I share the thoughts of Thomas.
The precision of arithmetic operations is 48 bits with a 32 bit mantissa. This corresponds to a precision of slightly more than 9 decimal digits is for the main target group good.
Who has real problems with it and why?
There exist more computing platforms in the world than Linux.
Variablen mit Nachkommastellen
Forumsregeln
Bitte beachte die Forumsregeln!
Bitte beachte die Forumsregeln!
Re: Variablen mit Nachkommastellen
Hello Thomas, Vleeuwen,
There is not real issue in my mind as I also said in my summary.
The RoboPro help suggest that also +, - etc. commands can work w/ floating point. This is not the case. It means you can not simply translate the logic of a RoboPro program using the +/- command from integer to floatingpoint. Its is just a matter of clarification.
The other is the claim from Quadro that there are issues w/ flaotingpoint conversions etc. As I said in my summary there is no evidence for this claim.
Hope this clears up the confusion Joachim
PS: Linux or not Linux this is not the question. Today's hardware floating point units support the IEEE 754 format and nothing else (except some exotic machines), independept of hte operating system
There is not real issue in my mind as I also said in my summary.
The RoboPro help suggest that also +, - etc. commands can work w/ floating point. This is not the case. It means you can not simply translate the logic of a RoboPro program using the +/- command from integer to floatingpoint. Its is just a matter of clarification.
The other is the claim from Quadro that there are issues w/ flaotingpoint conversions etc. As I said in my summary there is no evidence for this claim.
Hope this clears up the confusion Joachim
PS: Linux or not Linux this is not the question. Today's hardware floating point units support the IEEE 754 format and nothing else (except some exotic machines), independept of hte operating system
Re: Variablen mit Nachkommastellen
Can you give a simple example.
I am using a "int to float" converter (ch 12) and after that I am doing some float calculations.
This works.
I am using a "int to float" converter (ch 12) and after that I am doing some float calculations.
This works.
Re: Variablen mit Nachkommastellen
Hello Vleeuwen,
All the orange datapathes all work as said in my inital post, confirming UMUeller´s statement.
What does not work are the "blue" commands execpt the one with '='. The Help is not clear and when you add a datapath input to the blue command symbol then it offers the option to select 48-Float also for '+' or '-'.
But when you then run the program it complains that the instructions can not work with float.
regards Joachim
All the orange datapathes all work as said in my inital post, confirming UMUeller´s statement.
What does not work are the "blue" commands execpt the one with '='. The Help is not clear and when you add a datapath input to the blue command symbol then it offers the option to select 48-Float also for '+' or '-'.
But when you then run the program it complains that the instructions can not work with float.
regards Joachim
Re: Variablen mit Nachkommastellen
I can reproduce what your describe.
-> "No attached input can handle message '-'.
-> "No attached input can handle message '+'.
It looks like that the RoboPro-designer did not implemented Floating point handlers for these commands.
In fact you only need the "=" command to store FP-results in a variable, this in combination with the Operators.
This work around will solve the problems.
Step counting is normally only done with integers.
-> "No attached input can handle message '-'.
-> "No attached input can handle message '+'.
It looks like that the RoboPro-designer did not implemented Floating point handlers for these commands.
In fact you only need the "=" command to store FP-results in a variable, this in combination with the Operators.
This work around will solve the problems.
Step counting is normally only done with integers.