HP35s slower than HP32SII?
Ralf Fritzsch
yesterday I wondered about the fact that a simple program ran about twice
as long on my HP35s compared to my HP32SII. The program was nothing
complicated, DSEing a loop 1000 times and within the body of the loop
decrementing a counter and summing up the inverses of the counter. Both
programs look exactly the same, the only difference is GTO W on the 32SII
vs. GTO W001 on the 35s.
Has anyone out there an explanation for this behaviour?
-------------------------------------------------------------------------
Ralf Fritzsch
Bundesanstalt fuer Wasserbau Federal Waterways Engineering and Research
Dienststelle Kueste Institute - Department Hamburg
-------------------------------------------------------------------------
Unix _IS_ user friendly - it's just selective about who its friends are.
-------------------------------------------------------------------------
Raymond Del Tondo
there were related threads on the forum of www.hpmuseum.org .
Since the hardware of the 32S/II and the 33s/35s are totally different,
you could expect some differences;-)
The Saturn-based machine has a small and somehow optimized 'OS'
with a restricted, but efficient 'programming language' .
Seems either the SPL with 6502 core is not the optimal machinery for a
calculator,
or the routines written in C/C++ are not optimized for performance, or
both...
Maybe from a code efficiency point of view,
it would have been better to stay with a CPU which was developed
especially for calculators, namely the Saturn,
even if it were a hardware simulation inside an ASIC or alike.
HTH
Raymond
"Ralf Fritzsch" <> schrieb im Newsbeitrag
news:5lc2mtF...@mid.individual.net...
JT
new models of HP calculators. Especially interesting was
the comparison between the new HP35s and the Hp33s.
The 33s is just about the fastest scientific ever made but
the 35s has much improved programming, etc. Looks like the
best of both worlds is split into two devices... I'd love to see
the 35s built around the hardware of the 33s.
JT
On Sep 19, 3:50 am, Fritz...@Hamburg.BAW.DE (Ralf Fritzsch) wrote:
> Hi there,
>
> yesterday I wondered about the fact that a simple program ran about twice
> as long on my HP35s compared to my HP32SII.
> Has anyone out there an explanation for this behaviour?
cyrille de Brebisson
couple of reasons there...
1: the 32sII CPU (saturn based) is Made to do deciaml floating point
operations (with registers designed to hold the number and build in decimal
function). the 35S CPU is a 8 bit micro with no dcb, so everything has to be
done with memory read/write all the time....
2: the 32SII was programmed in assembly by guys who were gurus! the 35S is
programmed in C on a CPU that is NOT designed to be programed in C and
therefore generates vastly inefficient assembly
3: the 33SII only supports reals, while the 35S also supports complex and
vectors, that adds complexity to the '+' operation
4: the 32sII CPU has direct access to all it's memory (512K), inclluding
internal ram (fast access) the 35s has only 64K of memory and rom is
switched all the time, causing slow down, in addition, the RAM where the
user program is stored is external and serialy accessed in Software, so
reading the user instructions is slow
I think that all of these together are some of the main contributors to the
slowness...
cyrille
"Ralf Fritzsch" <Frit...@Hamburg.BAW.DE> wrote in message
news:5lc2mtF...@mid.individual.net...
A.L.
<cyr...@hp.com> wrote:
>
>2: the 32SII was programmed in assembly by guys who were gurus! the 35S is
>programmed in C on a CPU that is NOT designed to be programed in C and
>therefore generates vastly inefficient assembly
That what?...
A.L.
jc.ran...@gmail.com
calculators. It's truly useful to me, looks great, and so I bought two
of them. I'll probably buy more as time goes on, especially if an
improved version is introduced.
keep up the good work - JCR
On Sep 19, 6:23 am, "cyrille de Brebisson" <cyri...@hp.com> wrote:
> hello,
>
> couple of reasons there...
> 1: the 32sII CPU (saturn based) is Made to do deciaml floating point
> operations (with registers designed to hold the number and build in decimal
> function). the 35S CPU is a 8 bit micro with no dcb, so everything has to be
> done with memory read/write all the time....
> 2: the 32SII was programmed in assembly by guys who were gurus! the 35S is
> programmed in C on a CPU that is NOT designed to be programed in C and
> therefore generates vastly inefficient assembly
> 3: the 33SII only supports reals, while the 35S also supports complex and
> vectors, that adds complexity to the '+' operation
> 4: the 32sII CPU has direct access to all it's memory (512K), inclluding
> internal ram (fast access) the 35s has only 64K of memory and rom is
> switched all the time, causing slow down, in addition, the RAM where the
> user program is stored is external and serialy accessed in Software, so
> reading the user instructions is slow
>
> I think that all of these together are some of the main contributors to the
> slowness...
>
> cyrille
>
> "Ralf Fritzsch" <Fritz...@Hamburg.BAW.DE> wrote in message
>
> news:5lc2mtF...@mid.individual.net...
>
>
>
> > Hi there,
>
> > yesterday I wondered about the fact that a simple program ran about twice
> > as long on my HP35s compared to my HP32SII. The program was nothing
> > complicated, DSEing a loop 1000 times and within the body of the loop
> > decrementing a counter and summing up the inverses of the counter. Both
> > programs look exactly the same, the only difference is GTO W on the 32SII
> > vs. GTO W001 on the 35s.
>
> > Has anyone out there an explanation for this behaviour?
> > -------------------------------------------------------------------------
> > Ralf Fritzsch
> > Bundesanstalt fuer Wasserbau Federal Waterways Engineering and Research
> > Dienststelle Kueste Institute - Department Hamburg
> > -------------------------------------------------------------------------
> > Unix _IS_ user friendly - it's just selective about who its friends are.
> > -------------------------------------------------------------------------- Hide quoted text -
>
> - Show quoted text -
A.L.
>Despite the flaws, I really like it, reminds me of the old HP
>calculators. It's truly useful to me, looks great, and so I bought two
>of them.
Oh... I understand: left for left hand, righ for right hand...
I agree that it "looks great" (vovever, looks less gerat if you
switch it on), but this is abuut all.
However:
a. Display is UGLY,
b. There is only one type of regression (linear). All 20 bucks
calculators have more. Moreover, all 20 bucks calculators have
statistics table based, where you can see and edit data table,
c. Having 30K memory without any way to save programs somewhere - on
PC, tape recoreder, flash drive is a nonsense.
Compare this calculator with CASIO fx-9860G. Casio costs about 60
bucks, the same as HP-35.
A.L.
cyrille de Brebisson
> c. Having 30K memory without any way to save programs somewhere - on
> PC, tape recoreder, flash drive is a nonsense.
the problem is that allowing comuniation with PC would bar the calcualtor
from a certain number of exams.... shame...
cyrille
msto...@yahoo.com
now for
awhile, so I thought I'd add my current views.
As JCR wrote, I really like it. The speed hasn't been an issue for me,
and I prefer the RPN style keystroke programming to that used
on the graphing calcs. My gripes are that the very top row of icons
in the display are nearly impossible to read (they fall in the shadow
of the bezel), it has definitely missed a few keystrokes, and I'm
waiting to see just what the battery life will be like.
On Monday, I needed to test a few different combinations of ball
endmill diameters and depths-of-cut vs width of cut combinations
that would normally have me looking around for a piece of scratch
paper to quickly diagram out and come up with the equations. Instead
I remembered the equation writer on the 38S and typed in:
R^2=(R-D)^2+(W/2)^2 and was able to easily try out different
sizes and depths until I found what I needed.
For the features it has, and more importantly those it lacks, I think
the
price should come down a bit. But I feel better leaving this one sit
out on the
shop floor than I did with my (much harder to replace and more
expensive) 41CX.
Dave Boyd
news:1190194625....@22g2000hsm.googlegroups.com:
> On Sep 19, 3:50 am, Fritz...@Hamburg.BAW.DE (Ralf Fritzsch) wrote:
>> Hi there,
>>
>> yesterday I wondered about the fact that a simple program ran about
>> twice as long on my HP35s compared to my HP32SII.
>
>> Has anyone out there an explanation for this behaviour?
>
> new models of HP calculators. Especially interesting was
> the comparison between the new HP35s and the Hp33s.
> The 33s is just about the fastest scientific ever made but
> the 35s has much improved programming, etc. Looks like the
> best of both worlds is split into two devices... I'd love to see
> the 35s built around the hardware of the 33s.
It is built around substantially the same hardware. The speed
differences are due to the new capabilities of the software running the
calc. You pay a speed price for the flexibility. For example, In the
32SII, and the 33S, all registers could contain only real numbers. Now,
in the 35S, they might contain a real, or a complex, or a binary integer.
So, probably (depending on how they implemented it), every register
access, every pop or push operation, etc, must now have some code
surrounding it to check for the data type in the register. Adding an
integer to another integer requires different code than adding a real to
a real; and adding an integer to a real, the integer must be converted to
a real first; etc, etc. So it runs a bit slower on the same programs,
even though the CPU is the same and it runs at the same speed as a 33S.
As for the 32SII, the CPU and software bear no relation apart from
functionality.
--
Dave Boyd
"If we hit that bullseye, the rest of the dominoes will fall
like a house of cards. Checkmate." -Capt. Zapp Brannigan, D.O.O.P.
Joel Kolstad
news:ib92f3pc2tmp76hp6...@4ax.com...
> a. Display is UGLY,
Subjective -- I think it's OK.
> b. There is only one type of regression (linear). All 20 bucks
> calculators have more. Moreover, all 20 bucks calculators have
> statistics table based, where you can see and edit data table,
You could program your 35s to do this, and it'd probably end up being just
about as nice or better than what those $20 machines do.
> c. Having 30K memory without any way to save programs somewhere - on
> PC, tape recoreder, flash drive is a nonsense.
This is a marketing problem (HP apparently felt they'd lose much of the
academic market if they included such a feature), not a technical problem.
> Compare this calculator with CASIO fx-9860G. Casio costs about 60
> bucks, the same as HP-35.
HP has always had a "premium" price tag associated with it. :-)
---Joel
A.L.
<JKolstad7...@yahoo.com> wrote:
>"A.L." <alew...@fala2005.com> wrote in message
>news:ib92f3pc2tmp76hp6...@4ax.com...
>> a. Display is UGLY,
>
>Subjective -- I think it's OK.
I don't
>
>> b. There is only one type of regression (linear). All 20 bucks
>> calculators have more. Moreover, all 20 bucks calculators have
>> statistics table based, where you can see and edit data table,
>
>You could program your 35s to do this, and it'd probably end up being just
>about as nice or better than what those $20 machines do.
>
Yes, I can program other types of regression. I could also program
tables. Byt why and for what?...
>> c. Having 30K memory without any way to save programs somewhere - on
>> PC, tape recoreder, flash drive is a nonsense.
>
>This is a marketing problem (HP apparently felt they'd lose much of the
>academic market if they included such a feature), not a technical problem.
>
They DON'T have academic market, anyway. And many people here
pointed out that HP targets engineering market, not academic market.
I would be pretty reluctant to key in i don't-know-how-long program
for varius regression models, knowing that there is no way to store
it and reload if necessary.
Mentioned CASIO has 1 MB frash ROM, and I can shift programs and
data to flash to be sure that stuff will be not gone if I replace
batteries. By the way, it has USB port, too...
>> Compare this calculator with CASIO fx-9860G. Casio costs about 60
>> bucks, the same as HP-35.
>
>HP has always had a "premium" price tag associated with it. :-)
>
Other big nonsense is "vectors". Totally useless.
A.L.
A.L.
Taking into account how popular is HP in schools, this is really
serious argument...
A.L.
Ralf Fritzsch
question.
Anyway, speed is no issue to me, since I have a lot of programmable
calculators (both HP, and TI) to choose from. But I like very much the
old-fashioned look and feel of this calculator, and especially the
jealous compliments of my colleagues :-)
Until the HP35s I used a HP15C for every day calculations, and very
quickly the HP35s took over this position.
Once again, thanks to all who answered!
richwood
Another difference I found is that a number placed in line in a
program loop rather than in a register slows down the 35s much more
than the 33s. It seems that the 35s handling of in line numbers is
much less efficient than the 33s is, possibly due to the additional
number types it handles.
Rich Wood
Charles Perry
>
> They DON'T have academic market, anyway. And many people here
> pointed out that HP targets engineering market, not academic market.
> I would be pretty reluctant to key in i don't-know-how-long program
> for varius regression models, knowing that there is no way to store
> it and reload if necessary.
The licensing exams for engineering in the US have gone from a "banned"
calculator list to an "approved list". This list is very, very short.
Currently the 33S is the only RPN calc on the list. The 35S will be on it
soon, imo.
<snip>
> Other big nonsense is "vectors". Totally useless.
>
> A.L.
In your opinion perhaps. Ask someone who needs that feature and you will
get a different answer.
Charles Perry P.E.
Miguel
> On Wed, 19 Sep 2007 12:32:55 -0700, "Joel Kolstad"
>
> >"A.L." <alewa...@fala2005.com> wrote in message
>
>
>
> Other big nonsense is "vectors". Totally useless.
>
> A.L.
How someone can make broad statements like these! How about trying to
program an accurate NPV and IRR analysis in a 32s or 32s II. That what
I did in my 35s with ease thanks to this useless feature and all those
useless indirect registers available. Now I just carry one calculator
because before I needed a 12c in addition to my 32s. Check the museum
forum to see what interesting things people are doing with this
calculator. And of course, I am aware of the shortcomings (I use it
every day), but I am getting a lot more benefits from my calculator
than inconveniences.
Regards,
Miguel
A.L.
<pipesan...@hotmail.com> wrote:
>
>"A.L." <alew...@fala2005.com> wrote in message
>news:8uo3f317dg5dv11k4...@4ax.com...
>> On Wed, 19 Sep 2007 12:32:55 -0700, "Joel Kolstad"
>> <JKolstad7...@yahoo.com> wrote:
>>
><snip>
>>
>> They DON'T have academic market, anyway. And many people here
>> pointed out that HP targets engineering market, not academic market.
>> I would be pretty reluctant to key in i don't-know-how-long program
>> for varius regression models, knowing that there is no way to store
>> it and reload if necessary.
>
>The licensing exams for engineering in the US have gone from a "banned"
>calculator list to an "approved list". This list is very, very short.
>Currently the 33S is the only RPN calc on the list. The 35S will be on it
>soon, imo.
Interesting, why?... What is wrong with using HP35 or more advanced
calculator?... Maybe beacuse exams in the USA check memory not
knowledge.
A.L.
A.L.
wrote:
>On Sep 19, 11:12 pm, A.L. <alewa...@fala2005.com> wrote:
>> On Wed, 19 Sep 2007 12:32:55 -0700, "Joel Kolstad"
>>
>> <JKolstad71HatesS...@yahoo.com> wrote:
>> >"A.L." <alewa...@fala2005.com> wrote in message
>>
>>
>>
>> Other big nonsense is "vectors". Totally useless.
>>
>> A.L.
>
>
>How someone can make broad statements like these! How about trying to
>program an accurate NPV and IRR analysis in a 32s or 32s II. That what
>I did in my 35s with ease thanks to this useless feature and all those
>useless indirect registers available. Now I just carry one calculator
>because before I needed a 12c in addition to my 32s. Check the museum
>forum to see what interesting things people are doing with this
>calculator. And of course, I am aware of the shortcomings (I use it
>every day),
Inetersting.... Using every day... For what?...
A.L.
Charles Perry
of a calculator to think for you.
Also to keep people from copying problems into the calc and sending them
several time zones away where others are beginning their tests.
Charles Perry P.E.
Charles Perry
??? For calculations perhaps?! For many of us, a calculator is a tool that
we use in our job, not a toy. I use my 35s nearly every day (unless I
happen to grab my 48GX first) for normal engineering calculations. Since I
work mostly in electric power, I use the complex number capabilities
extensively. The 35S is nearly the ideal number cruncher for what I need.
Anything more complicated and it is off to the simulation tools.
Charles Perry P.E.
bokubob
> wrote:
> >How someone can make broad statements like these! How about trying to
> >program an accurate NPV and IRR analysis in a 32s or 32s II. That what
> >I did in my 35s with ease thanks to this useless feature and all those
> >useless indirect registers available. Now I just carry one calculator
> >because before I needed a 12c in addition to my 32s.
On Sep 20, 7:26 pm, A.L. <alewa...@fala2005.com> wrote:
> Inetersting.... Using every day... For what?...
Did you read his post? He mentioned that he used vectors for NPV (Net
Present Value) and IRR (Initial Rate of Return), and that now he
doesn't need to carry his 12c (a financial calculator)..... So, I'd
have to guess something related to finance?
-Jonathan
JYA
How so?
One is run using a 4-bits processor with 64 bits registers ; while the
other is a 8 bit processor, with 2 16 bits working registers.
The only advantage of the 6502 is the enormous amount of addressing mode.
Other than the keyboard the hardware of those two machines couldn't be
any more different !
Jean-Yves
--
They who would give up an essential liberty for temporary security,
deserve neither liberty or security (Benjamin Franklin)
bokubob
>>> I'd love to see the 35s built around the hardware of the 33s.
Dave Boyd wrote:
>> It is built around substantially the same hardware. The speed
JYA wrote:
> Other than the keyboard the hardware of those two machines couldn't be
> any more different !
Actually, I think Dave was referring to the 33s and the 35s, in reply
to the post he quoted by JT. I think you mean the 32SII is different
from the 35s, yes?
-Jonathan
Dave Boyd
57g2000hsv.googlegroups.com:
> JT Wrote:
>>>> I'd love to see the 35s built around the hardware of the 33s.
>
> Dave Boyd wrote:
>>> It is built around substantially the same hardware. The speed
>
> JYA wrote:
>> Other than the keyboard the hardware of those two machines couldn't be
>> any more different !
>
> Actually, I think Dave was referring to the 33s and the 35s, in reply
> to the post he quoted by JT.
That was my intent, yes.
JYA
>
> Actually, I think Dave was referring to the 33s and the 35s, in reply
> to the post he quoted by JT. I think you mean the 32SII is different
> from the 35s, yes?
>