|
|
|
|
| Author |
Message |
William R. Walsh
Guest
|
 Posted: Thu Feb 22, 2007 4:37 am Post subject: Compaq Deskpro 386s20/n |
 |
|
What can I say? I saved it from an untimely end at a computer recyclers.
It's completely alive but unhappy about the state of its CMOS battery.
Specifically, it asks me for a configuration disk. I can "press F1 to
continue" and it will boot, but...I'd like to configure it if I could.
Amazingly enough, HP/Compaq's web site had not forgotten about the existence
of this machine. They have forgotten about all the files for it, though.
Drive A is a 5.25" floppy, although that is the least of my problems. I have
no problem writing out 5.25" floppies, either HD or 360K.
Can anyone help spin the clock back on this one?
William |
|
| Back to top |
|
 |
Fix your Windows Problems - FAST.
FREE Safe Scan Registry Check. Locate & Fix Errors in Minutes!
|
|
paulmd@efn.org
Guest
|
| Posted: Mon Feb 26, 2007 1:16 pm Post subject: Re: Compaq Deskpro 386s20/n |
|
|
On Feb 21, 8:06 pm, "William R. Walsh"
<newsgrou...@idontwantjunqueemail.walshcomptech.com> wrote:
| Quote: | What can I say? I saved it from an untimely end at a computer recyclers.
|
Untimely? It's going on 20 years old.
| Quote: | It's completely alive but unhappy about the state of its CMOS battery.
|
No suprise there.
Do you want the bad news first?
386 machines often have a Dallas Realtime clock that holds it's own
charge for about 10 years. Meaning it's a chip you have to replace,
not just a battery. Dallas no longer makes that part, and it's
replacement is incompatible with 386s. Last i looked, you could find
the part for around $20, from various resellers.
If on the other hand, you are lucky enough to have an actual battery,
it's not the standard cr2032. It's a more hard do find model.
| Quote: | Specifically, it asks me for a configuration disk. I can "press F1 to
continue" and it will boot, but...I'd like to configure it if I could.
Amazingly enough, HP/Compaq's web site had not forgotten about the existence
of this machine. They have forgotten about all the files for it, though.
Drive A is a 5.25" floppy, although that is the least of my problems. I have
no problem writing out 5.25" floppies, either HD or 360K.
Can anyone help spin the clock back on this one?
William
|
To be honest, given the vintage of the machine, if all you have to do
to make it work is press f1, you're doing pretty good. |
|
| Back to top |
|
|
Ben Myers
Guest
|
| Posted: Mon Feb 26, 2007 11:14 pm Post subject: Re: Compaq Deskpro 386s20/n |
|
|
If the RTC is a Dallas chip, the model number is most likely DS12887 and it
looks about the size of a black lego. It can be removed carefully by wedging it
out with a small flat-bladed screwdriver. I keep a few around, pulled from 486
boards. If that's what is in a 386s20/n, send me a fiver and I'll send out a
battery.
The battery might instead be one of the older ones that attaches to a
motherboard header with two leads and is affixed to the inside of the case with
velcro. Sorry, don't have any of these... Ben Myers
On 25 Feb 2007 23:16:25 -0800, "paulmd@efn.org" <paulmd@efn.org> wrote:
| Quote: | On Feb 21, 8:06 pm, "William R. Walsh"
newsgrou...@idontwantjunqueemail.walshcomptech.com> wrote:
What can I say? I saved it from an untimely end at a computer recyclers.
Untimely? It's going on 20 years old.
It's completely alive but unhappy about the state of its CMOS battery.
No suprise there.
Do you want the bad news first?
386 machines often have a Dallas Realtime clock that holds it's own
charge for about 10 years. Meaning it's a chip you have to replace,
not just a battery. Dallas no longer makes that part, and it's
replacement is incompatible with 386s. Last i looked, you could find
the part for around $20, from various resellers.
If on the other hand, you are lucky enough to have an actual battery,
it's not the standard cr2032. It's a more hard do find model.
Specifically, it asks me for a configuration disk. I can "press F1 to
continue" and it will boot, but...I'd like to configure it if I could.
Amazingly enough, HP/Compaq's web site had not forgotten about the existence
of this machine. They have forgotten about all the files for it, though.
Drive A is a 5.25" floppy, although that is the least of my problems. I have
no problem writing out 5.25" floppies, either HD or 360K.
Can anyone help spin the clock back on this one?
William
To be honest, given the vintage of the machine, if all you have to do
to make it work is press f1, you're doing pretty good.
|
|
|
| Back to top |
|
|
William R. Walsh
Guest
|
| Posted: Mon Mar 05, 2007 3:16 am Post subject: Re: Compaq Deskpro 386s20/n |
|
|
Hi!
| Quote: | Untimely? It's going on 20 years old.
|
It runs and it works. What more do you want? I can think of useful things to
do with it, although I do realize that I'm probably in something of a
minority.
Actually, there was. :-) Someone had gone to the trouble of replacing it.
They cut into the original Compaq wiring and grafted a more standard battery
into the system.
| Quote: | 386 machines often have a Dallas Realtime clock that holds it's own
charge for about 10 years.
|
They really do quite better than that in real life. Ten years is the
lifetime of the memory without any external power. I have some of these that
are over 10 years old and still keeping excellent time. The same seems to be
true of their "bigger brother" the DS1387.
The DS12887 and DS12887+ (ROHS compliant) are still available and work fine
in every machine I ever had to put one in. I'm not sure where you got the
idea that these "don't work" in a 386. I've put new parts in 286, 386SX and
386 systems and never had a problem. Quite a few pin compatible replacements
exist from other makers.
In any event, this Compaq uses a large 4.5 or 3V (haven't checked yet)
battery pack, not a Dallas module.
| Quote: | To be honest, given the vintage of the machine, if all you have to do
to make it work is press f1, you're doing pretty good.
|
Which is fine, apart from the fact that it was stuck in January 1st, 1980
and the normal DOS TIME/DATE commands couldn't fix this so it would survive
rebooting. I did manage to ask around and get ahold of a suitable
configuration disk. This let me set the time and date without incident (and
the battery now seems to be holding) but I've lost the hard disk. The
program is new enough to properly configure the system, but it tells me the
hard disk is not supported. I know nobody's replaced it, since it has Compaq
stickers and part numbers on it. I need to figure out what the exact
operating parameters of the disk are and see if I can manually pick one from
the listing that will work. It's a 120MB Conner Peripherals drive.
William |
|
| Back to top |
|
|
William R. Walsh
Guest
|
| Posted: Mon Mar 05, 2007 3:21 am Post subject: Re: Compaq Deskpro 386s20/n |
|
|
Hi!
| Quote: | The battery might instead be one of the older ones that attaches to a
motherboard header with two leads and is affixed to the inside of the
case with velcro. Sorry, don't have any of these... Ben Myers
|
It's definitely more like this type of battery and I do have a source who
can get them brand new for a pretty good price.
For now I've put the system aside (see my earlier post) and have been
working on reviving a Compaq 386DX/33 EISA box. I plan to run FreeDOS with
this system. (I always wanted to work with a 386DX/33...imagine what a nice
system that would have been back when it was new. Fortunately, its battery
is up (the system appears to charge it when running) and the configuration
partition was easily restored from disk images on the HP/Compaq site. Now
just to try and wait patiently for my EISA adapters to come in from eBay...)
William |
|
| Back to top |
|
|
|
|
paulmd@efn.org
Guest
|
| Posted: Mon Mar 05, 2007 5:33 am Post subject: Re: Compaq Deskpro 386s20/n |
|
|
On Mar 4, 1:16 pm, "William R. Walsh"
<newsgrou...@idontwantjunqueemail.walshcomptech.com> wrote:
| Quote: | Hi!
Untimely? It's going on 20 years old.
It runs and it works. What more do you want?
|
Web surfing?
| Quote: | I can think of useful things to
do with it, although I do realize that I'm probably in something of a
minority.
|
I don't have a problem with using ancient beasts. It's just spending
money on fixing them. If you're just having fun, go ahead, nothing
wrong with that. But you can get a working pentuim 3 for under $50.
| Quote: | No suprise there.
Actually, there was. :-) Someone had gone to the trouble of replacing it.
They cut into the original Compaq wiring and grafted a more standard battery
into the system.
386 machines often have a Dallas Realtime clock that holds it's own
charge for about 10 years.
They really do quite better than that in real life. Ten years is the
lifetime of the memory without any external power. I have some of these that
are over 10 years old and still keeping excellent time. The same seems to be
true of their "bigger brother" the DS1387.
The DS12887 and DS12887+ (ROHS compliant) are still available and work fine
in every machine I ever had to put one in. I'm not sure where you got the
idea that these "don't work" in a 386. I've put new parts in 286, 386SX and
386 systems and never had a problem. Quite a few pin compatible replacements
exist from other makers.
|
Last time i looked, which was last year, Dallas said they stopped
making the 12887 but was offering a 128887, which they said, right on
the spec sheet, more or less 'this wont work in your 386.'
| Quote: |
In any event, this Compaq uses a large 4.5 or 3V (haven't checked yet)
battery pack, not a Dallas module.
To be honest, given the vintage of the machine, if all you have to do
to make it work is press f1, you're doing pretty good.
Which is fine, apart from the fact that it was stuck in January 1st, 1980
and the normal DOS TIME/DATE commands couldn't fix this so it would survive
rebooting. I did manage to ask around and get ahold of a suitable
configuration disk. This let me set the time and date without incident (and
the battery now seems to be holding) but I've lost the hard disk. The
program is new enough to properly configure the system, but it tells me the
hard disk is not supported. I know nobody's replaced it, since it has Compaq
stickers and part numbers on it. I need to figure out what the exact
operating parameters of the disk are and see if I can manually pick one from
the listing that will work. It's a 120MB Conner Peripherals drive.
William |
|
|
| Back to top |
|
|
William R. Walsh
Guest
|
| Posted: Mon Mar 05, 2007 6:30 am Post subject: Re: Compaq Deskpro 386s20/n |
|
|
Hi!
It can be done. :-) (You knew I'd say it!)
http://greyghost.dyndns.org/8507/part2/DSC03100.JPG (1216x912, 70KB, from a
PS/2 Model 80 and an 8507 monochrome display...)
I generally agree with what you say about a Pentium III, although they don't
seem that cheap around here quite yet. I just find the old computers more
interesting than just about any PIII box you could find:
http://greyghost.dyndns.org/mcastuff/
http://www.walshcomptech.com/comp_coll.htm
| Quote: | Last time i looked, which was last year, Dallas said they stopped
making the 12887 but was offering a 128887, which they said, right on
the spec sheet, more or less 'this wont work in your 386.'
|
I think you mean the 12C887, which doesn't work since it handles the century
byte differently.
By the way, I was surprised upon returning to this group for the first time
in a while...it's evidently nowhere near as alive as it used to be. :-(
William |
|
| Back to top |
|
|
Ben Myers
Guest
|
| Posted: Mon Mar 05, 2007 6:31 am Post subject: Re: Compaq Deskpro 386s20/n |
|
|
EISA adapters for what? EISA graphics cards were not even close to being what
they were cracked up to be. EISA SCSI adapters were pretty respectable. EISA
network cards offered no advantage over ISA ones. I do not think that there
ever was an EISA 10/100 card, but 3COM produced its 3C515 10/100 ISA NIC.
You also need an EISA configuration utility program with configuration files for
the motherboard and all EISA add-in cards.
Wow, this takes me way back to the EISA versus MicroChannel article I worked on
for PC Magazine. The 386/33 was a much nicer box in its time than IBM's
ridiculously proprietary MicroChannel machines... Ben Myers
On Sun, 04 Mar 2007 21:21:11 GMT, "William R. Walsh"
<newsgroups1@idontwantjunqueemail.walshcomptech.com> wrote:
| Quote: | Hi!
The battery might instead be one of the older ones that attaches to a
motherboard header with two leads and is affixed to the inside of the
case with velcro. Sorry, don't have any of these... Ben Myers
It's definitely more like this type of battery and I do have a source who
can get them brand new for a pretty good price.
For now I've put the system aside (see my earlier post) and have been
working on reviving a Compaq 386DX/33 EISA box. I plan to run FreeDOS with
this system. (I always wanted to work with a 386DX/33...imagine what a nice
system that would have been back when it was new. Fortunately, its battery
is up (the system appears to charge it when running) and the configuration
partition was easily restored from disk images on the HP/Compaq site. Now
just to try and wait patiently for my EISA adapters to come in from eBay...)
William
|
|
|
| Back to top |
|
|
Ben Myers
Guest
|
| Posted: Mon Mar 05, 2007 6:32 am Post subject: Re: Compaq Deskpro 386s20/n |
|
|
Compaq had the QVision done, and then ATI had an EISA card for a while, but EISA
graphics were very quickly supplanted by first VL-bus then PCI. Few of the
graphics companies at the time wanted to deal with the added complexity of EISA
compared to ISA.
Yeah, the Adaptec 1740 is pretty good. I ran an EISA SCSI 486 system with a
Micronics motherboard as my main system for a while.
The 3c579 is 10mbit only, no 100mbit capability. About a year ago, my Compaq
EISA cards (convertible between Token Ring and Ethernet with a daughter card)
made their journey to the electronic board scrapper. Compaq made the TR card
and maybe one of the other companies specializing in TR did also. Like Proteon
and Madge. IBM did not.
EISA did outlive MicroChannel, which effectively bit the dust after 486s running
at 50 and 66MHz. I don't think there ever was a Pentium with MicroChannel, was
there? I don't think IBM ever did even a 100MHz 486 MCA system. EISA found
its way onto Pentium Pro and Pentium II server class boxes, and DEC Alpha
systems.
I have the old PC Magazine issue with the Bus Wars article here somewhere. I
saved many of the magazines with my articles in them. I also recall a trip
with PC Mag down to Boca, where I raised the hackles of the MicroChannel people
by telling them that the caching SCSI adapter controlled by an 80186 was
terribly mismatched with a 386, and that the adapter looked like it was designed
by accountants who found a lot of leftover 80186s and 30-pin SIMMs in a
warehouse somewhere.
Microchannel suffered from a lot of defects, but, along with EISA, paved the way
for the notion of embedding self-identification PALs in a device, so software
could easily tell what it was talking to. VL bus did not do this, and the
Intel PC guys picked right up on the idea when I suggested it to them. After
that, Plug and Play could become a reality. Of course, self-identification of
devices was not a new idea. It simply came from the mainframe world where some
of us worked in LARGE air-conditioned rooms with raised flooring.
To give perspective, PC Magazine back then ran maybe 500 or 600 pages every two
weeks, advertising computer boxes that cost thousands of dollars and monitors
costing up to a grand. Today, PC Magazine is a skinny publication, made so by
the internet ads and the lower prices of computers, incapable of sustaining the
expensive ad cost structure... Ben Myers
On Mon, 05 Mar 2007 04:49:56 GMT, "William R. Walsh"
<newsgroups1@idontwantjunqueemail.walshcomptech.com> wrote:
| Quote: | Hi!
EISA adapters for what? EISA graphics cards were not even close to
being what they were cracked up to be.
Compaq's QVision seems to be pretty darn respectable. I've run one of these
in a 486DX2/66 EISA box and it was more than fast enough at the time. It
also had a VRAM upgrade fitted, so it could manage high color at up to
1024x768 resolution. This 386DX/33 box has the same adapter, and the VRAM
upgrade is present on it as well.
EISA SCSI adapters were pretty respectable.
I found a seemingly new in box Adaptec AHA-1740 adapter with all the docs
and diskettes. Another preservation effort...image the disks and scan the
docs...
EISA network cards offered no advantage over ISA ones.
Really? Oh well, the price was right right for the three 3C579-TPs I got
ahold of.
I do not think that there ever was an EISA 10/100 card, but 3COM produced
its
3C515 10/100 ISA NIC.
Hmmm...I don't know. Guess I'd be surprised if there wasn't, as EISA did
hold on for a lot longer in the x86 world than MCA. I've heard of Pentium II
machines that had EISA slots. I'd really like to find a real EISA Token Ring
card.
Wow, this takes me way back to the EISA versus MicroChannel article
I worked on for PC Magazine. The 386/33 was a much nicer box in
its time than IBM's ridiculously proprietary MicroChannel machines...
Funny you'd mention that (got a copy of the article? I'd love to read
it...wct<atsign>walshcomptech<dot>com) as I have quite the Microchannel
collection:
http://greyghost.dyndns.org/mcastuff/
http://www.walshcomptech.com/comp_coll.htm
I guess you could say that I think microchannel was "insufficiently
appreciated" in its day, although I can very easily see why that was. But
they are rock solid reliable boxen... Oh, by the way, there does exist a
100Mbit Ethernet card for MCA-bus systems. Too bad it's a loosely converted
ISA design and a joke from a reliability perspective...
William
|
|
|
| Back to top |
|
|
paulmd@efn.org
Guest
|
| Posted: Mon Mar 05, 2007 6:34 am Post subject: Re: Compaq Deskpro 386s20/n |
|
|
On Mar 4, 7:28 pm, Ben Myers <ben_myers_spam_me_...@charter.net>
wrote:
| Quote: | EISA adapters for what? EISA graphics cards were not even close to being what
they were cracked up to be. EISA SCSI adapters were pretty respectable. EISA
network cards offered no advantage over ISA ones. I do not think that there
ever was an EISA 10/100 card, but 3COM produced its 3C515 10/100 ISA NIC.
|
I've seen a few by 3com, they were in my hands just long enough for me
to break off the gold leads, and put the rest in the High Grade
Circuit board bin.
Our linux guy tried to get one working an an SGI machine, running
Gentoo linux, couldn't find drivers except 'experimental' ones. And
then cursed when he remembered he could put an isa card in the eisa
slot. :)
| Quote: | You also need an EISA configuration utility program with configuration files for
the motherboard and all EISA add-in cards.
Wow, this takes me way back to the EISA versus MicroChannel article I worked on
for PC Magazine. The 386/33 was a much nicer box in its time than IBM's
ridiculously proprietary MicroChannel machines... Ben Myers
On Sun, 04 Mar 2007 21:21:11 GMT, "William R. Walsh"
newsgrou...@idontwantjunqueemail.walshcomptech.com> wrote:
Hi!
The battery might instead be one of the older ones that attaches to a
motherboard header with two leads and is affixed to the inside of the
case with velcro. Sorry, don't have any of these... Ben Myers
It's definitely more like this type of battery and I do have a source who
can get them brand new for a pretty good price.
For now I've put the system aside (see my earlier post) and have been
working on reviving a Compaq 386DX/33 EISA box. I plan to run FreeDOS with
this system. (I always wanted to work with a 386DX/33...imagine what a nice
system that would have been back when it was new. Fortunately, its battery
is up (the system appears to charge it when running) and the configuration
partition was easily restored from disk images on the HP/Compaq site. Now
just to try and wait patiently for my EISA adapters to come in from eBay...)
William |
|
|
| Back to top |
|
|
|
|
William R. Walsh
Guest
|
| Posted: Mon Mar 05, 2007 6:35 am Post subject: Re: Compaq Deskpro 386s20/n |
|
|
Hi!
| Quote: | EISA adapters for what? EISA graphics cards were not even close to
being what they were cracked up to be.
|
Compaq's QVision seems to be pretty darn respectable. I've run one of these
in a 486DX2/66 EISA box and it was more than fast enough at the time. It
also had a VRAM upgrade fitted, so it could manage high color at up to
1024x768 resolution. This 386DX/33 box has the same adapter, and the VRAM
upgrade is present on it as well.
| Quote: | EISA SCSI adapters were pretty respectable.
|
I found a seemingly new in box Adaptec AHA-1740 adapter with all the docs
and diskettes. Another preservation effort...image the disks and scan the
docs...
| Quote: | EISA network cards offered no advantage over ISA ones.
|
Really? Oh well, the price was right right for the three 3C579-TPs I got
ahold of.
| Quote: | I do not think that there ever was an EISA 10/100 card, but 3COM produced
its
3C515 10/100 ISA NIC.
|
Hmmm...I don't know. Guess I'd be surprised if there wasn't, as EISA did
hold on for a lot longer in the x86 world than MCA. I've heard of Pentium II
machines that had EISA slots. I'd really like to find a real EISA Token Ring
card.
| Quote: | Wow, this takes me way back to the EISA versus MicroChannel article
I worked on for PC Magazine. The 386/33 was a much nicer box in
its time than IBM's ridiculously proprietary MicroChannel machines...
|
Funny you'd mention that (got a copy of the article? I'd love to read
it...wct<atsign>walshcomptech<dot>com) as I have quite the Microchannel
collection:
http://greyghost.dyndns.org/mcastuff/
http://www.walshcomptech.com/comp_coll.htm
I guess you could say that I think microchannel was "insufficiently
appreciated" in its day, although I can very easily see why that was. But
they are rock solid reliable boxen... Oh, by the way, there does exist a
100Mbit Ethernet card for MCA-bus systems. Too bad it's a loosely converted
ISA design and a joke from a reliability perspective...
William |
|
| Back to top |
|
|
William R. Walsh
Guest
|
| Posted: Mon Mar 05, 2007 12:55 pm Post subject: Re: Compaq Deskpro 386s20/n |
|
|
Hi!
| Quote: | Compaq had the QVision done, and then ATI had an EISA card for a while
|
ATI GUP/Mach32? I have one in Microchannel and thought I saw one in EISA at
some time. I know PCI and VL-BUS versions also existed.
I haven't found the Microchannel one to be worth the expense of owning it.
The NT driver is marginal and the card itself is slooooow. IBM's own XGA-2
runs rings around it in high color modes, too bad Microsoft never saw fit to
write a driver that could fully exploit the card. (At least these woes were
resolved for Win9x users, thanks to the efforts of a third party with a
Win95 DDK.)
| Quote: | About a year ago, my Compaq EISA cards (convertible between Token Ring and
Ethernet with a daughter card) made their journey to the electronic board
scrapper. |
Gaaaak! I'm astounded to learn that Compaq had anything to do with TR that
they actually made. However, I do have some TR PCI cards that are IBM made
and "Compaq" branded.
| Quote: | EISA did outlive MicroChannel, which effectively bit the dust after 486s
running at 50 and 66MHz. I don't think there ever was a Pentium with
MicroChannel, was there?
|
Officially there were the Server 95 systems that came in Pentium 60 and
66MHz versions. The PC Server 500 (using the same planar as the Server 95
but a different case) featured a processor complex using a 90MHz Pentium
CPU. The 90MHz processor complex has been pushed to 233MHz MMX with varying
results, although 200MHz is easily obtained and stable. The PC Servers 5xx
and 720 were Pentium based systems with an MCA bus. In the case of the 720,
the system was a combination PCI/MCA unit. There also exist MCA versions of
the PC Server 3xx series and some 688* desktops with "Select A Bus". The 720
was capable of six-way SMP at up to 200MHz per processor card.
I have a Server 95a (the last PS/2 model) with a 90MHz Pentium processor
complex that's been rewired to accept a Pentium Overdrive 200 (run at 180)
MMX CPU. This thing is every bit of seriously sweet to use.
There were some Pentium fired NCR MCA-based servers. Nobody's ever seen one
that I know of.
| Quote: | I don't think IBM ever did even a 100MHz 486 MCA system.
|
They sure did. Models 76 and 77, both built around the "Lacuna" board either
came with or had everything you needed to run a 3.3V 486 CPU. These (and the
NCR 3350, which I also have) will take a Pentium OverDrive 63/83 and have an
L2 cache slot. Other 486 systems (like the Server 85 "X" models) don't take
a POD but will stand a 486-compatible upgrade, including the 133MHz AMD 486
chips.
The 8570-Bxx series machines were available with a "486 Power Platform"
running at 25MHz. It's been said that these were the first 486 computers
ever marketed. They are rare to say the least. Reports say that with some
diddling around, an AMD 486/133 processor upgrade works on them as well.
There were other models, many of which I have. If you didn't look at my
computer collection, you might want to. It's somewhat outdated in parts, but
I'm working on redoing it to feature all of the stuff I have. When MCA had
died off in the x86 world, it continued for another little while in the
RS/6000 world. I don't have any of those.
| Quote: | where I raised the hackles of the MicroChannel people by telling them that
the caching SCSI adapter controlled by an 80186 was terribly mismatched
with a 386, and that the adapter looked like it was designed
by accountants who found a lot of leftover 80186s and 30-pin SIMMs in a
warehouse somewhere.
|
Interesting. I did find with some benchmarking experiments that the short
late uncached (and 80C32 based) IBM MCA SCSI adapter was much faster than
the 80188 based caching ones. I've never been sure why, perhaps it was
better code optimization for the 80C32 microcontroller. Still, 8X CD burning
is possible on a 486 with this SCSI subsystem running the hard disks and an
NCR 53C710 driving the burner. I'd have to say IBM did a good job with these
in some ways, because I have a 9585 "X" machine with onboard cached SCSI
that is happily driving a 50GB Seagate SCA disk in a 3510 enclosure under
Windows NT 4.0. The boot-BIOS seems to only go up to 8GB, but an OS with
drivers that take over will do fine.
The so-called fast/wide adapters lost the cache and went with an 80186
clocked at 20MHz. These things haul...burning a CD at 12X is easy to do as
long as the hard disk can keep up. I've clocked this doing 10.1MB/sec
transfer. A later version with wide SE on the inside and external HVD goes
even faster (14.1MB/sec!). That one is driven by a 40MHz (!!!) 80186. These
adapters have been used to drive 250GB hard disks without incident.
| Quote: | To give perspective, PC Magazine back then ran maybe 500 or 600 pages
every two weeks, advertising computer boxes that cost thousands of dollars
and monitors costing up to a grand.
|
I remember those days and loved PC Magazine at the time. I was almost too
late to that particular time frame, but I was there and reading the
magazine. Too bad it isn't that way any more, although I don't think it
would still be here without having slimmed down as it did. I also remember
the days when Computer Shopper was a huge thing that could probably have
been used to defend yourself if need be!
Gosh, did that sound like an IBM commercial of some sort? :-) Sorry for the
rambling!
William |
|
| Back to top |
|
|
Ben Myers
Guest
|
| Posted: Mon Mar 05, 2007 9:14 pm Post subject: Re: Compaq Deskpro 386s20/n |
|
|
You got into MicroChannel a lot more than I did, and you have the real hardware
there to reinforce your memory. I forgot about the later hot-running Socket 4
Pentium boxes and the Pentium OverDrive. I may still have a Pentium OverDrive
kicking around here. For a while, I sold hand-assembled 486 upgrade kits built
around the AMD Am486-133 133MHz 486 workalike. Never got to try one in an IBM
box. It probably would not have worked.
It's possible someone else manufactured the Compaq EISA Ethernet/Token Ring card
for the Paq. It was certainly odd, with a little daughter card that attached
to give it one or the other network personality.
NCR was the other major player in MicroChannel, to the extent that NCR brand
systems were sold with MicroChannel boards inside. But NCR hedged its bets
with ISA and EISA systems, too. A lot of good that did them. One of my
ex-bosses from many years gone by was Pres of NCR for a while. He made his own
unique contribution to their demise as a computer name brand, and the NCR
product lines and product direction reflected his own chronic indecisiveness.
My benchmark tests (saturating the subsystem with I/O operations) showed that
the non-cached MCA SCSI adapter had more throughput than the 80188 cached one.
That's what led me to the assertion that really pissed off the IBM MicroChannel
guys. The 80188 bandwidth was too narrow, and could not run fast enough to do
all the necessary management of cached blocks while still meeting the demand
requests of the software on the motherboard. Was the on-board 80188 a
screaming 12MHz? IBM apparently recovered from the experience with a faster
controller, comparable, in fact, to the SCSI HBA used by mainstream Adaptec, who
also did a 1640 MCA SCSI host adapter... Ben Myers
On Mon, 05 Mar 2007 06:55:40 GMT, "William R. Walsh"
<newsgroups1@idontwantjunqueemail.walshcomptech.com> wrote:
| Quote: | Hi!
Compaq had the QVision done, and then ATI had an EISA card for a while
ATI GUP/Mach32? I have one in Microchannel and thought I saw one in EISA at
some time. I know PCI and VL-BUS versions also existed.
I haven't found the Microchannel one to be worth the expense of owning it.
The NT driver is marginal and the card itself is slooooow. IBM's own XGA-2
runs rings around it in high color modes, too bad Microsoft never saw fit to
write a driver that could fully exploit the card. (At least these woes were
resolved for Win9x users, thanks to the efforts of a third party with a
Win95 DDK.)
About a year ago, my Compaq EISA cards (convertible between Token Ring and
Ethernet with a daughter card) made their journey to the electronic board
scrapper.
Gaaaak! I'm astounded to learn that Compaq had anything to do with TR that
they actually made. However, I do have some TR PCI cards that are IBM made
and "Compaq" branded.
EISA did outlive MicroChannel, which effectively bit the dust after 486s
running at 50 and 66MHz. I don't think there ever was a Pentium with
MicroChannel, was there?
Officially there were the Server 95 systems that came in Pentium 60 and
66MHz versions. The PC Server 500 (using the same planar as the Server 95
but a different case) featured a processor complex using a 90MHz Pentium
CPU. The 90MHz processor complex has been pushed to 233MHz MMX with varying
results, although 200MHz is easily obtained and stable. The PC Servers 5xx
and 720 were Pentium based systems with an MCA bus. In the case of the 720,
the system was a combination PCI/MCA unit. There also exist MCA versions of
the PC Server 3xx series and some 688* desktops with "Select A Bus". The 720
was capable of six-way SMP at up to 200MHz per processor card.
I have a Server 95a (the last PS/2 model) with a 90MHz Pentium processor
complex that's been rewired to accept a Pentium Overdrive 200 (run at 180)
MMX CPU. This thing is every bit of seriously sweet to use.
There were some Pentium fired NCR MCA-based servers. Nobody's ever seen one
that I know of.
I don't think IBM ever did even a 100MHz 486 MCA system.
They sure did. Models 76 and 77, both built around the "Lacuna" board either
came with or had everything you needed to run a 3.3V 486 CPU. These (and the
NCR 3350, which I also have) will take a Pentium OverDrive 63/83 and have an
L2 cache slot. Other 486 systems (like the Server 85 "X" models) don't take
a POD but will stand a 486-compatible upgrade, including the 133MHz AMD 486
chips.
The 8570-Bxx series machines were available with a "486 Power Platform"
running at 25MHz. It's been said that these were the first 486 computers
ever marketed. They are rare to say the least. Reports say that with some
diddling around, an AMD 486/133 processor upgrade works on them as well.
There were other models, many of which I have. If you didn't look at my
computer collection, you might want to. It's somewhat outdated in parts, but
I'm working on redoing it to feature all of the stuff I have. When MCA had
died off in the x86 world, it continued for another little while in the
RS/6000 world. I don't have any of those.
where I raised the hackles of the MicroChannel people by telling them that
the caching SCSI adapter controlled by an 80186 was terribly mismatched
with a 386, and that the adapter looked like it was designed
by accountants who found a lot of leftover 80186s and 30-pin SIMMs in a
warehouse somewhere.
Interesting. I did find with some benchmarking experiments that the short
late uncached (and 80C32 based) IBM MCA SCSI adapter was much faster than
the 80188 based caching ones. I've never been sure why, perhaps it was
better code optimization for the 80C32 microcontroller. Still, 8X CD burning
is possible on a 486 with this SCSI subsystem running the hard disks and an
NCR 53C710 driving the burner. I'd have to say IBM did a good job with these
in some ways, because I have a 9585 "X" machine with onboard cached SCSI
that is happily driving a 50GB Seagate SCA disk in a 3510 enclosure under
Windows NT 4.0. The boot-BIOS seems to only go up to 8GB, but an OS with
drivers that take over will do fine.
The so-called fast/wide adapters lost the cache and went with an 80186
clocked at 20MHz. These things haul...burning a CD at 12X is easy to do as
long as the hard disk can keep up. I've clocked this doing 10.1MB/sec
transfer. A later version with wide SE on the inside and external HVD goes
even faster (14.1MB/sec!). That one is driven by a 40MHz (!!!) 80186. These
adapters have been used to drive 250GB hard disks without incident.
To give perspective, PC Magazine back then ran maybe 500 or 600 pages
every two weeks, advertising computer boxes that cost thousands of dollars
and monitors costing up to a grand.
I remember those days and loved PC Magazine at the time. I was almost too
late to that particular time frame, but I was there and reading the
magazine. Too bad it isn't that way any more, although I don't think it
would still be here without having slimmed down as it did. I also remember
the days when Computer Shopper was a huge thing that could probably have
been used to defend yourself if need be!
Gosh, did that sound like an IBM commercial of some sort? :-) Sorry for the
rambling!
William
|
|
|
| Back to top |
|
|
Guest
|
| Posted: Mon Mar 05, 2007 9:53 pm Post subject: Re: Compaq Deskpro 386s20/n |
|
|
Hi!
| Quote: | You got into MicroChannel a lot more than I did, and
you have the real hardware there to reinforce your
memory.
|
Yeah...when in school, I saw and used Model 25s (286/XT) that remote-
booted from an Advanced Netware 286 (!!) server OS running on a Model
80. (Why they used the 286 version I can only guess...)
I liked the machines then, and wanted to have one. In 1997 I finally
got a Model 70 and nothing's been quite the same since. Of course,
when the school finally got done with all of this stuff in the early
2000s, they called me before taking it all away to be recycled. They
had one big pickup truck bed's worth before I got there.
When I got done, they had nothing to take. :-)
| Quote: | For a while, I sold hand-assembled 486 upgrade kits
built around the AMD Am486-133 133MHz 486 workalike.
Never got to try one in an IBM box. It probably
would not have worked.
|
Bah. :-) I've got some from Evergreen and Kingston. They all work
great. Some early processor complexes did have problems, though.
| Quote: | It was certainly odd, with a little daughter
card that attached to give it one or the
other network personality.
|
Now I'm going to have to look for one of those. :-) It'll go along
well with my 487SX and other "weird stuff".
I've found the 3350 to be an interesting box. It's surprisingly
capable in some ways and lacking in others. The 77C22E video subsystem
is very capable per what spec sheets I've found...the drivers never
took advantage of all it could do.
| Quote: | The 80188 bandwidth was too narrow, and could
not run fast enough to do all the necessary
management of cached blocks while still meeting
the demand requests of the software on the
motherboard.
|
I've always wondered why the 80C32 cacheless controllers did so much
better on benchmarks. Thanks for that info.
The 80188 started running at 10MHz and later moved up to 16MHz on the
latest revision of the card. The Adaptec 1640--for having an annoying
1GB limit and only using a 16-bit slot--is definitely faster on just
about every hard disk benchmark I ever tried.
Hey, it's been great having this discussion, even if it's not strictly
Compaq related. If you should find those PC Magazine articles and
wouldn't mind passing a copy on, I'd love to see them.
William |
|
| Back to top |
|
|
DELETED
Guest
|
| Posted: Fri Apr 20, 2007 9:41 am Post subject: re:Compaq Deskpro 386s20/n |
|
|
| DELETED |
|
| Back to top |
|
|
|
|
|
|
| |
Watched Topics
Search
Register
Log in to check your private messages
Profile
Log in
Memberlist
Usergroups