Calculus

[goyoko]

Hello...

I have a question about calc.
I wish put in the main program a potencial function y=a*x^b but when I use math.h library with the pow() function the memory code or ROM memory it's used untill 89\%. How can I do the same but useless too memory.

Greating
CPB.
___________________________
This message was ported from CCS's old forum
Original Post ID: 10024

[Sherpa Doug]

:=Hello...
:=
:=I have a question about calc.
:=I wish put in the main program a potencial function y=a*x^b but when I use math.h library with the pow() function the memory code or ROM memory it's used untill 89\%. How can I do the same but useless too memory.
:=
:=Greating
:=CPB.

What types of numbers are you using? Could b be an integer? If so just use a loop. If b is not an integer could it be one of several fixed values? Maybe a switch case structure to solve for each of those values?

___________________________
This message was ported from CCS's old forum
Original Post ID: 10025

[goyoko]

:=:=Hello...
:=:=
:=:=I have a question about calc.
:=:=I wish put in the main program a potencial function y=a*x^b but when I use math.h library with the pow() function the memory code or ROM memory it's used untill 89\%. How can I do the same but useless too memory.
:=:=
:=:=Greating
:=:=CPB.
:=
:=What types of numbers are you using? Could b be an integer? If so just use a loop. If b is not an integer could it be one of several fixed values? Maybe a switch case structure to solve for each of those values?
:=

Hi.. Values are float. I'm covert a int16 variable to float variable to display in LCD.
___________________________
This message was ported from CCS's old forum
Original Post ID: 10026

[R.J.Hamlett]

:=Hello...
:=
:=I have a question about calc.
:=I wish put in the main program a potencial function y=a*x^b but when I use math.h library with the pow() function the memory code or ROM memory it's used untill 89\%. How can I do the same but useless too memory.
:=
:=Greating
:=CPB.
Seriously, float arithmetic, is a very heavy user of ROM space. This effect is less of a problem with larger chips (for instance on the 18F252, I solve a fairly complex set of arithmetic, involving power, log, sin etc., and still have a reasonable amount of space. The space used also declines on these chips, where the more efficient memory access code, and the presence of the hardware multiplier help a lot.
One solution that works well if you have a limited input range, is to use scaled integers, instead of float values. With a bit of care, this can help a lot. I notice you mention (in your other post), using float with a printf. This can really cost in terms of RPM useage. Dealing with a similar problem, I converted the result of my arithmetic to an integer by multiplying by 100, then printed this to a string in memory (using sprintf), and manually added the decimal point. This used less than 3/4 the space of simply printing the float value.

Best Wishes
___________________________
This message was ported from CCS's old forum
Original Post ID: 10028

[Sherpa Doug]

:=Hi.. Values are float. I'm covert a int16 variable to float variable to display in LCD.

I avoid floats at all costs. Scaled integers such as Mr Hamlett describes are the way to go with small chips. As others on this board have heard, I measure ships in integer millimeters and only convert to meters for output, which I do a string manipulation rather than numerical manipulation.
If your only reason for using a float is to print to a LCD, you can definately do better with ints.

___________________________
This message was ported from CCS's old forum
Original Post ID: 10039

[Mircea]

Recent versions of CCS compiler offer %w format specifier.

[ckielstra]

Thanks for making the effort to answer a question on this forum.
But do you really think the original poster is still waiting after 6,5 years ... ???

[John P]

If he's making a PIC do floating point math--he might be.

[Mircea]