Different speeds for HP-41CX and HP-41CV

[jedu]

Hi

After some time trying to get a HP-41 calculator I had the opportunity to get two: a HP-41CX halfnut and HP-41CV fullnut.

I just checked if they where more or less alike in terms of speed and there is a big difference executing 69!:

See: https://youtu.be/T-tmZWTPivA

I really don't know the reason... I have been checking in Internet and I haven't found any good reason for this to happen...

Any idea?

Best regards

Eduardo

[hth]

The time difference you see is the HP-41CX having to search longer to find the FACT instruction.

When you type a function, the HP-41 has to look it up. First it searches catalog 1 (user programs), then it searches catalog 2 (plug-in modules), finally it goes to catalog 3 (builtin functions) where it finds FACT.

The HP-41CX comes with 2 modules that need to be searched, the HP-41CV has an empty catalog 2 so it gets past faster.

Depending on what user programs you have, it may also affect the time it takes.

A better test is to enter 69 FACT into a program, then run it on both machines.

Håkan

[jedu]

(12-14-2016 05:43 PM)hth Wrote:  The time difference you see is the HP-41CX having to search longer to find the FACT instruction.

(...)

A better test is to enter 69 FACT into a program, then run it on both machines.

Håkan

Hi Håkan

You are absolutely right!

See it at

https://youtu.be/5xWAkNY5Ndk

I cite your answer. Minimal program: 69 FACT

Best regards and thank you

Eduardo

[aurelio]

and if you try with the 41C, I think is faster

see here

and from that site:


"HP-41 2.8 HP-41CV 2. 5 HP-41CX 2.6 HP-41CL (50x) 64.7

It is not quite clear why the 41C is faster than the CV and CX. Maybe branching to labels takes longer if more memory has to be searched for a label."

[hth]

(12-14-2016 08:04 PM)aurelio Wrote:  and if you try with the 41C, I think is faster

see here

and from that site:


"HP-41 2.8 HP-41CV 2. 5 HP-41CX 2.6 HP-41CL (50x) 64.7

It is not quite clear why the 41C is faster than the CV and CX. Maybe branching to labels takes longer if more memory has to be searched for a label."

There should not be a much of a difference.

The clock speed may differ slightly due to tolerances in the components. If a program is not packed, there may be differences due to invisible nulls. If one had compiled branches and the other did not, there may be differences (run twice to get it right). Program alignment (on register boundaries) may have some very minor influence, and so may different ROM versions.

Otherwise they execute the same instructions internally. Poll vectors are not normally called during program execution unless there are attached peripherals that wants attention.

The amount of available memory does not affect the speed of program runs at. It compiles label distances, and searching for a label takes place within the current program,.

There are some instructions that depend on the outside environment and may differ, like PSE and XEQ/GTO global labels and XROM calls, but no such instruction is used in the test program.

Håkan

[Jim Horn]

As mentioned, the tolerance of components give a tolerance in CPU speed. The 41's processor clock speed is determined by a LC oscillator, not a quartz one or other precision design.

This can be demonstrated by doing the BEEP function on a number of different 41s. Most will sound about the same but there will be some with audibly higher or lower pitch than the others.

For what it's worth, my early 41C (#1932A00186) had a lower pitch BEEP than any other '41 I ever encountered. Perhaps the earlier ones ended up slightly slower?

[Gene]

You know you MUST now record it and post an .mp3 or .wmv file of the beep, right ?

[aurelio]

(12-16-2016 08:51 PM)Gene Wrote:  You know you MUST now record it and post an .mp3 or .wmv file of the beep, right ?

It should be interesting collect mp3s from all the 41 owners and determinate up to them the clock of the processors

Actually I met a problem with the clock of one of my 41c, maybe different to the average: it does not allow the usage of certain modules.
I've planned to swap one of the capacitorsas from the service manual, I must find the time.

[Geoff Quickfall]

Then there is the 41CX versus 41CL beta board comparison:

Video one is a data base search and display, calc on left CL, calc on right CX:

https://youtu.be/ZK7VPqdyx2U

Video two is a data base search with a great circle formulae execution. The data base search looks up the Airport input and retrieves the Lat/Lon from the data base in extended memory, prompts for Ground Speed, then runs it through the GC routine to determine distance, track and estimated time on route:

https://youtu.be/37BBEY4AbQs

Geoff

[Garth Wilson]

(12-17-2016 05:58 PM)aurelio Wrote:  

(12-16-2016 08:51 PM)Gene Wrote:  You know you MUST now record it and post an .mp3 or .wmv file of the beep, right ?

It should be interesting collect mp3s from all the 41 owners and determinate up to them the clock of the processors

Actually I met a problem with the clock of one of my 41c, maybe different to the average: it does not allow the usage of certain modules.
I've planned to swap one of the capacitorsas from the service manual, I must find the time.

In the 1980's, I ordered the speed-up kit from EduCalc for my 41cx. It had a tiny switch to switch between normal speed and nearly double, IIRC. Before installing it, I read in its instructions that it should not be used with extended memory, HPIL, or merged modules. Since I had, and regularly used, all of these, I sent the kit back and did not install it.