8 Feb
2018
8 Feb
'18
10:18 p.m.
At some point I read an article or a transcript of an interview with an
Intel employee (or former employee) who had been involved with the design
of the 8085, describing how he had specified additional instructions over
those of the 8080, and they had been implemented in the silicon, but then
the decision was made to not document any of the new instructions other
than RIM and SIM.
I no longer recall which Intel employee that was, and can't find the
article or interview. Does anyone else remember that, and perhaps have a
copy?
Eric
8 Feb
8 Feb
11:56 p.m.
On 02/08/2018 07:18 PM, Eric Smith via cctalk wrote:
At some point I read an article or a transcript of an
interview with an
Intel employee (or former employee) who had been involved with the design
of the 8085, describing how he had specified additional instructions over
those of the 8080, and they had been implemented in the silicon, but then
the decision was made to not document any of the new instructions other
than RIM and SIM.
I no longer recall which Intel employee that was, and can't find the
article or interview. Does anyone else remember that, and perhaps have a
copy?
Do you mean Cort Allen? His email a couple of years ago was:
manofquest at aol.com
He wrote;
Actually, all of these instructions were 100% tested when they were
manufactured at Intel. The test program used to test these devices
contained all of these instructions and was written to do an extensive
test on ALL instructions in many combinations and order. So, unless the
8085 broke after it was shipped, these instructions were tested and were
working.
I know this because I was the Test Engineer at Intel that developed the
8085 Production Test Program. I wrote all the code for this test and it
ran on a Megatest @8000 test system.
9 Feb
9 Feb
8:22 a.m.
On 02/08/2018 11:56 PM, Chuck Guzis via cctalk wrote:
On 02/08/2018 07:18 PM, Eric Smith via cctalk wrote:
Also all of the competitors and second sources that made 8085s had the
added instructions.? The are indeed handy too.
Allison
At some point I read an article or a transcript
of an interview with an
Intel employee (or former employee) who had been involved with the design
of the 8085, describing how he had specified additional instructions over
those of the 8080, and they had been implemented in the silicon, but then
the decision was made to not document any of the new instructions other
than RIM and SIM.
I no longer recall which Intel employee that was, and can't find the
article or interview. Does anyone else remember that, and perhaps have a
copy?
Do you mean Cort Allen? His email a couple of years ago was:
manofquest at aol.com
He wrote;
Actually, all of these instructions were 100% tested when they were
manufactured at Intel. The test program used to test these devices
contained all of these instructions and was written to do an extensive
test on ALL instructions in many combinations and order. So, unless the
8085 broke after it was shipped, these instructions were tested and were
working.
I know this because I was the Test Engineer at Intel that developed the
8085 Production Test Program. I wrote all the code for this test and it
ran on a Megatest @8000 test system.
12:23 p.m.
On 02/09/2018 05:22 AM, allison via cctalk wrote:
Also all of the competitors and second sources that
made 8085s had the
added instructions.? The are indeed handy too.
Calmos (then later Tundra) Semiconductor actually documents them in
detail in their CA80C85 datasheet, but those were from the early 1980s.
What was interesting was that even through we were working pretty
tightly with Intel in 1976, not a word of the added instructions was
uttered by our sales guy, "Fast Eddie", who usually was pretty well
informed about the goings on over in Santa Clara. Bill Davidow was on
our BOD as well.
That's a shame; we could have used those instructions too.
--Chuck
1:29 p.m.
On 2/9/18 9:23 AM, Chuck Guzis via cctalk wrote:
Calmos (then later Tundra) Semiconductor actually
documents them in
detail in their CA80C85 datasheet, but those were from the early 1980s.
the info is still out there
http://saxelec.com/doc/T8000.pdf
1:57 p.m.
On Fri, Feb 9, 2018 at 11:29 AM, Al Kossow via cctalk <cctalk at classiccmp.org
wrote:
On 2/9/18 9:23 AM, Chuck Guzis via cctalk wrote:
For those that don't want to read through the entire datasheet,
https://electronicerror.blogspot.com/2007/08/undocumented-flags-and-instruc…
has a good summary.
Warner
Calmos (then later Tundra) Semiconductor actually
documents them in
detail in their CA80C85 datasheet, but those were from the early 1980s.
the info is still out there
http://saxelec.com/doc/T8000.pdf
2:43 p.m.
On 02/09/2018 10:57 AM, Warner Losh via cctalk wrote:
For those that don't want to read through the
entire datasheet,
https://electronicerror.blogspot.com/2007/08/undocumented-flags-and-instruc…
has a good summary.
RIM and SIM were always documented for the 8085. Not doing so would
have made it difficult to use the added features, such as "half" and
TRAP interrupts, in addition to bit-bang serial I/O.
In the Calmos document cited by Al, the discussion of added flags and
instructions begins on PDF page 39 and is much clearer.
--Chuck
3:26 p.m.
On 02/09/2018 12:23 PM, Chuck Guzis via cctalk wrote:
On 02/09/2018 05:22 AM, allison via cctalk wrote:
Interesting story the NEC 8085A was a reverse engineer and was faithful to
the then (pre 1980) hidden instructions.? NEC learned hard form the 8080
as it was not compatible (they belived the datasheet) and would lead to the
8080AFC (A version, fully compatable).
If anyone didn't that's a first as every part I could get did at least
after 1978.
I new of the instruction back in 78-79 when I got a Netronics Explorer85
(8085 powered with a short s100 bus).? I bought and built it to play with
8085 and the magic instructions.
IBM PC early also did that... other vendors of the 8088 didn't work in the
64K and early 256K IBM PCs due to a slight timing variation that
reputedly intel helped design in.? One gate fixed it but it was undocumented
thing you could getaway with only with Intel 8088s.
The industry was loaded with that the 6502 series also had that going on
as well as the 6809 and others.
Allison
Also all of the competitors and second sources
that made 8085s had the
added instructions.? The are indeed handy too.
Calmos (then later Tundra)
Semiconductor actually documents them in
detail in their CA80C85 datasheet, but those were from the early 1980s.
What was interesting was that even through we were working pretty
tightly with Intel in 1976, not a word of the added instructions was
uttered by our sales guy, "Fast Eddie", who usually was pretty well
informed about the goings on over in Santa Clara. Bill Davidow was on
our BOD as well.
That's a shame; we could have used those instructions too.
--Chuck
4:59 p.m.
On 02/09/2018 04:18 PM, Sean Conner wrote:
It was thus said that the Great allison via cctalk
once stated:
Not a processor I use much bt there is a version of the 6809 that has
additional modes.
Thats the Hitachi 63b09 wich is 6809E compatable save for it has an
extended
mode and more registers.
I use that as an example of compatibility or extensions another more
familiar
is the NEC V20 fits in the 8088 socket but has enhanced performance and
native
8080 emulation mode.
Allison
The industry was loaded with that the 6502 series
also had that going on
as well as the 6809 and others.
Do you have any information about undocumented
opcodes for the 6809?
-spc
5:41 p.m.
On 02/09/2018 01:59 PM, allison via cctalk wrote:
I use that as an example of compatibility or
extensions another more
familiar is the NEC V20 fits in the 8088 socket but has enhanced
performance and native 8080 emulation mode.
The V-series may be a good example of
why Intel didn't want to publicize
the added 8085 instructions.
Aside from 8080 emulation, the V-series also has some x86 "orphan"
instructions, such as the BCD string instructions that were never
carried over into any other x86 design.
It reminds me a bit of the 386 bit-string instructions that were
publicized early on, but never implemented AFAIK in any 80386.
--Chuck
5:53 p.m.
On Fri, Feb 9, 2018 at 3:41 PM, Chuck Guzis via cctalk <
cctalk at classiccmp.org> wrote:
The V-series may be a good example of why Intel
didn't want to publicize
the added 8085 instructions.
Maybe. What I'd heard from multiple sources was that they only wanted the
8085 to replace the 8080, so it was supposed to be "better" in terms of
being a lower-cost 8080 replacement, needing fewer support chips (except an
address latch, but that's cheaper than an 8228/8238), but they didn't want
it to have a better instruction set that might put it into sockets that
might otherwise get filled with an 8086/8088.
4:24 p.m.
On 02/09/2018 12:26 PM, allison via cctalk wrote:
IBM PC early also did that... other vendors of the
8088 didn't work
in the 64K and early 256K IBM PCs due to a slight timing variation
that reputedly intel helped design in. One gate fixed it but it was
undocumented thing you could getaway with only with Intel 8088s.
Hmm, must have been *very* early PCs, as mine had an AMD 8088 in it--but
wasn't AMD a licensee of the Intel part? But then, so was NEC.
I think IMSAI used the "different" NEC 8080s in their floppy controller,
as they were inexpensive (because of the incompatibility) and in the
particular application, it didn't matter.
I do recall very early 8085s having a "reset" bug, necessitating some
external glue, but it's been too long for me to remember the details.
--Chuck
2:59 p.m.
On Thu, Feb 8, 2018 at 9:56 PM, Chuck Guzis via cctalk <
cctalk at classiccmp.org> wrote:
On 02/08/2018 07:18 PM, Eric Smith via cctalk wrote:
That's not the interview I was thinking of, but it's definitely interesting!
At some point I read an article or a transcript
of an interview with an
Intel employee (or former employee) who had been involved with the design
of the 8085, describing how he had specified additional instructions over
those of the 8080, and they had been implemented in the silicon, but then
the decision was made to not document any of the new instructions other
than RIM and SIM.
I no longer recall which Intel employee that was, and can't find the
article or interview. Does anyone else remember that, and perhaps have a
copy?
Do you mean Cort Allen? His email a couple of years ago was:
manofquest at aol.com
2360
days inactive
2360
days old
13 comments
6 participants
participants (6)
-
aek＠bitsavers.org -
allisonportable＠gmail.com -
cclist＠sydex.com -
imp＠bsdimp.com -
spacewar＠gmail.com -
spc＠conman.org