singer@XN.LL.MIT.EDU (Matthew R. Singer) (03/12/88)
Before investing in microport's full 386 system, I have a few questions
if anyone cares to answer:
1) Can the 386 system compile for a 286 machine (Large model)?
If not, since I need the system for product development,
can V/AT and V/386 reside on the same machine?
2) Does the 386 system implement shared text?
3) I have a 386, with an XT class hard drive for MS-DOS. I want
to add a large AT class drive for Unix. Can I leave the old
disk in place for running MS-DOS? Can I boot dos from one
and unix from the other? Can I even boot dos or do you
have to use dosmerge?
4) Does the 386 system page? Or just swap...
Danke...
Matt Singerewv@violet.berkeley.edu (Eric Varsanyi) (03/12/88)
In article <929@xn.LL.MIT.EDU> singer@XN.LL.MIT.EDU (Matthew R. Singer) writes: >1) Can the 386 system compile for a 286 machine (Large model)? No, the supplied compiler cannot produces 286 executables. The system does seem to be able to run COFF 286 executables. > If not, since I need the system for product development, > can V/AT and V/386 reside on the same machine? I've heard gripes here that people cannot get the 286 product up on a 386, theoretically it should work, the 386 is downward compatible. >2) Does the 386 system implement shared text? Yes, shared libraries are available. Libc and the networking libraries come with the system as shared libraries. (Non shared versions are also included). >3) I have a 386, with an XT class hard drive for MS-DOS. I want > to add a large AT class drive for Unix. Can I leave the old > disk in place for running MS-DOS? Can I boot dos from one > and unix from the other? Can I even boot dos or do you > have to use dosmerge? Be carefull booting between DOS and UNIX, DOS doesn't understand the intel VTOC and will probably trash it. Best bet is to boot Unix off the HD and boot DOS off the floppy drive (then go to D: or where ever your hard disk is at). >4) Does the 386 system page? Or just swap... It pages very nicely, a sar -p shows paging activity. I would suggest getting 4 Meg of real storage, 2M just doesn't do the job. The low 1 meg seems to be taken by the kernel and memory mapped I/O, the above-the-line memory (1M in a 2M system) is split between the buffer cache (default 100K) and free pages. When running one compile, my system pages at about 13/sec. ---------------------------------------------------------------------------- Eric Varsanyi ewv@violet.berkeley.edu !ucbvax!violet!ewv Any opinions expressed are mine and are not necessarily those of Cray ----------------------------------------------------------------------------
louis@auvax.UUCP (Louis Schmittroth) (03/13/88)
In article <7602@agate.BERKELEY.EDU|, ewv@violet.berkeley.edu (Eric Varsanyi) writes: | In article <929@xn.LL.MIT.EDU> singer@XN.LL.MIT.EDU (Matthew R. Singer) writes: | >4) Does the 386 system page? Or just swap... | It pages very nicely, a sar -p shows paging activity. I would suggest getting | 4 Meg of real storage, 2M just doesn't do the job. The low 1 meg seems to | be taken by the kernel and memory mapped I/O, the above-the-line memory | (1M in a 2M system) is split between the buffer cache (default 100K) and | free pages. When running one compile, my system pages at about 13/sec. | ---------------------------------------------------------------------------- | Eric Varsanyi ewv@violet.berkeley.edu | !ucbvax!violet!ewv | Any opinions expressed are mine and are not necessarily those of Cray | ---------------------------------------------------------------------------- I have a chance to buy a very nice Made in Canada 386 machine which comes with 2MB of memeory using 256kb chips on the motherboard. The only other option is 8MB using 1mb chips, but this would cost me $2000 more. Will uport 386 run OK with 1 or 2 users with only 2MB memory? My strategy would be to wait until 1mb chips come down in price and install them later. -- Louis Schmittroth My employer has no opinions. Computer Science Athabasca University ...{ubc-vision, ihnp4}!alberta!auvax!louis
wnp@killer.UUCP (Wolf Paul) (03/14/88)
In article <7602@agate.BERKELEY.EDU> ewv@violet.berkeley.edu (Eric Varsanyi) writes: >In article <929@xn.LL.MIT.EDU> singer@XN.LL.MIT.EDU (Matthew R. Singer) writes: >>1) Can the 386 system compile for a 286 machine (Large model)? >No, the supplied compiler cannot produces 286 executables. The system does >seem to be able to run COFF 286 executables. > >> If not, since I need the system for product development, >> can V/AT and V/386 reside on the same machine? >I've heard gripes here that people cannot get the 286 product up on a 386, >theoretically it should work, the 386 is downward compatible. > Well, if the 386 compiler will not compile 286 code, but the 386 UNIX will run 286 COFF binaries, you might be able to run the 286 Development System under your 386 UNIX kernel. Of course you'd have to use not only the 286 compiler, assembler & linker, but the 286 libraries and include files as well. It should even be possible to set up a shell script which does a "chroot" to a separate "286 directory hierarchy" to automatically access a 286-specific /usr/lib/, /usr/include/, etc. Has anybody tried this? -- Wolf N. Paul * 3387 Sam Rayburn Run * Carrollton TX 75007 * (214) 306-9101 UUCP: ihnp4!killer!dcs!wnp ESL: 62832882 INTERNET: wnp@EESDES.DAS.NET or wnp@dcs.UUCP TLX: 910-280-0585 EES PLANO UD One Austrian's Opinion: Waldheim must go!
karl@ddsw1.UUCP (Karl Denninger) (03/16/88)
In article <3679@killer.UUCP> wnp@killer.UUCP (Wolf Paul) writes: $In article <7602@agate.BERKELEY.EDU> ewv@violet.berkeley.edu (Eric Varsanyi) writes: $>In article <929@xn.LL.MIT.EDU> singer@XN.LL.MIT.EDU (Matthew R. Singer) writes: $>>1) Can the 386 system compile for a 286 machine (Large model)? $>No, the supplied compiler cannot produces 286 executables. The system does $>seem to be able to run COFF 286 executables. WARNING: These '286 executables must be READABLE to execute, not just executable. Scratch any idea of security for your compiled modules. $>> If not, since I need the system for product development, $>> can V/AT and V/386 reside on the same machine? $>I've heard gripes here that people cannot get the 286 product up on a 386, $>theoretically it should work, the 386 is downward compatible. The '286 product runs on our '386 systems as does the '386 product (we sell Televideo). $Well, if the 386 compiler will not compile 286 code, but the 386 UNIX will run $286 COFF binaries, you might be able to run the 286 Development System under $your 386 UNIX kernel. Of course you'd have to use not only the 286 compiler, $assembler & linker, but the 286 libraries and include files as well. $ $It should even be possible to set up a shell script which does a "chroot" $to a separate "286 directory hierarchy" to automatically access a $286-specific /usr/lib/, /usr/include/, etc. This works... we use it here for cross-development. The 'installit' program on the '386 can and will read '286 format installit disks; this is how we finally got '286 compilation capability on our Microport/386 system. Just don't forget that you need the shared libraries in the chrooted area, or nothing will work at all! :-) Two notes about this: 1) '286 SDB will not work on the 386 runtime. Thus you cannot DEBUG 80286 code on the '386 (yuck!), for this, you need a '286 as well. Core dumps in a '286 COFF binary are a *bitch* to find without a workable debugger! 2) The 'chrooted' area must have all the directories and files which are normally required for operation. If you put your chrooted area on the same filesystem as your primary operating files, you can at least use 'ln' and save significant amounts of space. This requires that you have both licenses, too -- which makes quite a dent in the pocketbook..... ------ Karl Denninger | Data: +1 312 566-8912 Macro Computer Solutions, Inc. | Voice: +1 312 566-8910 ...ihnp4!ddsw1!karl | "Quality solutions for work or play"
neighorn@catlabs.UUCP (Steven C. Neighorn) (04/17/88)
In article <7602@agate.BERKELEY.EDU> ewv@violet.berkeley.edu (Eric Varsanyi) writes: >In article <929@xn.LL.MIT.EDU> singer@XN.LL.MIT.EDU (Matthew R. Singer) writes: >>1) Can the 386 system compile for a 286 machine (Large model)? >No, the supplied compiler cannot produces 286 executables. The system does >seem to be able to run COFF 286 executables. > With the following expections from my 'running 286 binaries on V/386' days: 1) Unify's Microport 286 version of Accell (The development environment). Most of the RDBMS stuff worked Ok, but trying anything fancy with Accell resulted in a 500k(!) core file. 2) A ported version of Sendmail 5.51. "ioctl" problems here, and the venerable 'not a typewriter' error we have all grown to love and cherish. 3) The Dataflex DB compiler. The rest of Dataflex works fine, but we have to keep a uport 286 system hanging around to compile new Dataflex stuff. #1 and #2 were fixed by getting a new version/recompiling. #3 is still a sore spot. -- Steven C. Neighorn ...!tektronix!{psu-cs,reed,ogcvax}!qiclab!catlabs!neighorn Portland Public Schools "Where we train young Star Fighters to defend the (503) 249-2000 ext 337 frontier against Xur and the Ko-dan Armada"