Why no SOLVE & INTEGRATE commands on the 41C?

[David Hayden]

Why doesn't the 41C have built-in SOLVE and INTEGRATE commands? These were introduced with the 34C if I understand correctly. The 41C has so many more functions, so why didn't they include solve and integrate?

I see that root finders are in the Standard pac and Math pac, and the Math pac also has a numerical integrator.

I have a feeling that there's some interesting history here....

Dave

[Maximilian Hohmann]

Hello!

Maybe because HP wanted to sell expensive software modules together with their calculators? Around the time when the HP-41 was on the market, the letters "HP" were an acronym for "high price", at least in my part of the word. Ultimately this is the reason why they lost the education market to the competition.

Regards
Max

[Dave Britten]

Gotta be able to sell those Math Pacs somehow, right?

(I say that mostly in jest. My more realistic guess is that it was just a matter of mainframe ROM space, and programming features that could support such applications were a better long-term investment.)

[Gene]

Good question...

1) Would the mcode to do the solve and integrate would have fit in the OS? If not, what mcode and functions would you have dropped?

2) Assuming a YES answer to #1, since the HP 41 byte table is known...what functions would you drop to include them?


Purely speculation here... no wrong answer.

[Valentin Albillo]

(01-14-2020 07:29 PM)David Hayden Wrote:  Why doesn't the 41C have built-in SOLVE and INTEGRATE commands? These were introduced with the 34C if I understand correctly. The 41C has so many more functions, so why didn't they include solve and integrate?

Probably because of ROM-space limitations. The HP-41C also lacks such things as RCL arithmetic and (pseudo) Random-number generation, factorial for non-integer arguments (i.e.: gamma function), etc.

Anyway, the Advantage ROM does provide microcode versions of Solve and Integrate (the ones implemented in the HP-34C and HP-15C) in a very convenient 12 Kb ROM, so the functionality is available for the 41C all the same.

Quote:I see that root finders are in the Standard pac and Math pac, and the Math pac also has a numerical integrator.

Most algorithms in the Standard and Math pac are pretty subpar and slow, so I suggest giving them a wide berth.

If you want a good RPN Integrate for the HP-41C, consider this (3-page PDF):

       HP-41C 3-point Gaussian Integration

and for a good, versatile RPN Root finder for the HP-41C, have a look at this (4-page PDF):

       HP-41C Finding Roots of Equations

Regards.
V.

[Sylvain Cote]

Did you look at these HP-41 Math oriented ROM's?

• Advantage Pac from HP
• Solve & Integrate ROM from Ángel Martin
• Equation Solver ROM from Ángel Martin
• SandMath 4x4 from Ángel Martin (with parts from Jean-Marc Baillard)
• ... and several others here
  

[rprosperi]

There's also an important marketing consideration. The 34C was still a relatively new machine, and it had both of these features. There no doubt was plenty of market demand from folks that needed these features but didn't need (or could not afford) all that a 41 offered. The 41 could have included them (notwithstanding issues like ROM space and available opcodes) but that would have essentially 'obsoleted' the 34C less than a year after introduction.

Look closely at all the fancy feature analysis brochures of the period comparing the various models. Every model has some features none of the other models offered, thus letting HP offer a wider product family at many price points. In the 70s/80s, these were important marketing considerations demanded by retailers that wanted to have offerings at many price points, and the smart manufacturers tried hard to supply something at as many price points as they could.

By offering these features in a HP Application Pac, it offered parity with the 41C (plus even more) but at about twice the price.

While it seems like an odd oversight looking back, I think they had pretty clear vision looking forward.

[hth]

Lots of good points are given above.

I want to add that it is highly likely that the HP-41 was originally budgeted for having a 16K OS as can be seen from the use of page 0-2, then 3 empty and 4 being the diagnostic ROM. Thus, they originally set aside 16K for the machine but in the end only used 12K, leaving page 3 unused (until the 41CX).

Most likely they thought the would need that space but perhaps found that they could get away with 12K if they dropped a few features. Given that the HP-41 is so versatile and expandable, they probably decided to step back a bit and saved a 40KBit ROM chip from being inserted into each produced HP-41 unit.

Fitting it into opcode table would not have been too hard, it has several rather dubious instructions, ADV, CLRG, DEC, OCT are good candidates to be left out. There are also a couple of unused slots, though I do not know how easy it would be to make use of them.

[Csaba Tizedes]

(01-14-2020 11:43 PM)Valentin Albillo Wrote:  The HP-41C also lacks such things as (...) etc.

Linear regression... (Greatest fault)

Cs.

[Valentin Albillo]

(01-15-2020 06:08 AM)Csaba Tizedes Wrote:  

(01-14-2020 11:43 PM)Valentin Albillo Wrote:  The HP-41C also lacks such things as (...) etc.

Linear regression... (Greatest fault)

That was included in my "etc.", as well as hyperbolics, show mantissa, self-tests, ..., etc.

Regards.
V.

[J-F Garnier]

My personal feeling (or maybe did I read something about it) is that the 41C was introduced as the first member of the new series 40, as there were the series 20 and 30 with several models of different capabilities. Just think that the 41C had 64 registers only (same amount than the later HP15C). The initial marketing plan was (maybe) to develop a complete series 40 with higher performance models.

The 41C was then so a success and instead of introducing a 42C, 44C, etc, HP developed extension modules (Time, X-functions, I/O, ...), and introduced the 41CV (just more RAM) and the 41CX (Time and X-functions integrated, and a few extra new functions) without change so much the 41C architecture.

We had to wait for the Advantage module (around 1985, very late in the 41C lifetime) to get the proper Solve and Integrate features on the HP-41C.

Just speculations, unless we could get a direct feedback of the HP people of that time.

Edited: Yes, I was sure I read something about it. In his book "A Guide to HP Handheld...", Wlodek wrote "The HP-41 model ... was to be the simplest model of a series, and indeed the HP-41 was first planned as a fairly simple model".

J-F

[Ángel Martin]

(01-15-2020 04:46 AM)hth Wrote:  I want to add that it is highly likely that the HP-41 was originally budgeted for having a 16K OS as can be seen from the use of page 0-2, then 3 empty and 4 being the diagnostic ROM. Thus, they originally set aside 16K for the machine but in the end only used 12K, leaving page 3 unused (until the 41CX).

IIRC the X-Func/Mem was introduced more or less at the same time as the 15C, or maybe earlier. At that point the trend to use extension modules really made sense, given the success of the mainframe. Later on came the CX, and then page#3 was used for the XF and CAT extensions, thus no room left for other stuff. The 41Advantage Pac was how they dealt with the crunch.

(01-15-2020 04:46 AM)hth Wrote:  Fitting it into opcode table would not have been too hard, it has several rather dubious instructions, ADV, CLRG, DEC, OCT are good candidates to be left out. There are also a couple of unused slots, though I do not know how easy it would be to make use of them.

I'd love to figure out how to activate those unused slots (the infamour "W comes to mind...)

Best,
ÁM

[Dave Britten]

(01-15-2020 04:46 AM)hth Wrote:  Fitting it into opcode table would not have been too hard, it has several rather dubious instructions, ADV, CLRG, DEC, OCT are good candidates to be left out. There are also a couple of unused slots, though I do not know how easy it would be to make use of them.

No way! I actually find a use for those two now and then. And back when the 41 was new, octal-based machines like the PDP 11 were still fairly common.

[Andres]

(01-14-2020 07:40 PM)Maximilian Hohmann Wrote:  Hello!

Maybe because HP wanted to sell expensive software modules together with their calculators? Around the time when the HP-41 was on the market, the letters "HP" were an acronym for "high price", at least in my part of the word. Ultimately this is the reason why they lost the education market to the competition.

Regards
Max

The code for a root finder was provided in the Standard Applications booklet which came with the HP-41C. And examples of code for solving integrals using Simpson's Rule or similar methods were easy to find, including HP literature (i.e.: HP 25 Applications Manual), so the code and methods were readily available for anyone interested in such problems. The Application Pacs modules were a separate product indeed, but I don't think HP decided to drive their sales by reducing the native HP-41C function set, which was extensive and complete at the time. Just my opinion,

[edryer]

(01-15-2020 02:06 PM)Dave Britten Wrote:  

(01-15-2020 04:46 AM)hth Wrote:  Fitting it into opcode table would not have been too hard, it has several rather dubious instructions, ADV, CLRG, DEC, OCT are good candidates to be left out. There are also a couple of unused slots, though I do not know how easy it would be to make use of them.

No way! I actually find a use for those two now and then. And back when the 41 was new, octal-based machines like the PDP 11 were still fairly common.

Agreed, Octal was low level predominant and remained so for at last a few more years until hex (and 16-bit) architectures were commonplace.

[In 1979 I have no doubt the PDP-11 was the predominant mini however already in decline with the VAXen.]

[Gene]

Cmon! I did my assembly work on the PDP-11 in that timeframe too, so OCT was helpful haha.

010102 MOV R1,R2

Of course OCT isn't very useful with that one is it?

JMP addresses, yes.

[Dave Britten]

And octal still lingers today in Unix permissions. The chmod calculator I wrote for the 42S and 65 uses both of the DEC and OCT conversion functions.

If you buy/build an Altairduino or PiDP and do any front-panel programming, you might find it useful too.

[hth]

(01-15-2020 02:06 PM)Dave Britten Wrote:  

(01-15-2020 04:46 AM)hth Wrote:  Fitting it into opcode table would not have been too hard, it has several rather dubious instructions, ADV, CLRG, DEC, OCT are good candidates to be left out. There are also a couple of unused slots, though I do not know how easy it would be to make use of them.

No way! I actually find a use for those two now and then. And back when the 41 was new, octal-based machines like the PDP 11 were still fairly common.

I suppose that I am too young, I started with 6502 and it was hex. I never dealt with octal apart from the ancient HP-41 NOMAS list files from HP. Octal is pure pain for me.

Anyway, the point is that there are a couple of dubious instructions that we perhaps could have lived fine without.

[hth]

(01-15-2020 01:47 PM)Ángel Martin Wrote:  I'd love to figure out how to activate those unused slots (the infamour "W comes to mind...)

Best,
ÁM

That can probably be figured out, for the 41CL I suppose? What would you use it for?

I once dreamt that I should make a Forth sub-system for the 41 and allow (public) Forth words to be accessed like ordinary HP-41 instructions. In programs they could be represented by one of those unused opcodes as a prefix before some Forth-token number that gives the actual word.
Then I woke up and realized that when we are about 5 MCODE programmers left, how many would program their system in Forth? 0, 1 or 2? Lots of work for little gain as reality is today...

[Thomas Okken]

Octal was alive and well in the early '80s. I remember seeing core dumps from the university's CDC-175 that were pages full of octal numbers.

In fact, I still use it today, in escape sequences in C strings!