10 Aug
2006
10 Aug
'06
11:59 a.m.
I'm working on adding double integer (32-bit) instructions to the HP 1000
simulator under SIMH. The 1000 F-Series computer came standard with these
instructions, and I have documentation on the opcodes used. The E-Series
did not have them, but there was a "specials group" product -- HP number
93585A -- that added firmware support for these instructions. However, the
opcodes were different from the F-Series opcodes to avoid conflict with
other E-Series microcode options.
Does anyone have information on the opcodes used in the 93585A product?
I have found a reference to a manual, HP number 93585-90007, "93585A Double
Integer Firmware Installation and Operating Manual," but I cannot locate a
copy.
-- Dave
10 Aug
10 Aug
10:09 p.m.
Virtually *every* HP21MX, 'E-series' machine will have the FFT microcode
that implements several long integer formats (2 and 3 word).
----- Original Message -----
From: "J. David Bryan" <jdbryan at acm.org>
To: "Classic Computing List" <cctech at classiccmp.org>
Sent: Thursday, August 10, 2006 11:59 AM
Subject: HP 1000 double integer instructions info
I'm working on adding double integer (32-bit)
instructions to the HP 1000
simulator under SIMH. The 1000 F-Series computer came standard with these
instructions, and I have documentation on the opcodes used. The E-Series
did not have them, but there was a "specials group" product -- HP number
93585A -- that added firmware support for these instructions. However,
the
opcodes were different from the F-Series opcodes to avoid conflict with
other E-Series microcode options.
Does anyone have information on the opcodes used in the 93585A product?
I have found a reference to a manual, HP number 93585-90007, "93585A
Double
Integer Firmware Installation and Operating Manual," but I cannot locate a
copy.
-- Dave
11 Aug
11 Aug
10:44 a.m.
On 10 Aug 2006 at 22:09, Bob Shannon wrote:
Virtually *every* HP21MX, 'E-series' machine
will have the FFT microcode
that implements several long integer formats (2 and 3 word).
"FFT" = "Fast Fourier Transform?" I'm unfamiliar with that
product, and I
can't find it in any of the 1000 MEF ordering guides that I have. Do you
have a part number?
Are you possibly thinking of "FFP" = "Fast FORTRAN Processor?" That
was a
common microcode option that added two- and three-word floating-point
instructions, but no integer instructions.
-- Dave
6:45 p.m.
Sorry for the typo, I did mean FFP, Fast Fortran.
Standard on F-series, and I've only ever found one 2113 E-series CPU that
did not have it installed (it had been highly customized with user
microcode).
----- Original Message -----
From: "J. David Bryan" <jdbryan at acm.org>
To: "General Discussion: On-Topic Posts Only" <cctech at classiccmp.org>
Sent: Friday, August 11, 2006 10:44 AM
Subject: Re: HP 1000 double integer instructions info
On 10 Aug 2006 at 22:09, Bob Shannon wrote:
Virtually *every* HP21MX, 'E-series'
machine will have the FFT microcode
that implements several long integer formats (2 and 3 word).
"FFT" = "Fast Fourier Transform?" I'm unfamiliar with that
product, and I
can't find it in any of the 1000 MEF ordering guides that I have. Do you
have a part number?
Are you possibly thinking of "FFP" = "Fast FORTRAN Processor?" That
was a
common microcode option that added two- and three-word floating-point
instructions, but no integer instructions.
-- Dave
12 Aug
12 Aug
5:39 p.m.
On 11 Aug 2006 at 18:45, Bob Shannon wrote:
Sorry for the typo, I did mean FFP, Fast Fortran.
Standard on F-series, and I've only ever found one 2113 E-series CPU that
did not have it installed....
On a related topic, are the RTE-6/VM microcode ROMs commonly present in
found E- and F-Series machines? There are two sets of three; the HP part
numbers are:
92084-80001, 2, 3 (or) -80001, 80102, 80103 (or) -80007, 8, 9
and:
92084-80004, 5, 6.
-- Dave
7:03 p.m.
J. David wrote...
On a related topic, are the RTE-6/VM microcode ROMs
commonly present in
found E- and F-Series machines? There are two sets of three; the HP part
numbers are:
I have a set. Of all the HP 21MX E/F machines I've had run through my
basement (at least 30+ machines), only one had the RTE-6/VM microcode.
Odd, but true. Oh, they are on bitsavers, so we could burn additional sets
easy enough.
Jay West
13 Aug
13 Aug
12:58 p.m.
RTE IV/B firmware is more commonly found in my experience.
RTE 6VM firmware is not ~rare~, but it is a little uncommon.
There is also a huge collection of microcode that is associated with
a 'Booster Microcode' board that extends the ALU / datapath.
This microcode and hardware is used to accelerate FFT's, and is
found in E-series CPU's used in HP's vibration control and analysis
systems (like a HP 54410B processor).
As far as I'm aware, there is no documentation for this configuration.
Is anyone aware of documentation for the 54410 variant of the 2113?
----- Original Message -----
From: "J. David Bryan" <jdbryan at acm.org>
To: "General Discussion: On-Topic Posts Only" <cctech at classiccmp.org>
Sent: Saturday, August 12, 2006 5:39 PM
Subject: Re: HP 1000 double integer instructions info
On 11 Aug 2006 at 18:45, Bob Shannon wrote:
Sorry for the typo, I did mean FFP, Fast
Fortran.
Standard on F-series, and I've only ever found one 2113 E-series CPU that
did not have it installed....
On a related topic, are the RTE-6/VM microcode ROMs commonly present in
found E- and F-Series machines? There are two sets of three; the HP part
numbers are:
92084-80001, 2, 3 (or) -80001, 80102, 80103 (or) -80007, 8, 9
and:
92084-80004, 5, 6.
-- Dave
6:51 p.m.
On 13 Aug 2006 at 12:58, Bob Shannon wrote:
RTE IV/B firmware is more commonly found in my
experience.
RTE 6VM firmware is not ~rare~, but it is a little uncommon.
Interesting, thanks. I had thought that virtually everyone would have
upgraded from IVB to 6/VM at the time. Sounds as though more shops stuck
with IVB than I had imagined.
-- Dave
11 Aug
11 Aug
5:02 a.m.
J. David Bryan schrieb:
I'm working on adding double integer (32-bit)
instructions to the HP 1000
simulator under SIMH. The 1000 F-Series computer came standard with these
instructions, and I have documentation on the opcodes used. The E-Series
did not have them, but there was a "specials group" product -- HP number
93585A -- that added firmware support for these instructions. However, the
opcodes were different from the F-Series opcodes to avoid conflict with
other E-Series microcode options.
Does anyone have information on the opcodes used in the 93585A product?
There is an
opcode list in the appendix of the MACRO-1000 assembler
manual. This manual is donloadable from the RTE section at HP.
--
Holger Veit
11:08 a.m.
On 11 Aug 2006 at 11:02, Holger Veit wrote:
There is an opcode list in the appendix of the
MACRO-1000 assembler
manual.
I hadn't thought to look there. Thanks!
Appendix C, "HP 1000 Computer Instruction Set," lists the double integer
instructions for the E-Series as in the range 105320-105337, which are the
opcodes for the SIS firmware on the F-Series. Seems reasonable.
However, the first page of that appendix states, "Special or custom
firmware (for example, double integer and third-party firmware) is not
included in table C-1." So I'm unsure of the origin of those opcodes.
The reference to the "specials" product 93585A came in a "Real-Time
Interface" (an Interex publication) article from January 1994, written by
Esther Heller and Don Pottenger of HP DSD. The date of the MACRO/1000
manual is December 1992. So I guess that the manual is giving the opcodes
of the 93585A microcode, even though it states that it isn't.
If anyone has any corroborating evidence, perhaps from an old HP buying
guide, I'd feel more confident.
-- Dave
13 Aug
13 Aug
2:57 p.m.
J. David Bryan schrieb:
On 11 Aug 2006 at 11:02, Holger Veit wrote:
105200-105237 are the FFP codes Note that some of these codes are
overlayed by the Double integer instrs.
Actually, on the F the DBI instrs are scattered over multiple locations;
for date code >1920, the F seems to have these instrs built in -
apparently for some sort of forward compatibility to the upcoming
A-series systems:
105014 .DAD
105034 .DSB
105054 .DMP
105074 .DDI
105114 .DSBR
105134 .DDIR
105203 .DNG
105204 .DCO
105210 .DIN
105211 .DDE
105212 .DIS
105213 .DDS
105214 .NGL
There is an opcode list in the appendix of the
MACRO-1000 assembler
manual.
I hadn't thought to look there. Thanks!
Appendix C, "HP 1000 Computer Instruction Set," lists the double integer
instructions for the E-Series as in the range 105320-105337, which are the
opcodes for the SIS firmware on the F-Series. Seems reasonable.
However, the first page of that appendix states,
"Special or custom
firmware (for example, double integer and third-party firmware) is not
included in table C-1." So I'm unsure of the origin of those opcodes.
C-1 lists the mnemonics; this is infact incomplete. C-2, in constrast
contains much more opcodes, several of which I haven't yet identified
completely. You find:
- the mentioned DBI instrs,
- FPP,
- FFP,
- RTE-IV/B EMA,
- RTE-6/VM VMA,
- DS/1000,
- SIS,
- RTE/6 OS,
- VIS,
- DMS
- and all the standard instrs + EIG.
The reference to the "specials" product
93585A came in a "Real-Time
Interface" (an Interex publication) article from January 1994, written by
Esther Heller and Don Pottenger of HP DSD. The date of the MACRO/1000
manual is December 1992. So I guess that the manual is giving the opcodes
of the 93585A microcode, even though it states that it isn't.
Looks as if. You might also look at the 5950-3767 M-E-F CE handbook
(bitsavers) on pp 11-22 ff. which also lists the SIGNAL/1000 opcodes.
--
Holger
6:51 p.m.
On 13 Aug 2006 at 20:57, Holger Veit wrote:
Note that some of these codes are overlayed by the
Double integer
instrs. Actually, on the F the DBI instrs are scattered over multiple
locations;
Yes, part are in the FP area, the rest in the FFP area. I'd guess that's
because the E-Series microcode for both became much smaller when it was
adapted for the F-Series hardware floating point box. So HP reused those
emptier areas (and ROMs), rather than allocate a new block from the "HP
Reserved" microcode spaces.
C-2, in constrast contains much more opcodes, several
of which I
haven't yet identified completely.
Which ones? Note that the A-Series added a lot of new instructions for the
new hardware, e.g., code and data separation (CDS) instructions.
You might also look at the 5950-3767 M-E-F CE handbook
(bitsavers) on
pp 11-22 ff. which also lists the SIGNAL/1000 opcodes.
Had already. :-) It has the F-Series opcodes for the double integer
instructions, but not the E-Series ones.
-- Dave
6559
days inactive
6562
days old
11 comments
5 participants
participants (5)
-
bshannon＠tiac.net -
holger.veit＠ais.fraunhofer.de -
holger.veit＠iais.fraunhofer.de -
jdbryan＠acm.org -
jwest＠classiccmp.org