Amazing! An HP-32SII is faster than an HP-35s



#24

I've always preferred my old HP-32SII over my HP-35s for many reasons: better keys, better display, more solid feel, better primary key assignment such as separate STO and RCL keys etc. However, I've always assumed that at least the HP-35s would be faster with it's more modern processor, but such is not the case. I wrote a simple benchmark program to loop on the calculation of natural logarithm and exponential functions, and found out that the HP-35s takes 50% longer to run this benchmark. All this despite the fact that the batteries in the HP-32SII last much longer. IMO, the HP-35s not only is *not* an improvement over the much older Pioneer calc, it is actually a step backwards. How sad ! :-(


#25

Interesting. But I think it gets more complicated, in terms of speed, if you do other tests you will get results which favor the 35s...


#26

Can you provide an example ?


#27

Sure...

1) Storing 2400 numbers into registers. 35s is much faster than the 33s or 32s2.

2) Sorting 500 numbers in registers.

etc.

:-)


#28

Not to forget the super high speed when it comes to draining batteries in the 35s:-)

(Has been mentionned somewhere here, but suits well as a knock-down...)

#29

We discussed this years ago. As the 35s has to consider more datatypes, it is expected to have slower execution speed. The 32SII is slower than the 32S, btw.

The 35s is not a step backwards. It just has too many bugs :-(.

#30

Quote:
I wrote a simple benchmark program to loop on the calculation of natural logarithm and exponential functions

Michael, part of this may be the looping commands, ISG/DSE. I don't have a 35s handy, but I found out a long time ago that the looping commands on the 33s were not terribly efficient, in terms of speed (I even put a postit note in my 33s manual: "Don't use ISG/DSE"). If you have some significant loops, it is much faster on the 33s to code them without using ISG/DSE, even though it requres more instructions. The 33s itself is a pretty fast machine, but the looping commands are not.

I don't know if the 35s inherited this looping slowness from the 33s.

I do know that the 32sii looping commands are fast.

#31

Speed is not the only criterion to assert improvement. I wrote this Cash Flow Analysis program using vectors and indirect addressing. Something like this cannot be programmed on a HP 32sii (which I love, Pioneers are my favorites).

Regards,

Miguel


#32

Actually, the HP-32SII also has indirect addressing capabilities and the vector capabilities of the HP-35s are very limited (3 dimension maximum, no cross-product). The HP-35s does indeed have a few additional capabilities such as a catalogue of physics constants, but not enough IMO to offset its other deficiencies. Certain operations, such as rectangular-polar coordinate conversions are simpler and more convenient on the 32SII than the 35s.

Edited: 3 Aug 2011, 10:38 a.m.


#33

Two egregious errors in the 35s were the elimination of a "simple" rectangular to polar method (35s requires the imaginary number business and even worse was the totally messed up and elaborate base conversion which is next to useless for normal things.


#34

Quote:
[...] 35s requires the imaginary number business [...]
That would have been ok for infrequent use, imo. Unfortunately, and I guess that was the point the developer simply forgot, polar coordinates cannot be easily composed/decomposed from and to stack, respectively - one has to type them in :-/.

After all those years, I still cannot understand why there never was a new revision of the firmware, or even better, a 35sII w/o those silly cursor keys.


#35

Yes, exactly the problem. Here you have an advanced calculator, with gbs of memory, full programmabiliy, and yet you cannot pull R-P/P-R to/from the stack? Just ridiculous!

It is a great mystery that HP has not reworked this machine. They put real effort into straightening out the 33s. Why not the 35S?


#36

Maybe a larger production run and fewer sales than with the 33s. It would be interesting to know production dates of recently sold calculators. I wouldn't be too surprised if all were produced in 2007 and 2008 at last.

#37

The point was that the 35s allows to do things and to create applications that are not possible or very difficult on the 32sii, even if it is buggy as everyone knows, that is enough not to dismiss it completely. It is really nice to have some financial programs in my scientific calculator, so not to carry a 12c around. The 35s allows having this done very easily so I carry it everywhere letting me to keep my old pioneers save at home. Just check the link.

Vector capabilities are limited but combined with indirect addressing and the amount of memory in this calculator, as in this case, then a lot of very fun programs may be done.

Just my two cents...

Miguel

Edited: 3 Aug 2011, 7:51 p.m.


#38

Quote:
It is really nice to have some financial programs in my scientific calculator

Then use the 32sii with the TVM program listed in its manual. And if you ever need to do work in other number bases, you'll find it's a breeze on the 32sii but you'll probably take a hammer to the calculator if you try to do it on the 35s.

I'd use a 33s before a 35s, and the 32sii tops both of them. But that's just my opinion, and we're all entitled to one.


#39

I use this and this instead. The first is not in the manual (cannot be done in the 32sii, either) and the second is more accurate. TVM is not the only trick the 35s can do, financially ;-)

Miguel.


Edited: 3 Aug 2011, 8:54 p.m.


#40

Thing is, both the 32SII and 35s are scientific calculators, and that's how I use them. Anyway, I don't do a lot of fancy programming on either of those calculators; instead, I use my HP-50g.

But getting back to the topic of my orginal post, namely relative speed of these two calcs, I pulled up an example that is shown in the manuals for both calculators, finding the root of a cubic polynomial using the solver. Again, the 32SII is faster than the 35s, so it seems that in many real world ways the 35s is not an improvement.


#41

Michael,

While I like the 32Sii it is near the bottom of my list when I grab for a calculator so maybe I should just stay out of this :-). However, while it is true that the 32Sii is faster than the 35S in many areas this is largely due (as discussed earlier) to the greater number of data types that the 35S can handle.

So, the question becomes, aside from speed how else can these calculators be compared? Let's start by accepting the fact that the 35S has bugs that are problematic for some (so far I have not been effected by them) and that this in and of itself may make the 35S unusable for many. For those it may be that the 32Sii or the 33S are the only choices.

For me, however, programming capability is a big draw. I play with these things to program them. It has been many years since I have needed the "higher math" capability of any of my calculators (I think the last time was when I took my Astronomy Comprehensive Exams to graduate from college and that was more years ago than I am willing to share--I will say that the 41 I used back then was in it's relative infancy).

In the area of programming the 35S has many nice additions over the 32Sii and even the 33S.

1. A less cluttered keyboard thanks to the sloping face keys. At least I find it easier to use the keyboard on the 35S than I ever did on the 32/33 (leaving aside the really silly placement of STO as a shifted function).

2. Much more memory than the 32Sii. For some programs this is essential although I love trying to squeeze every last byte out of a program to make it fit in the 32.

3. Many more memory registers, although you have to address them indirectly. Depending on what you are trying to do this can be a huge plus.

4. Two indirect addressing register (I and J) instead of one. The 32/33 do get a little back here since, with them, the indirect register (i) is separate from the 26 primary registers while with the 35S the I and J registers are part of the 26 primary register set.

5. Can jump to line numbers. For example GTO A100. This allows much better utilization of the available memory than is offered by the 33S which is limited to 26 labels. Good luck trying to use all of the 32K available with 26 labels. I guess one could have lots (and LOTS) of equations. With the 35S one can have 26 separate programs labeled A-Z whereas this is not possible with the 32/33 for anything but the most trivial programs. At an extreme with the 35S one could even have multiple programs with a single alpha label.

6. The ENTER key is in the right place (this is a dig at the 33S and obviously does not refer to the 32Sii which also has it's ENTER key in the right place).

7. The arrow keys are primary which is a bonus when single stepping. In this case the 33S beats the 32Sii as it also has it's arrow keys in the primary plane.

I might be able to think of other features that are improved on the 35S but my point here is not to try to convince you that the 35S is better (as noted earlier it may not even be usable for you) but to show that it does have certain advantageous features. Whether your requirements make use of these features is for you to decide and from what you said it would seem that the 32 suits your needs better and is the better solution for you.

Finally, as I stated earlier, the 32Sii is one of the LAST calculators I reach for. This is also true of the 35S. My calculators of choice are (somewhat in order) the 41, 42, 48 (S/X, G/X), and (believe it or not) the 39GS which I like playing with for a completely different set of reasons. The WP34S is also becoming a go-to machine for me (Thanks Pauli, Walter, and Marcus!) and, oh yes, I like the 17bii for it's solver. All these machines have one thing in common: Separate programs (functions in the 17bii) that can be named by multi-character labels that make it clear what the program/function is supposed to do. Maybe I am getting old but trying to remember what the program labeled 'A' is supposed to do just seems like too much trouble.

Cheers,

Marwan


#42

Quote:
Let's start by accepting the fact that the 35S has bugs that are problematic for some (so far I have not been effected by them)

You've never needed to checksum a program (for archival purposes e.g.)? That is one bug I think just about everyone who programs the 35s will have hit :-)

Is it a show stopper? No. However, it does indicate a carelessness of implementation that is unforgivable in what is supposed to be a precision instrument. Lets face it, long gone are the days when calculators producing dubious results is tolerable. They are ubiquitous and cheap. They should just work.

I'd even go as far as claiming that most if not all of the 35s bugs are indicative of such carelessness :-(


That all said, I actually quite like the 35s: lots of memory, decent keyboard, okay display. It isn't, however, my grab for calculator. To the extent I gave mine to one of my children for their entertainment.


- Pauli


#43

Yes, I am aware of the checksum bug, and yes, I guess it is one that does "concern" me. But as you say, not a show stopper. I guess, having programmed on so many machines that do not offer this feature, I never really missed it.

Although, again, I agree with you that the bugs in the 35S show a lamentable carelessness, what I find worse is that those same bugs, to the best of my knowledge, have NEVER been fixed! That is truly a sad commentary on how much importance the developers place on their work. It would have been wonderful if the calculator was flashable and HP provided ongoing updates to cure it of it's ills.


#44

If it were flashable, the 34s would have used that as a basis instead :-)


- Pauli

#45

Quote:
1. A less cluttered keyboard thanks to the sloping face keys. At least I find it easier to use the keyboard on the 35S than I ever did on the 32/33 (leaving aside the really silly placement of STO as a shifted function).
Yes indeed, the arrangement of the functions/keys is way better on the 35s. I love the 32SII and got used to it, but the key assignment is somewhat strange. STO being a shifted function was no show-stopper to me. I don't use it often.
#46

The 33s did that, too, without messing up all the useful functionality of the32sii. In fact the 33s improved a few things, once the bugs were fixed--which HP did make sure to do.

#47

I have a HP-35s, but for everyday calculations I always find myself using unconsciouslessly either my old HP-11C or my newer (but old styled) HP-12C.

Who knows....

-- Antonio


#48

Antonio,

I find myself in the nearly the same situation (minus the 12C). My trusty old 11C, S/N from May '82 (and on my 3rd or 4th set of batteries!), is still my first choice for most calculations though my new crop of 34S machines (one at home, one at work) are what I am reaching for when I have to do HEX or logical stuff.

I must say that I love the complex math on the 35s. I find it more intuitive than the 42S (though I only have free42). I too miss the R->P and P->R functions but Pauli's published workarounds help out there.

I have an emulator for 15C on my iPad but find the complex math somewhat of a challenge to keep straight with the (effectively) limited stack (even with the parallel imaginary stack). The 34S takes a bit to get used to with the complex math but the 8-level stack sure helps.

I really can not get my knickers in a knot over the issues with the 35S. Probably because I don't use it for much programming. The keyboard is worlds ahead of the 33S and the display is positively lovely. If the form factor were smaller (and maybe landscape!) and it was flashable, what a machine it could become! Sigh!


Possibly Related Threads...
Thread Author Replies Views Last Post
  An amazing day: Giving a talk at HP about their calculators Geir Isene 9 1,597 12-16-2013, 06:14 PM
Last Post: aurelio
  How much faster is the HP Prime than the HP 50g ? Michael de Estrada 13 1,313 11-13-2013, 07:38 AM
Last Post: DeboT
  HP 35S: Resistors in Series or Parallel (Portred from HP-33E) Eddie W. Shore 2 458 10-14-2013, 05:29 AM
Last Post: R. Pienne
  Interesting Base Conversions - Porting a 1975 HP 25 Program to the HP 35S Eddie W. Shore 1 462 10-13-2013, 07:49 PM
Last Post: BruceH
  HP 35s y^x vs. yrootx Richard Berler 8 892 09-28-2013, 12:32 PM
Last Post: Andrew Feist
  HP 32sii Fault Carlos Estinos 9 945 09-26-2013, 01:06 PM
Last Post: Csaba Tizedes (Hungary)
  Joys of eBay: HP-32S, HP-32SII, HP-42S, HP-16C, ... Sasu Mattila 7 883 09-23-2013, 04:39 PM
Last Post: Julián Miranda (Spain)
  HP 35s Richard Berler 1 342 09-19-2013, 01:48 AM
Last Post: Thomas Radtke
  HP-35s Richard Berler 21 1,840 09-12-2013, 11:15 PM
Last Post: Dale Reed
  OT: Reversing Sinclair's amazing 1974 calculator hack - half the ROM of the HP-35 Egan Ford 6 1,010 09-02-2013, 10:54 PM
Last Post: Thomas Klemm

Forum Jump: