Back to home page

DOS ain't dead

Forum index page

Log in | Register

Back to index page
Thread view  Board view
DOS386(R)

01.05.2008, 05:02
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h (Developers)

RayeR wrote # :

> Anyway I have a 386SX notebook laying around so I can test himemx there
> I tested himemx v3.32 on intel 386SX/25MHz

Great :-) How much RAM ? FPU ? Bugs (early 80386's reportedly were extremely buggy) ? Please keep it. I have a few things that should occasionally get tested on an 80386.

Rugxulo reportedly has an 80486 ... Freq ? RAM ? FPU ? CPUID ?

Anyone else has 80386 ? Anyone has a 16-bit PC ? Preferably 8086 :hungry:

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

Rugxulo(R)

Homepage

Usono,
01.05.2008, 05:44

@ DOS386
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> RayeR wrote # :
>
> > Anyway I have a 386SX notebook laying around so I can test himemx there
> > I tested himemx v3.32 on intel 386SX/25MHz
>
> Great :-) How much RAM ? FPU ?

He already said 4 MB RAM and 387 FPU.

> Rugxulo reportedly has an 80486 ... Freq ? RAM ? FPU ? CPUID ?

486 Sx/25 w/ 8 MB RAM (without CPUID support).

> Anyone else has 80386 ? Anyone has a 16-bit PC ? Preferably 8086 :hungry:

Jason Burgon (GVFM) has a 386 as does Blair Campbell. Trixter (of Oldskool.org) has an 8088. ATV (and two others??) on FASM's forum has a 286.

DOS386(R)

01.05.2008, 05:50

@ Rugxulo
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> 486 Sx/25 w/ 8 MB RAM (without CPUID support).

FPU ?

> Jason Burgon (GVFM) has a 386 as does Blair Campbell. Trixter (of
> Oldskool.org) has an 8088. ATV (and 2 others??) on FASM's forum has a 286.

COOL. Any 8086 around ?

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

Rugxulo(R)

Homepage

Usono,
01.05.2008, 05:57

@ DOS386
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> > 486 Sx/25 w/ 8 MB RAM (without CPUID support).
>
> FPU ?

"Sx" means no FPU.

> > Jason Burgon (GVFM) has a 386 as does Blair Campbell. Trixter (of
> > Oldskool.org) has an 8088. ATV (and 2 others??) on FASM's forum has a
> 286.
>
> COOL. Any 8086 around ?

8088 is basically a slower 8086 (8-bit external data bus instead of 16-bit) or such.

RayeR(R)

Homepage

CZ,
01.05.2008, 14:12

@ Rugxulo
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> > > 486 Sx/25 w/ 8 MB RAM (without CPUID support).
> >
> > FPU ?
>
> "Sx" means no FPU.

BTW as I readed, SX has FPU physically on silicon die but it was hardware disabled (even if this block was OK). It's cheaper to do one CPU line than two :) And 487 shouldn't be a coproccessor but full-featured 486DX with slightly different pinout :)

---
DOS gives me freedom to unlimited HW access.

Steve(R)

Homepage E-mail

US,
01.05.2008, 21:14

@ RayeR
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> BTW as I readed, SX has FPU physically on silicon die but it was hardware
> disabled (even if this block was OK). It's cheaper to do one CPU line than
> two :)

Not disabled, but defective. Intel's method was to have one production line, then test chips. Good FPU > sell as DX. Bad FPU > sell as SX. Similarly with speed - run chip at increasing frequencies, then label according to where errors begin.

> And 487 shouldn't be a coproccessor but full-featured 486DX with
> slightly different pinout :)

Correct.

RayeR(R)

Homepage

CZ,
02.05.2008, 01:15

@ Steve
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> Not disabled, but defective. Intel's method was to have one production
> line, then test chips. Good FPU > sell as DX. Bad FPU > sell as SX.
> Similarly with speed - run chip at increasing frequencies, then label
> according to where errors begin.

Do you think they had as many defective CPUs as they sold SX processors? I think that math coprocessor itself takes only small surface of silicon die, sure much smaller than cache. Probability of def. FPU would be smaller than def. cache. If intel would have so many defects they will not survive. And for flexibility of suplying variable damand of DX/SX CPUs it's good idea to be able to quickly turn DX to SX after silicon dies was made and just put different marking on it. Ad CPU speed - I agree.

---
DOS gives me freedom to unlimited HW access.

Steve(R)

Homepage E-mail

US,
02.05.2008, 09:24

@ RayeR
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> Do you think they had as many defective CPUs as they sold SX processors?

FPUs - yes.

> If intel would have so many defects they will not survive.

In the early 486 days, Intel had big troubles, and a lot of competition - but they moved ahead with the Pentium fast enough to save their world.

> And
> for flexibility of suplying variable damand of DX/SX CPUs it's good idea to
> be able to quickly turn DX to SX after silicon dies was made and just put
> different marking on it.

There was no demand for the SX before Intel started selling it - the major point of the 486 was having the CPU and FPU on the same die. Disabling the FPU, so as to be able to sell the chip for less, was definitely not part of the original plan. The sequence of events was 1) Find defects in FPU, 2) What do we do with this crap, 3) Try to sell what's left at a discount. No. 3 worked - low-cost PC makers liked it, sold SX machines as a cheaper upgrade from the 386.

RayeR(R)

Homepage

CZ,
02.05.2008, 16:04

@ Steve
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> In the early 486 days, Intel had big troubles, and a lot of competition -
> but they moved ahead with the Pentium fast enough to save their world.

I don't have any intel statistics, just seems to me as very bad efectivity. How many % they had thrashed with bad cache? Later they got idea to make celeron but there was not 486 without cache :)

> 2) What do we do with this crap, 3) Try to sell what's left at a discount.
> No. 3 worked - low-cost PC makers liked it, sold SX machines as a cheaper
> upgrade from the 386.

Yes, and then I would expect when they improved technologic process and had produced less bad SX it was still demanded by market bacause became popular as cheap CPU. So could turn some DX to SX if needed. I don't have any proof but do you have a proof that all SX ever made have really bad FPU? I think it just a speculation...

---
DOS gives me freedom to unlimited HW access.

Steve(R)

Homepage E-mail

US,
03.05.2008, 08:01

@ RayeR
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> I don't have any intel statistics, just seems to me as very bad
> efectivity.

Right - Intel had bad manufacturing problems at the time.

> How many % they had thrashed with bad cache?

I don't know.

> > 2) What do we do with this crap, 3) Try to sell what's left at a discount.
> > No. 3 worked - low-cost PC makers liked it, sold SX machines as a cheaper
> > upgrade from the 386.
>
> Yes, and then I would expect when they improved technologic process and
> had produced less bad SX it was still demanded by market bacause became
> popular as cheap CPU.

Right.

> So could turn some DX to SX if needed. I don't have
> any proof but do you have a proof that all SX ever made have really bad
> FPU? I think it just a speculation...

After the broken 486SXes started making money for Intel, they figured out how to make better 486DXes, and also a few 486es without the FPU. So they kept SX supplies up for a while longer, at lower cost and higher profit, until the Pentium replaced all 486es.

http://foldoc.org/index.cgi?query=486sx&action=Search

RayeR(R)

Homepage

CZ,
03.05.2008, 14:12

@ Steve
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> http://foldoc.org/index.cgi?query=486sx&action=Search

LOLz, to drill a hole through processor to reenable FPU, good joke :-D

---
DOS gives me freedom to unlimited HW access.

marcov(R)

02.05.2008, 23:19

@ Steve
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> There was no demand for the SX before Intel started selling it - the major
> point of the 486 was having the CPU and FPU on the same die. Disabling the
> FPU, so as to be able to sell the chip for less, was definitely not part
> of the original plan. The sequence of events was 1) Find defects in FPU,
> 2) What do we do with this crap, 3) Try to sell what's left at a discount.
> No. 3 worked - low-cost PC makers liked it, sold SX machines as a cheaper
> upgrade from the 386.

Note there was another cheaper 486, the SLC, which had halved memory bandwidth (like the 386sx), and its brother the DLC (similar, but with copro)

Steve(R)

Homepage E-mail

US,
03.05.2008, 08:03

@ marcov
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> Note there was another cheaper 486, the SLC, which had halved memory
> bandwidth (like the 386sx), and its brother the DLC (similar, but with
> copro)

Those were from Cyrix, intended as plugin replacements for the 386.

IBM also made a 486SLC, but it was based on the Intel SX, with added cache.

RayeR(R)

Homepage

CZ,
03.05.2008, 14:14

@ Steve
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> Those were from Cyrix, intended as plugin replacements for the 386.
>
> IBM also made a 486SLC, but it was based on the Intel SX, with added
> cache.

I've seen this from Texas Instruments (it had same 386 PGA) but still missing it in my CPU collection.

---
DOS gives me freedom to unlimited HW access.

Steve(R)

Homepage E-mail

US,
03.05.2008, 15:20

@ RayeR
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> > Those were from Cyrix, intended as plugin replacements for the 386.
> >
> > IBM also made a 486SLC, but it was based on the Intel SX, with added
> > cache.
>
> I've seen this from Texas Instruments (it had same 386 PGA) but still
> missing it in my CPU collection.

Texas Instruments made some of the Cyrix Cx486SLC chips, and later a TI486SLC based on the Cyrix design. IBM also made some of the Cx486SLC chips, in addition to its own Intel-based 486SLC (very nice chip, too, used in many IBM PCs and notebooks).

(Cyrix was a design house only - all manufacturing was contracted out).

marcov(R)

02.05.2008, 23:17

@ RayeR
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> > Not disabled, but defective. Intel's method was to have one production
> > line, then test chips. Good FPU > sell as DX. Bad FPU > sell as SX.
> > Similarly with speed - run chip at increasing frequencies, then label
> > according to where errors begin.
>
> Do you think they had as many defective CPUs as they sold SX processors? I
> think that math coprocessor itself takes only small surface of silicon die,
> sure much smaller than cache.

(the cache wasn't that big in those days. 8k or so, except later stepping DX4's that had 16k).

But both are right as far as I know. In the beginning they were a way to create a market the partially broken ones (just like the first Celeron's were P-II's with defective cache). But when yields go up, a special line (mendocinno in Celeron's case), or "good" CPUs are marked as the inferior product.

RayeR(R)

Homepage

CZ,
03.05.2008, 14:23

@ marcov
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> (the cache wasn't that big in those days. 8k or so, except later stepping
> DX4's that had 16k).

If you look on the silicon chip of uncovered 486, you will see quite large uniform area of cache which covers about 40% of chip surface. I'm sure that FPU block is smaller.

---
DOS gives me freedom to unlimited HW access.

marcov(R)

01.05.2008, 14:13

@ DOS386
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> > Anyway I have a 386SX notebook laying around so I can test himemx there
> > I tested himemx v3.32 on intel 386SX/25MHz
>
> Great :-) How much RAM ? FPU ? Bugs (early 80386's reportedly were
> extremely buggy) ? Please keep it. I have a few things that should
> occasionally get tested on an 80386.

I might have a 386SX 25 mobo somewhere with 8MB, no FPU (but it has a socket). Don't know if I still have a multi card to match it.

rr(R)

Homepage E-mail

Berlin, Germany,
02.05.2008, 16:43

@ DOS386
 

Good PC's needed (80386, 8086, 4004 :-D) report what you h

> Anyone else has 80386 ? Anyone has a 16-bit PC ? Preferably 8086 :hungry:

I recently sold my Acer AnyWare 386s laptop, but I still own the PC-alike palmtops Atari Portfolio (8088) and HP 200LX (80186).

Steve(R)

Homepage E-mail

US,
05.05.2008, 01:41

@ DOS386
 

4004

Here are some views of the first 4004 machine. Note the built-in narrow-carriage printer. (Monitor not shown).

http://www.intel4004.com/images/busi_calc.gif

http://www.intel.com/museum/archives/images/busicom.jpg

Rugxulo(R)

Homepage

Usono,
05.05.2008, 18:45

@ Steve
 

4004

> Here are some views of the first 4004 machine. Note the built-in
> narrow-carriage printer. (Monitor not shown).
>
> http://www.intel4004.com/images/busi_calc.gif
>
> http://www.intel.com/museum/archives/images/busicom.jpg

Wow, home computers began almost as glorified super calculators. And now they're home arcades, gaming machines, multimedia (e.g. tv) replacements, word processors, communications (somewhat replacing telephone and snail mail), database, etc.

Steve(R)

Homepage E-mail

US,
06.05.2008, 04:58

@ Rugxulo
 

4004

> Wow, home computers began almost as glorified super calculators. And now
> they're home arcades, gaming machines, multimedia (e.g. tv) replacements,
> word processors, communications (somewhat replacing telephone and snail
> mail), database, etc.

That was the calculator that the 4004 was designed for. Then Intel expanded it into the 8xxx chips, and the rest we know. It's truly amazing, isn't it? From one little project, all our fancy toys.

Back to index page
Thread view  Board view
15192 Postings in 1365 Threads, 250 registered users, 12 users online (0 registered, 12 guests)
DOS ain't dead | Admin contact
RSS Feed
powered by my little forum