I am running MSDOS and Win 3.1 in VirtualBox. Don't ask. I would like to
take an older computer and install MSDOS on it because it would be better
than running from VirtualBox. What would be the restrictions on the CPU?
Also if the computer has more than 64K and maybe 640K of memory is there a
way to restrict the memory amount? I am asking here because Paul and others
will have the answers. By the way if you complain about the later Windows
versions go back and look at 3.1.

<Bill>

On Thu, 30 Jun 2022 at 08:50:54, Bill Bradshaw <bradshaw@gci.net> wrote
(my responses usually FOLLOW):
>I am running MSDOS and Win 3.1 in VirtualBox. Don't ask. I would like to
>take an older computer and install MSDOS on it because it would be better
>than running from VirtualBox. What would be the restrictions on the CPU?

I _think_ any "old" PC still actually working would be capable of
running MSDOS (and probably Windows 3.1); I'd say it's more a matter of
hardware reliability problems.

I would avoid multicore processors - I think in most cases they'll _run_
DOS/W3.1, but almost certainly would only _use_ one of the cores, so
there's little point.

>Also if the computer has more than 64K and maybe 640K of memory is there a
>way to restrict the memory amount?

Oh, you're talking _really_ old computers! (Ones with a _real_ on-off
switch!) I'm not sure I'd try to go back that far, because of hardware
reliability concerns. AFAICR, DOS doesn't _use_ anything above 640K - or
possibly 1M - even if it's present, without you jumping through hoops
(things like memory managers etc.) even if it is there; I think my first
proper "PC" had 4M, and I don't remember that causing DOS any problems
(I used to use it in DOS most of the time, mostly only using Windows
when I wanted to play with the fledgling WWW). If anything, it was more
a matter of the problem being _getting_ DOS to use more memory than it
wanted to - playing with HIMEM, QEMM, and other such things. If what you
want to do with it are things that'll sit happily in the memory DOS is
happy with without fiddling with it, it should be fine.

(You say "Don't ask", but I _am_ curious: what old things - software -
_are_ you running? Will you be using floppies [either size] - if so you
actually _want_ an older motherboard so it still has the floppy-drive
connector on it, though those usually remained on the mobo until
relatively recently, compared to DOS/W3.x.)

> I am asking here because Paul and others
>will have the answers. By the way if you complain about the later Windows
>versions go back and look at 3.1.

They had their problems. (Not least the lack of certain networking
built-in - I remember Trumpet Winsock - but we managed! Again, curious
about what you are _wanting_ to run.)
>
><Bill>
>
>
John
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

If a cluttered desk is characteristic of a cluttered mind, what does an empty
desk mean ?

On 6/30/2022 12:50 PM, Bill Bradshaw wrote:
> I am running MSDOS and Win 3.1 in VirtualBox. Don't ask. I would like to
> take an older computer and install MSDOS on it because it would be better
> than running from VirtualBox. What would be the restrictions on the CPU?
> Also if the computer has more than 64K and maybe 640K of memory is there a
> way to restrict the memory amount? I am asking here because Paul and others
> will have the answers. By the way if you complain about the later Windows
> versions go back and look at 3.1.
>
> <Bill>

The processors and the chipsets, continued to have
support for quite a while. I'd say maybe P4 and ICH5
would be a good place to stop. I've run Win98 on a Core2
and it just ignored the second CPU core. The VIA chipset
had Win98 drivers (I selected the VIA chipset on purpose,
because of the chance of finding drivers). Today, you need
archive.org support for your adventure, and it's hard to
say what drivers will be lost to time.

https://en.wikipedia.org/wiki/Real_mode

"for backward compatibility, all x86 CPUs start in
real mode when reset, though it is possible to emulate
real mode on other systems when starting on other modes."

"Decline

The changing towards the NT kernel resulted in the
operating system not needing DOS to boot the computer
as well as being unable to use it. The need to restart the
computer in real mode MS-DOS declined after Windows 3.1x
until it was no longer supported in Windows ME. The
only way of currently running DOS applications that
require real mode from within newer versions of Windows
is by using emulators such as DOSBox or x86 virtualization products."

Processors are started using an onion-skin approach. At
first, there isn't even any mapped RAM to run code
(register based code). The caches are turned off at first.
Whatever code a developer runs, it is up to that developer
to decide when the "next bit is to be turned on". This is
how newer hardware runs older OSes.

Intel kept "Compatibility Mode" for their Southbridges,
so OSes like Windows 98 would see four SATA ports named
"Primary IDE Master", "Primary IDE Slave",
"Secondary IDE Master", "Secondary IDE Slave",
and having a usable storage is a requirement for
the older OSes. The graphics frame buffer is mapped
to a well known address. These are "comfort foods"
for an older OS. You could look in a motherboard
user manual, on a four or six port SATA board and
see if it still had a Compatibility Mode.

But I don't have dates, as to when Intel started
removing some of these things. The machine I'm typing on,
doesn't have Compatibility Mode. It doesn't have a floppy.
It doesn't have a PCI slot (boo/hiss, I need that). I would
think items with P4 processors would be a reasonably limit,
to start.

Paul

On 6/30/22 09:50, Bill Bradshaw wrote:
> I am running MSDOS and Win 3.1 in VirtualBox. Don't ask. I would like to
> take an older computer and install MSDOS on it because it would be better
> than running from VirtualBox. What would be the restrictions on the CPU?
> Also if the computer has more than 64K and maybe 640K of memory is there a
> way to restrict the memory amount? I am asking here because Paul and others
> will have the answers. By the way if you complain about the later Windows
> versions go back and look at 3.1.
>
> <Bill>
>
>

Any used computer stores near by?

In alt.comp.os.windows-10, on Thu, 30 Jun 2022 08:50:54 -0800, "Bill
Bradshaw" <bradshaw@gci.net> wrote:

>I am running MSDOS and Win 3.1 in VirtualBox. Don't ask. I would like to
>take an older computer and install MSDOS on it because it would be better
>than running from VirtualBox. What would be the restrictions on the CPU?
>Also if the computer has more than 64K and maybe 640K of memory is there a
>way to restrict the memory amount?

I take your question to mean, Can I limit the amount of memory used to
less than what's available, less than 640K for example. There was a
way to do that, and I'm sure it still works, but darned if I remember
how.

> I am asking here because Paul and others
>will have the answers. By the way if you complain about the later Windows
>versions go back and look at 3.1.
>
><Bill>
>

In article <2vfsbhtok309saqnmpo096cs19cflm9vns@4ax.com>,
micky says...

> I take your question to mean, Can I limit the amount of memory used to
> less than what's available, less than 640K for example. There was a
> way to do that, and I'm sure it still works, but darned if I remember
> how.
>

Using RAMDRIVE.SYS, Set up a Ram Drive of an appropriate size in
Config.sys.

--
Ken

J. P. Gilliver (John) wrote:
> (You say "Don't ask", but I _am_ curious: what old things - software -
> _are_ you running? Will you be using floppies [either size] - if so
> you actually _want_ an older motherboard so it still has the
> floppy-drive connector on it, though those usually remained on the
> mobo until relatively recently, compared to DOS/W3.x.)
>
I wrote a program completed in 1993. It has to run on a fat partition and
works just fine using DOSBox. I installed MSDOS in VirtualBox and the
program works great on it. I need to create a devopement environmen where I
can make some changes they want. Fortunately I still have the code and old
deelopment programs. I do have the computer I used to create the program in
my crawl space somewhere but I am trying to avoid using it,

<Bill>

In alt.comp.os.windows-10, on Fri, 1 Jul 2022 06:28:42 +0100,
Unsteadyken <unsteadyken@gmail.com> wrote:

>In article <2vfsbhtok309saqnmpo096cs19cflm9vns@4ax.com>,
>
>micky says...
>
>> I take your question to mean, Can I limit the amount of memory used to
>> less than what's available, less than 640K for example. There was a
>> way to do that, and I'm sure it still works, but darned if I remember
>> how.
>>
>
>Using RAMDRIVE.SYS, Set up a Ram Drive of an appropriate size in
>Config.sys.

That sounds good. FTR, I think there is another way too. Is the dos ng
still functioning? What about -- I hate to say it -- a dos web forum?
My gosh, I hate webforums.

On 6/30/22 13:44, J. P. Gilliver (John) wrote:
> On Thu, 30 Jun 2022 at 08:50:54, Bill Bradshaw <bradshaw@gci.net> wrote
> (my responses usually FOLLOW):
>> I am running MSDOS and Win 3.1 in VirtualBox.  Don't ask.  I would
>> like to
>> take an older computer and install MSDOS on it because it would be better
>> than running from VirtualBox.  What would be the restrictions on the CPU?
>
> I _think_ any "old" PC still actually working would be capable of
> running MSDOS (and probably Windows 3.1); I'd say it's more a matter of
> hardware reliability problems.
>
> I would avoid multicore processors - I think in most cases they'll _run_
> DOS/W3.1, but almost certainly would only _use_ one of the cores, so
> there's little point.
>
>> Also if the computer has more than 64K and maybe 640K of memory is
>> there a
>> way to restrict the memory amount?
>
> Oh, you're talking _really_ old computers! (Ones with a _real_ on-off
> switch!)

I suppose you mean one where the switch itself disconnects power, and
"off" is really OFF (draws ZERO current).

> I'm not sure I'd try to go back that far, because of hardware
> reliability concerns. AFAICR, DOS doesn't _use_ anything above 640K - or
> possibly 1M - even if it's present, without you jumping through hoops
> (things like memory managers etc.) even if it is there;

Actually a little more (IIRC, 1M+64K-16b) considering a real-mode bug.
Later versions of DOS could use that.

I think my first
> proper "PC" had 4M, and I don't remember that causing DOS any problems
> (I used to use it in DOS most of the time, mostly only using Windows
> when I wanted to play with the fledgling WWW).

i had about the same, using DOS most of the time. I needed Windows for
the web, as well as a card called "PC Tele-Vision".

> If anything, it was more
> a matter of the problem being _getting_ DOS to use more memory than it
> wanted to - playing with HIMEM, QEMM, and other such things. If what you
> want to do with it are things that'll sit happily in the memory DOS is
> happy with without fiddling with it, it should be fine.
>
> (You say "Don't ask", but I _am_ curious: what old things - software -
> _are_ you running? Will you be using floppies [either size] - if so you
> actually _want_ an older motherboard so it still has the floppy-drive
> connector on it, though those usually remained on the mobo until
> relatively recently, compared to DOS/W3.x.)

There are 3.5-inch floppy drives that connect by USB, although I've
never seen 5.25-inch ones that use USB.

>>  I am asking here because Paul and others
>> will have the answers.  By the way if you complain about the later
>> Windows
>> versions go back and look at 3.1.
>
> They had their problems. (Not least the lack of certain networking
> built-in - I remember Trumpet Winsock - but we managed! Again, curious
> about what you are _wanting_ to run.)
>>
>> <Bill>
>>
>>
> John

--
Mark Lloyd
http://notstupid.us/

"File Linking Error. Your mistake is now in every file."

On 6/30/22 11:50, Bill Bradshaw wrote:
> I am running MSDOS and Win 3.1 in VirtualBox. Don't ask. I would like to
> take an older computer and install MSDOS on it because it would be better
> than running from VirtualBox. What would be the restrictions on the CPU?
> Also if the computer has more than 64K and maybe 640K of memory is there a
> way to restrict the memory amount?

I remember seeing a program called EATMEM that does nothing but take up
memory, but have no idea of where you'd get that.

Why do you want to limit memory?

> I am asking here because Paul and others
> will have the answers. By the way if you complain about the later Windows
> versions go back and look at 3.1.

Yes, 3.1 was bad. That's why 95 seemed so good.

BTW, I have run Windows 95 in a VM recently (actually in order to test
Internet Explorer 1, which is limited to Win 95). Initially, it would
not allow internet access over the network. At that time, MS expected
you to not want to use TCP/IP over ethernet. The solution wasn't that
hard, once I remembered what I had done 20 years go.

> <Bill>
>
>

--
Mark Lloyd
http://notstupid.us/

"File Linking Error. Your mistake is now in every file."

On 6/30/22 15:32, Paul wrote:

[snip]

> https://en.wikipedia.org/wiki/Real_mode
>
>    "for backward compatibility, all x86 CPUs start in
>     real mode when reset, though it is possible to emulate
>     real mode on other systems when starting on other modes."

The 80376 didn't have real mode:
https://en.wikipedia.org/wiki/Intel_80376

>    "Decline
>
>     The changing towards the NT kernel resulted in the
>     operating system not needing DOS to boot the computer
>     as well as being unable to use it. The need to restart the
>     computer in real mode MS-DOS declined after Windows 3.1x
>     until it was no longer supported in Windows ME. The
>     only way of currently running DOS applications that
>     require real mode from within newer versions of Windows
>     is by using emulators such as DOSBox or x86 virtualization products."

Or you could use a DOS boot floppy.

[snip]

--
Mark Lloyd
http://notstupid.us/

"File Linking Error. Your mistake is now in every file."

On Fri, 1 Jul 2022 at 12:06:37, Mark Lloyd <not.email@all.invalid> wrote
(my responses usually FOLLOW):
>On 6/30/22 13:44, J. P. Gilliver (John) wrote:
[]
>> Oh, you're talking _really_ old computers! (Ones with a _real_
>>on-off switch!)
>
>I suppose you mean one where the switch itself disconnects power, and
>"off" is really OFF (draws ZERO current).

Well, yes - I think the ATX power supply was where the "soft-off" came
in, i. e. the on/off switch was just a momentary-contact thing, and the
PSU and motherboard between them handled the control. (I think that's
when the concept of the "5V standby" supply came in too - I think some
supplies even did that supply the old-fashioned way, with a 50 [or 60]
Hz transformer, rather than switched-mode.)

But I was really thinking of the great old - genuine IBM, I think,
though there may have been compatibles - '286 and earlier machines, that
had a very thick metal case, and a BIG red lever switch on the side that
went CLUNK. (We were still using such a machine where I worked well into
the 201x - we were in avionics maintenance, and there was some old piece
of kit for which the test software/hardware had been created on it; it
still worked, and that piece of kit came in so rarely it wasn't worth
anyone's time to replace it. [Plus, I doubt there were many, even in
that department, who'd have been able to!])
>
>> I'm not sure I'd try to go back that far, because of hardware
>>reliability concerns. AFAICR, DOS doesn't _use_ anything above 640K -
>>or possibly 1M - even if it's present, without you jumping through
>>hoops (things like memory managers etc.) even if it is there;
>
>Actually a little more (IIRC, 1M+64K-16b) considering a real-mode bug.
>Later versions of DOS could use that.

Yes, I knew there were all sorts of wrinkles to it.
>
> I think my first
>> proper "PC" had 4M, and I don't remember that causing DOS any
>>problems (I used to use it in DOS most of the time, mostly only using
>>Windows when I wanted to play with the fledgling WWW).
>
>i had about the same, using DOS most of the time. I needed Windows for
>the web, as well as a card called "PC Tele-Vision".

Ah yes - video capture cards (or even with a tuner). Quite a lot of
people kept old kit going for those, because they (bus-based cards in a
motherboard) worked better than the first two or three generations of
ones that worked via USB. (I think there's at least one person on the
groups I take who still says his old whatever card gives better
results.) [There was an intervening period when FireWire held sway there
- it was about the only wide-use thing that I ever saw FireWire used
for.]
[]
>>- _are_ you running? Will you be using floppies [either size] - if so
>>you actually _want_ an older motherboard so it still has the
>>floppy-drive connector on it, though those usually remained on the
>>mobo until relatively recently, compared to DOS/W3.x.)
>
>There are 3.5-inch floppy drives that connect by USB, although I've
>never seen 5.25-inch ones that use USB.

(Me neither.) I thought about those, but DOS support for USB was
virtually non-existent - even '95 was fairly hopeless, and even '98OSR2
was a matter of fingers crossed. I suspect finding a USB floppy drive
for which you could find drivers for DOS or even W3.1 would be a forlorn
quest, though I could be wrong.
[]
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

We no longer make things, but sell each other consultancy on how to run
consulatancies better. (Michael Cross, Computing 1999-3-4 [p. 28].)

On Fri, 1 Jul 2022 at 12:31:42, Mark Lloyd <not.email@all.invalid> wrote
(my responses usually FOLLOW):
>On 6/30/22 15:32, Paul wrote:
>
>[snip]
>
>> https://en.wikipedia.org/wiki/Real_mode
>>    "for backward compatibility, all x86 CPUs start in
>>     real mode when reset, though it is possible to emulate
>>     real mode on other systems when starting on other modes."
>
>The 80376 didn't have real mode:
>https://en.wikipedia.org/wiki/Intel_80376
[]
Well I'm blowed! I had to follow that link out of curiosity, as I'd
never heard of the '376, so wondered if it was a mis-type - but I see it
was a variant more aimed at integrated systems (as the 80186 had been).

I remember the '386SX didn't have a math co-processor - I think it
implemented it in software, or you could add it separately (was that the
'387?), or pay more for the '386DX which did.

The bit Paul quoted did say "all x86 CPUs", which arguably wouldn't
include the 376 (-:
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

We no longer make things, but sell each other consultancy on how to run
consulatancies better. (Michael Cross, Computing 1999-3-4 [p. 28].)

On 7/1/2022 3:12 PM, J. P. Gilliver (John) wrote:
> On Fri, 1 Jul 2022 at 12:06:37, Mark Lloyd <not.email@all.invalid> wrote (my responses usually FOLLOW):
>> On 6/30/22 13:44, J. P. Gilliver (John) wrote:
> []
>>>  Oh, you're talking _really_ old computers! (Ones with a _real_ on-off  switch!)
>>
>> I suppose you mean one where the switch itself disconnects power, and "off" is really OFF (draws ZERO current).
>
> Well, yes - I think the ATX power supply was where the "soft-off" came in, i. e. the on/off switch was just a momentary-contact thing, and the PSU and motherboard between them handled the control. (I think that's when the concept of the "5V standby" supply came in too - I think some supplies even did that supply the old-fashioned way, with a 50 [or 60] Hz transformer, rather than switched-mode.)
>
<snip>

The Pavouk sample schematic can show you some of the
older ATX ideas.

http://www.pavouk.org/hw/en_atxps.html

The "second supply" in the lower left corner, runs
as an SMPS (switcher) off the main DC capacitor used
by the main supply. The output of the switcher is
higher than 5V. It seems to form a supply for the
supervisory chip for the main switcher. So maybe the
switcher makes +12V for the TL494, and uses a linear
regulator (7805) to make +5VSB from that (weak) +12V.

By using a 7805, the device is "doomed" in terms of
efficiency. People used to make fun of how bad some
of the +5VSB were on supplies. And how a "puff of heat"
would come out of the supply, where the +5VSB cct had been
cooking while the PC sleeps.

Modern ATX use switchers where the output is the
right one for the job. And efficiency is more of a
concern, due to the marketing buzz from "gold" or
"platinum" rating or whatever. You won't find a 7805
in there.

Apparently, there is a new ATX standard 3.0 about to enter
the arena, which will fracture the supply market again.
Roughly speaking, if you ever *did* want to buy a
new supply for that old 440BX in the garage, now
is the time to buy your spare, not two years from now.
Two years from now, all the older ATX could be scooted
to the curb. It all depends on whether there is still
any repair demand, as to how well the older standards
will be supported. I would think COVID has largely cleaned
out the repair market anyway - you might not be able to
find a Sparkle even if you wanted one (the Sparkle would
work best on a True Sine UPS). Why a Sparkle ? well, modern
supplies offer 3.3V @ 20A and 5.0V @ 20A, and some
older applications need more than that. My NForce2 board
and 9800Pro video card together, drew something like 25 amps
from the 5V. The 12V in that case, had almost no load at all
(just disk drive motors).

The strongest ATX in that regard, produced +5V at 40A,
and it was hard to say why the wiring didn't melt, as
the wiring might not have been meant for all of that.
Some dual socket boards (two Athlons running off 5V VCores),
two pins on the ATX main connector used to burn, because
they were "too close to the edge" on current. This was
back in the cheapness-era where nobody would ever consider
putting 2oz copper for power layers. It's partly a PCB design
issue where current hogging wasn't properly controlled
(some of that requires via-stitching). (The 5V pins would
not have burned, if the four 5V pins shared the current flow
exactly equally. Which was not the case.)

It's because of the shift from running CPUs on 5V, to running
them on 12V, that these issues went away. Even 12V isn't high
enough... Doing 12V at 80A isn't all that clever, and takes
gobs of solder on the mezzanine board in the PSU (board with
the modular connector array on it).

https://www.anandtech.com/show/17447/the-gigabyte-ud1000gm-pg5-1000w-psu-review

The back of the mezzanine on that one is great looking. Most of
the current flow is likely on inner layers. It would take
me a bit of time to find one of the older gobbed up ones
(where the solder was applied with a spatula).

https://www.techpowerup.com/review/gigabyte-ud1000gm-pg5-1000-w/3.html

Paul

On Fri, 1 Jul 2022 12:16:22 -0500, Mark Lloyd wrote:

> Yes, 3.1 was bad. That's why 95 seemed so good.

Not so bad in 1992 (was it). Windows just got better and better in
those days.

micky <NONONOmisc07@fmguy.com> wrote:
> In alt.comp.os.windows-10, on Fri, 1 Jul 2022 06:28:42 +0100,
> Unsteadyken <unsteadyken@gmail.com> wrote:

> >In article <2vfsbhtok309saqnmpo096cs19cflm9vns@4ax.com>,
> >
> >micky says...
> >
> >> I take your question to mean, Can I limit the amount of memory used to
> >> less than what's available, less than 640K for example. There was a
> >> way to do that, and I'm sure it still works, but darned if I remember
> >> how.
> >>
> >
> >Using RAMDRIVE.SYS, Set up a Ram Drive of an appropriate size in
> >Config.sys.

> That sounds good. FTR, I think there is another way too. Is the dos ng
> still functioning? What about -- I hate to say it -- a dos web forum?
> My gosh, I hate webforums.

http://reddit.com/r/dos and https://www.reddit.com/r/MSDOS/
--
July and Canada Day, eh?
Note: A fixed width font (Courier, Monospace, etc.) is required to see this signature correctly.
/\___/\ Ant(Dude) @ http://aqfl.net & http://antfarm.home.dhs.org.
/ /\ /\ \ Please nuke ANT if replying by e-mail.
| |o o| |
\ _ /
( )

Bill Bradshaw wrote:
> I am running MSDOS and Win 3.1 in VirtualBox. Don't ask. I would
> like to take an older computer and install MSDOS on it because it
> would be better than running from VirtualBox. What would be the
> restrictions on the CPU? Also if the computer has more than 64K and
> maybe 640K of memory is there a way to restrict the memory amount? I
> am asking here because Paul and others will have the answers. By the
> way if you complain about the later Windows versions go back and look
> at 3.1.
> <Bill>

Everything is setup in VirtualBox. Using WinXP Pro, TASM, and Borland C++
4.5. Now just to get my memory refreshed regarding the efficient use of
these old programs again. Fortuantely when I created the program code I
included many comments regarding how it worked. Thanks for the responses.

<Bill>

On 7/1/22 14:19, J. P. Gilliver (John) wrote:

[snip]

> I remember the '386SX didn't have a math co-processor - I think it
> implemented it in software,

You could use software, or IIRC you could use a 287.

> or you could add it separately (was that the
> '387?), or pay more for the '386DX which did.

386DX did not have a coprocessor. Integrated coprocessors were
introduced with the 486.

IIRC, the differences in the 386SX where a multiplexed data bus
(internally 32-bit, but using 16-bit hardware) and a limited address bus
(only 24 bits available, max RAM of 16 MiB).

[snip]

--
Mark Lloyd
http://notstupid.us/

On Sat, 2 Jul 2022 at 11:57:31, Mark Lloyd <not.email@all.invalid> wrote
(my responses usually FOLLOW):
>On 7/1/22 14:19, J. P. Gilliver (John) wrote:
>
>[snip]
>
>> I remember the '386SX didn't have a math co-processor - I think it
>>implemented it in software,
>
>You could use software, or IIRC you could use a 287.
>
>> or you could add it separately (was that the '387?), or pay more for
>>the '386DX which did.
>
>386DX did not have a coprocessor. Integrated coprocessors were
>introduced with the 486.
>
>IIRC, the differences in the 386SX where a multiplexed data bus
>(internally 32-bit, but using 16-bit hardware) and a limited address
>bus (only 24 bits available, max RAM of 16 MiB).
>
>[snip]
>
Ah yes, I remember they changed what the difference between the SX
(budget) and DX was; one sort the SX had shared address/data buses and
the DX didn't (so had more pins), the other sort the SX didn't have a
maths processor built in (you could add one externally), the DX did. It
was certainly confusing.
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

The average age of a single mum in this country is 37
- Jane Rackham, RT 2016/5/28-6/3

J. P. Gilliver (John) wrote:

> Ah yes, I remember they changed what the difference between the SX (budget) and
> DX was; one sort the SX had shared address/data buses and the DX didn't (so had
> more pins), the other sort the SX didn't have a maths processor built in (you
> could add one externally), the DX did. It was certainly confusing.

You might be muddling 386/486 and SX/DX suffixes?

I think for 386 CPUs, SX meant 16bit bus, DX meant 32bit bus

but for 486 CPUs, SX meant no FPU, DX/DX2/DX4 meant with FPU

On Mon, 4 Jul 2022 at 07:38:05, Andy Burns <usenet@andyburns.uk> wrote
(my responses usually FOLLOW):
>J. P. Gilliver (John) wrote:
>
>> Ah yes, I remember they changed what the difference between the SX
>>(budget) and DX was; one sort the SX had shared address/data buses
>>and the DX didn't (so had more pins), the other sort the SX didn't
>>have a maths processor built in (you could add one externally), the
>>DX did. It was certainly confusing.
>
>You might be muddling 386/486 and SX/DX suffixes?
>
>I think for 386 CPUs, SX meant 16bit bus, DX meant 32bit bus
>
>but for 486 CPUs, SX meant no FPU, DX/DX2/DX4 meant with FPU

I wasn't muddling them - I remembered that that was the case. They
changed what the difference meant at the same point they changed from
38x to 48x. I remember the FPU you could buy separately was the x87,
though not how that worked (presumably mobos had a socket for it!).

I'd forgotten the DX2 and DX4 variants - weren't those 2 or 4 times
clock internally (so the start of having processor clock faster than
rest of system, though I think it only happened - at least, you only got
the DX/DX2/DX4 _choice_ - at the 486, with if it was the case on the
Pentium, no choice in the matter).

I remember there was an add-on module you could add to 486 systems that
gave you a Pentium (it might even have been called a 586) without having
to replace the whole of your computer. Since I think it involved such a
processor that still had to run on (I think) 5V, it didn't half run hot!
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

How do you govern a country that seems to have decided that facts are the work
of the devil? - Andy Hamilton on HIGNFY, 2010

Andy Burns <usenet@andyburns.uk> wrote:

>but for 486 CPUs, SX meant no FPU, DX/DX2/DX4 meant with FPU

As I recall, the SX actually had an FPU, but it was disabled. You
could then add a 487 "coprocessor" which was actually a
fully-functional 486. But you had to leave the SX plugged into the
motherboard. If you took it out to use in another computer the
"coprocessor" wouldn't work - even though the SX was no longer doing
anything at all.

--
Tim Slattery
tim <at> risingdove <dot> com

On 7/3/22 19:14, J. P. Gilliver (John) wrote:

[snip]

> Ah yes, I remember they changed what the difference between the SX
> (budget) and DX was; one sort the SX had shared address/data buses and
> the DX didn't (so had more pins), the other sort the SX didn't have a
> maths processor built in (you could add one externally), the DX did. It
> was certainly confusing.

There was a 487, which was IIRC a relabeled 486DX, used to add
coprocessor to a 486SX.

--
Mark Lloyd
http://notstupid.us/

"Only two things are infinite, the universe and stupidity. I am not so
sure about the former..."

On 7/4/22 01:38, Andy Burns wrote:
> J. P. Gilliver (John) wrote:
>
>> Ah yes, I remember they changed what the difference between the SX
>> (budget) and DX was; one sort the SX had shared address/data buses and
>> the DX didn't (so had more pins), the other sort the SX didn't have a
>> maths processor built in (you could add one externally), the DX did.
>> It was certainly confusing.
>
> You might be muddling 386/486 and SX/DX suffixes?
>
> I think for 386 CPUs, SX meant 16bit bus, DX meant 32bit bus

16-bit off-chip DATA bus and 24-bit ADDRESS bus. People seem to have
stopped recognizing the difference.

> but for 486 CPUs, SX meant no FPU, DX/DX2/DX4 meant with FPU

Yes, those with a number after the DX multiplied the clock frequency.
IIRC DX4 was really a 3x multiplier.

--
Mark Lloyd
http://notstupid.us/

"Only two things are infinite, the universe and stupidity. I am not so
sure about the former..."

Mark Lloyd wrote:

> Andy Burns wrote:
>
>> I think for 386 CPUs, SX meant 16bit bus, DX meant 32bit bus
>
> 16-bit off-chip DATA bus and 24-bit ADDRESS bus.

Oh, yeah I'd forgotten that address bus was thinned down as well.

> People seem to have stopped
> recognizing the difference.

Not exactly relevant to many people any more, maximum 16MB memory

>> but for 486 CPUs, SX meant no FPU, DX/DX2/DX4 meant with FPU
>
> Yes, those with a number after the DX multiplied the clock frequency. IIRC DX4
> was really a 3x multiplier.