Intel strikes back with a parallel x86 design

wrote:

keith writes:


On Mon, 10 Oct 2005 00:06:35 +0000, Jason Lee Eckhardt wrote:

I guess as long as software only used the BIOS routines, and avoided
hardware programming, things would run. But, the BIOS routines were
so hideously slow, that alot of software avoided them.

Come on! At 4.77MHz, BIOS ROM wasn't any slower than DRAM, (which also
had the refresh penalty). Remember, it was an 8-bit bus, with memory on
the ISA bus.


The problem with the BIOS was bad code (IIRC, at least the display routines
and serial code were best avoided), not slow ROM access...

You're right:

Serial code did not really exist at all in the BIOS (i.e. polled only!),
whereas the display bios actually did work, but the performance was
really bad if you wanted to do anything that required scrolling, or any
other kind of in-frame-buffer movement:

My own terminal emulator code became much faster after I wrote a full
replacement that did everything in a set of linked list memory buffers,
only copying the actual result to the frame buffer when idle.

Terje
--
-
"almost all programming can be viewed as an exercise in caching"

In article ,
Jason Lee Eckhardt wrote:
In article ,
keith wrote:
On Mon, 10 Oct 2005 00:06:35 +0000, Jason Lee Eckhardt wrote:

In article ,
keith wrote:
On Fri, 07 Oct 2005 14:49:47 +0000, Jason Lee Eckhardt wrote:



snip


IIRC, the Jr had no DMA controller at all. It did have an 8253 PIT,
though. I don't recall if it used that for refresh.

What did the memoy refresh? ...just asking, I don't remember.

I suspect it was the 8253 PIT (programmable interval timer) I mentioned,
though I cannot say for certain without digging out the PC Jr Tech Ref
manual. Of course, a small bit of extra circuitry would be
necessary in order to generate the dummy read cycles (whereas the
original PC used a combination of one 8237 DMA channel and one
8253 timer channel).


After digging out the Jr Tech Ref, it appears that the 8253 was
not used for refresh. Instead, an interesting combination of the
6845 CRT controller and a custom gate array did the job. The sync
signals from the 6845 feed the gate array, which itself drives the
RAS and CAS of the RAMs. The gate array is also part of the video
graphics subsystem (providing the enhanced graphics modes, RAMDAC,
etc).

On Tue, 11 Oct 2005 18:56:16 +0000, Jason Lee Eckhardt wrote:

In article ,
Jason Lee Eckhardt wrote:
In article ,
keith wrote:

After digging out the Jr Tech Ref, it appears that the 8253 was
not used for refresh. Instead, an interesting combination of the
6845 CRT controller and a custom gate array did the job. The sync
signals from the 6845 feed the gate array, which itself drives the
RAS and CAS of the RAMs. The gate array is also part of the video
graphics subsystem (providing the enhanced graphics modes, RAMDAC,
etc).

Now that you mention this, I do remember that the display controller did
the refresh. Anything that does a sequential access to memory will
refresh memory as a side effect. Thanks.

--
Keith

In comp.sys.ibm.pc.hardware.chips keith wrote:
I didn't like any of the RadioShaft computers other than the original
TRS80.

I loved the Model 100 and 102 portables...

--
Nate Edel http://www.cubiclehermit.com/

"I do have a cause, though. It is Obscenity. I'm for it." - Tom Lehrer

(MSCHAEF.COM) wrote in
:


My understanding is that IBM's original offerings were these:

CGA
- 80x25 and 40x25 text with an 8x8 character cell and color attributes
- 640x200 Monochrome Graphics
- 320x200 4-color Graphics Black + (White/Cyan/Magenta or
Red/Yellow/Green)

MDA
- 80x25 text with a 14x9 character cell

These adapters had different base addresses and could be run
on the same machine.


I'm not sure when it came along. But I remeber(wasn't that old at the
time... maybe about) we had this Tandy 1000, which had "Tandy 16 color" I
just remeber I thought it ruled because the graphics on it was better than
a lot of others kids Dad's computers. I think that was later though,
because it wasn't for a few years until we bought a real computer game for
it. Anyone know what time frame in here that Tandy 16 came out?

I remeber sitting at the dos prompt, trying to figure out how to play the
game "driver.exe" I could see that file listed on the dos 3.2 floppy, and
really wanted to figure out how to play it... :

Grant Schoep wrote:
(MSCHAEF.COM) wrote in
:


My understanding is that IBM's original offerings were these:

CGA
- 80x25 and 40x25 text with an 8x8 character cell and color attributes
- 640x200 Monochrome Graphics
- 320x200 4-color Graphics Black + (White/Cyan/Magenta or
Red/Yellow/Green)

MDA
- 80x25 text with a 14x9 character cell

These adapters had different base addresses and could be run
on the same machine.



I'm not sure when it came along. But I remeber(wasn't that old at the
time... maybe about) we had this Tandy 1000, which had "Tandy 16 color" I
just remeber I thought it ruled because the graphics on it was better than
a lot of others kids Dad's computers. I think that was later though,
because it wasn't for a few years until we bought a real computer game for
it. Anyone know what time frame in here that Tandy 16 came out?

I remeber sitting at the dos prompt, trying to figure out how to play the
game "driver.exe" I could see that file listed on the dos 3.2 floppy, and
really wanted to figure out how to play it... :


I think the timeframe was mid to late 80's, after the XT came out. From
memory, it had an 80186 CPU (or 80188) which could be led to some
performance enhancements if you used assembler. It also supported 768k
(six) instead of 640k due to the controllers being in a better place.

--
-bill davidsen )
"The secret to procrastination is to put things off until the
last possible moment - but no longer" -me

On Wed, 26 Oct 2005 20:53:50 +0000, Bill Davidsen wrote:

Grant Schoep wrote:
(MSCHAEF.COM) wrote in
:


My understanding is that IBM's original offerings were these:

CGA
- 80x25 and 40x25 text with an 8x8 character cell and color attributes
- 640x200 Monochrome Graphics
- 320x200 4-color Graphics Black + (White/Cyan/Magenta or
Red/Yellow/Green)

MDA
- 80x25 text with a 14x9 character cell

These adapters had different base addresses and could be run
on the same machine.



I'm not sure when it came along. But I remeber(wasn't that old at the
time... maybe about) we had this Tandy 1000, which had "Tandy 16 color" I
just remeber I thought it ruled because the graphics on it was better than
a lot of others kids Dad's computers. I think that was later though,
because it wasn't for a few years until we bought a real computer game for
it. Anyone know what time frame in here that Tandy 16 came out?

I remeber sitting at the dos prompt, trying to figure out how to play the
game "driver.exe" I could see that file listed on the dos 3.2 floppy, and
really wanted to figure out how to play it... :


I think the timeframe was mid to late 80's, after the XT came out. From
memory, it had an 80186 CPU (or 80188) which could be led to some
performance enhancements if you used assembler. It also supported 768k
(six) instead of 640k due to the controllers being in a better place.

The 80186/8 weren't fully PC compatable. The IBM PC devlopers, in their
infinite wisdom, squated on "reserved" interrupts that were reserved for
the 80186/8, relegating the thing to the embedded market. ...too bad, it
was a nice processor. Some tried to make it compatable, but there were
always problems with anything that wrote to the hardware, which was
more than "common".

--
Keith

keith wrote:
On Wed, 26 Oct 2005 20:53:50 +0000, Bill Davidsen wrote:


Grant Schoep wrote:

(MSCHAEF.COM) wrote in
:



My understanding is that IBM's original offerings were these:

CGA
- 80x25 and 40x25 text with an 8x8 character cell and color attributes
- 640x200 Monochrome Graphics
- 320x200 4-color Graphics Black + (White/Cyan/Magenta or
Red/Yellow/Green)

MDA
- 80x25 text with a 14x9 character cell

These adapters had different base addresses and could be run
on the same machine.



I'm not sure when it came along. But I remeber(wasn't that old at the
time... maybe about) we had this Tandy 1000, which had "Tandy 16 color" I
just remeber I thought it ruled because the graphics on it was better than
a lot of others kids Dad's computers. I think that was later though,
because it wasn't for a few years until we bought a real computer game for
it. Anyone know what time frame in here that Tandy 16 came out?

I remeber sitting at the dos prompt, trying to figure out how to play the
game "driver.exe" I could see that file listed on the dos 3.2 floppy, and
really wanted to figure out how to play it... :


I think the timeframe was mid to late 80's, after the XT came out. From
memory, it had an 80186 CPU (or 80188) which could be led to some
performance enhancements if you used assembler. It also supported 768k
(six) instead of 640k due to the controllers being in a better place.


The 80186/8 weren't fully PC compatable. The IBM PC devlopers, in their
infinite wisdom, squated on "reserved" interrupts that were reserved for
the 80186/8, relegating the thing to the embedded market. ...too bad, it
was a nice processor. Some tried to make it compatable, but there were
always problems with anything that wrote to the hardware, which was
more than "common".

Good point, although the Tandy 1000 did run most PC software. It run a
Tandy version of MS-DOS, which let you run 768k of memory. I never tried
to run anything else on the system, other things came along.

The 80186 was nice in many way for embedded, just as you noted. IIRC
there was an interrupt controller and some 1 bit parallel port which
could be taught to do RS-232 just by a driver to the correct voltage.

Speaking of which, wasn't the 8086 5 and 12v and the 8018[68] single 5v?
The memory is going...

--
-bill davidsen )
"The secret to procrastination is to put things off until the
last possible moment - but no longer" -me

On Thu, 27 Oct 2005 16:49:49 +0000, Bill Davidsen wrote:

keith wrote:
On Wed, 26 Oct 2005 20:53:50 +0000, Bill Davidsen wrote:


Grant Schoep wrote:

(MSCHAEF.COM) wrote in
:



My understanding is that IBM's original offerings were these:

CGA
- 80x25 and 40x25 text with an 8x8 character cell and color attributes
- 640x200 Monochrome Graphics
- 320x200 4-color Graphics Black + (White/Cyan/Magenta or
Red/Yellow/Green)

MDA
- 80x25 text with a 14x9 character cell

These adapters had different base addresses and could be run
on the same machine.



I'm not sure when it came along. But I remeber(wasn't that old at the
time... maybe about) we had this Tandy 1000, which had "Tandy 16 color" I
just remeber I thought it ruled because the graphics on it was better than
a lot of others kids Dad's computers. I think that was later though,
because it wasn't for a few years until we bought a real computer game for
it. Anyone know what time frame in here that Tandy 16 came out?

I remeber sitting at the dos prompt, trying to figure out how to play the
game "driver.exe" I could see that file listed on the dos 3.2 floppy, and
really wanted to figure out how to play it... :


I think the timeframe was mid to late 80's, after the XT came out. From
memory, it had an 80186 CPU (or 80188) which could be led to some
performance enhancements if you used assembler. It also supported 768k
(six) instead of 640k due to the controllers being in a better place.


The 80186/8 weren't fully PC compatable. The IBM PC devlopers, in their
infinite wisdom, squated on "reserved" interrupts that were reserved for
the 80186/8, relegating the thing to the embedded market. ...too bad, it
was a nice processor. Some tried to make it compatable, but there were
always problems with anything that wrote to the hardware, which was
more than "common".

Good point, although the Tandy 1000 did run most PC software. It run a
Tandy version of MS-DOS, which let you run 768k of memory. I never tried
to run anything else on the system, other things came along.

Right, as long as every access to hardware went through a *DOS* interrupt,
the Tandy DOS worked. If it went to BIOS or diddled with bits directly,
not so good. That's clear by your 768K number, though a *compatable*
could have 704K with no problems (until much later when the EGA came out).

The 80186 was nice in many way for embedded, just as you noted. IIRC
there was an interrupt controller and some 1 bit parallel port which
could be taught to do RS-232 just by a driver to the correct voltage.

I don't remember any one-bit parallel ports, but it did have
integrated address decode logic (another issue in the DOS map), that I
suppose could be used for such. It also had integrated DMA and timers,
along with a few instruction improvements (stack sorts of thigns IIRC).

Speaking of which, wasn't the 8086 5 and 12v and the 8018[68] single 5v?
The memory is going...

Sheesh! ;-) Interesting note though, the 80186 and 80188 were origianlly
the same die, with only a bondout option difference. I think that was the
first time that was done.

--
Keith