eudora Light Version 1.5.4 (32)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Thu, 06 Jun 1996 15:44:41 -0600
To: [email protected]
From: Kay-Uwe Kasemir <[email protected]>
Subject: PC-port status
Content-Length: 4764
>Current status of the EPICS -> PC port
>======================================
>
>The first goal was to port the IOC (target) portion of EPICS
>onto a PC, using VxWorks 5.3 Tornado on a Pentium PC running NT4.0
>as the development platform.
>
>While some parts of the EPICS _host_ software are already ported to NT,
>this was done using a MS C compiler with specially build new
>Makefiles ("Project-Files"?). Because of the huge number of Makefiles &
>Shell-scripts used to build EPICS, I tried to use them instead of being
>forced to write new project files.
>
>While Tornado comes whith GNU make + bash (Born again shell), it turned out
>that Tornado-make calls $(SHELL) with "/C", which is ok for the default
>SHELL=cmd.exe but prevents using SHELL=bash.
>
>Fortunately I found a rather complete GNU-suite at Cygnus
>(http://www.cygnus.com/misc/gnu-win32) containing gcc, bash, make,
>bison, ....
>The additional benefit was the included gcc, the fact that the Cygnus-bash even
>emulates soft-links ("ln -s") that many Makefiles use, and a partial
>complete clib:
>
> "The second part of the project (which we are currently in) is developing a
> C library that is enough for GNU programs (fileutils, sh-utils, textutils,
> bash, make) to be compiled on a Win32 system without any modification
> to those programs. From their point of view, a Windows NT box is no
> different than any other Unix box."
>
>This clib already allows you to call socket(), bind(), gethostbyname(), fork(),
>getpid(), ... so I was not only able to use the current Makefiles & scripts,
>in addition I could also build e.g. antelope and the other small tools that
>are needed on the host to compile the target portion of EPICS.
>
>Basically I only needed to provide new config files CONFIG.Vx.pc486,
>CONFIG.Unix.NT
>(because with Cygnus-gnu makes NT look like a flavor of Unix) and change
>all executables' filenames in the Makefiles from "foo" to "foo$(EXE)" with
>EXE=.exe (Cygnus gnu also uses .exe-postfixes for executables).
>
>
>The current state is that I managed to
>1) compile the target portion of EPICS
>2) load a simple database onto a PC 100MHz Pentium target
>
>It's like the "makeExampleApp" database:
>
> calc ai
> +------------------------+ +-------------------+
> |SCAN: 1 second | | |
> |CALC:(A<B)?(A+C):D VAL|---->-------| INP VAL |---\
> | | | | |
> /->--|INPA | | | |
> | 9-|INPB | | | |
> | 1-|INPC | | | |
> | 0-|INPD | | | |
> | +------------------------+ +-------------------+ |
> | |
> \-----------------------------------<------------------------------/
>
>(This is not a new form of ASCII-graphical-dct but merely an illustration)
>The records are scanned correctly as I can tell so far, the calc record is
>ramping from 0 to 9.
>
>dbt results for the calc record (100 executions):
>
>100MHz Pentium 3550 +/- 74
>NITRO60 CPU 3625 +/- 74
>
>(The NITRO60 CPU is running the "older" EPICS R3.12)
>
>This result suggests that the PC target is at least as fast as a speedy
>VME-based CPU.
>You should also bear in mind that most 100MHz Pentium boards can also be
>equipped with
>a 120MHz, 150MHz,... CPU.
>
>Channel access is *NOT* working so far because of several differences in the
>byte order of the CPU 68k and Ix86 architectures. We are working on this.
>
>---------------------------------------------------------------------
>[email protected]
>Kay-Uwe Kasemir AOT-8, MS H820
> Los Alamos National Laboratory
>Tel. (505) 665 8891 Los Alamos, NM 87545
---------------------------------------------------------------------
[email protected]
Kay-Uwe Kasemir AOT-8, MS H820
Los Alamos National Laboratory
Tel. (505) 665 8891 Los Alamos, NM 87545
'-|-`
______| |______
(______\_/______)
(_\/_\_(X)_/_\/_)
O'=`O
\ o
=[
d
----- End Included Message -----
- Navigate by Date:
- Prev:
Software Engineering Position SIBLEY
- Next:
DBR_STRING get for DBF_DOUBLE (Possible Problem) Marty Kraimer
- Index:
1994
1995
<1996>
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
- Navigate by Thread:
- Prev:
Software Engineering Position SIBLEY
- Next:
DBR_STRING get for DBF_DOUBLE (Possible Problem) Marty Kraimer
- Index:
1994
1995
<1996>
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
|