Post by thread:[PIC]:MPLAB floating point in C

Does anyone have math libraries for using floating point variables in c. I am working on a PIC and I have seen microchip's AN575. That would be very useful if the application I was working with was in asm. I have looked all over for help on doing this in c. Maybe someone here can help. Thanks Richard McElroy CST Corporation 450 S. Cips St. Watseka, IL 60970 815-432-6859 ext. 775 -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

What are you trying to do? Float <--> integer conversion is just a matter of casting, and I haven't seen a C compiler recently that didn't support basic operations (+,-,/,*). Standard C math functions will do a lot... are you using a compiler without floating point support? Dale -- "Curiosity is the very basis of education and if you tell me that curiosity killed the cat, I say only the cat died nobly." - Arnold Edinborough On Fri, 22 Feb 2002, Rich Mcelroy wrote: > Does anyone have math libraries for using floating point variables in c. I > am working on a PIC and I have seen microchip's AN575. That would be very > useful if the application I was working with was in asm. I have looked all > over for help on doing this in c. Maybe someone here can help. Thanks -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Can you post more of the code... like the variable declarations for Percent, A & B, etc. I've never used Bytecraft's compiler, but something like that shouldn't be a challenge... Dale -- "Curiosity is the very basis of education and if you tell me that curiosity killed the cat, I say only the cat died nobly." - Arnold Edinborough On Fri, 22 Feb 2002, Rich Mcelroy wrote: > I'm using Bytecraft's compiler for MPLAB. it does basic math, but the > results I'm getting indicate that it is droping anything after the integer > on my calculations. I'm trying to calculate a percentage: > > Percent = A/B * 100; something like that. I have been getting a zero for > percent everytime. -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

----- Original Message ----- From: "Rich Mcelroy" <spam_OUTrmcelroyTakeThisOuTCSTSURVEY.COM> To: <.....PICLISTKILLspam@spam@MITVMA.MIT.EDU> Sent: Friday, February 22, 2002 9:48 AM Subject: Re: [PIC]:MPLAB floating point in C > I'm using Bytecraft's compiler for MPLAB. it does basic math, but the > results I'm getting indicate that it is droping anything after the integer > on my calculations. I'm trying to calculate a percentage: > > Percent = A/B * 100; something like that. I have been getting a zero for > percent everytime. Try multiplying A by 100 first and then divide by B. eg. x = (A * 100) / B IOW, eliminate the fractional parts by scaling them to integers before doing the divide. michael brown -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

On 22-Feb-02 Rich Mcelroy wrote: > I'm using Bytecraft's compiler for MPLAB. it does basic math, but the > results I'm getting indicate that it is droping anything after the integer > on my calculations. I'm trying to calculate a percentage: > > Percent = A/B * 100; something like that. I have been getting a zero for > percent everytime. What are the types of A & B please ? If they are integers, then the result of the division will be an integer. I would try (100.0 * A) / B That should "promote" the result to be a floating point type. See Page 44 of the Second edition of K&R. Peter. -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

A is a signed long, and B is an unsigned long. two large values. -----Original Message----- From: Peter Onion [.....ponionKILLspam.....SRD.BT.CO.UK] Sent: Friday, February 22, 2002 10:04 AM To: EraseMEPICLISTspam_OUTTakeThisOuTMITVMA.MIT.EDU Subject: Re: [PIC]:MPLAB floating point in C On 22-Feb-02 Rich Mcelroy wrote: > I'm using Bytecraft's compiler for MPLAB. it does basic math, but the > results I'm getting indicate that it is droping anything after the integer > on my calculations. I'm trying to calculate a percentage: > > Percent = A/B * 100; something like that. I have been getting a zero for > percent everytime. What are the types of A & B please ? If they are integers, then the result of the division will be an integer. I would try (100.0 * A) / B That should "promote" the result to be a floating point type. See Page 44 of the Second edition of K&R. Peter. -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

On 22-Feb-02 Rich Mcelroy wrote: > Multiplying A by 100 first would work if it wasn't already a large number. > I have declared it as an unsigned long and it is usually around 4500-6000, > if I multiply by 100 I overflow it. Hmmmm I guess the long must only be 16 bits ? Not quite what K&R intended. Ok, so multiply by 1.0 as that is a float and will cause the result to be a float. ((1.0 * A) / B) * 100.0 That should do the trick I think. Peter -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

On Fri, 22 Feb 2002, Rich Mcelroy wrote: > A is a signed long, and B is an unsigned long. two large values. That's probably the problem. Dunno about Bytecraft, but the compiler probably isn't quite intiutive enough to know that you want to perform an operation with two longs and get a float result. It's probably going to do a long mult, then try to map the result into a float variable... the results aren't going to be pretty, I've found. Try this: float Percent; signed long A; unsigned long B; Percent = (float)A/(float)B; Percent *= 100; You may want to take a look at the resulting code size. I find my compiler often generates much smaller, cleaner code with something like this: float Percent, fA, fB; signed long A; unsigned long B; fA = (float)A; fB = (float)B; Percent = fA/fB; Percent *= 100; Dale -- "Curiosity is the very basis of education and if you tell me that curiosity killed the cat, I say only the cat died nobly." - Arnold Edinborough -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Huh? No float support at all? Wow. CCS, Hi-Tech, even CC5X do. Dale -- "Curiosity is the very basis of education and if you tell me that curiosity killed the cat, I say only the cat died nobly." - Arnold Edinborough On Fri, 22 Feb 2002, Rich Mcelroy wrote: > Dale, > > that would be great, and I actually tried to do that first, but Bytecraft > doesn't recognize "float" as a type def. Do you know of a compiler for > MPLAB that does? -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

That's why then Do somehing like Percent = (double) A / (double) B Cast K&R page 45 {Quote hidden}> -----Original Message----- > From: pic microcontroller discussion list > [PICLISTspam_OUTmitvma.mit.edu] On Behalf Of Rich Mcelroy > Sent: 22 February 2002 16:30 > To: @spam@PICLISTKILLspammitvma.mit.edu > Subject: Re: [PIC]:MPLAB floating point in C > > > A is a signed long, and B is an unsigned long. two large values. > > -----Original Message----- > From: Peter Onion [KILLspamponionKILLspamSRD.BT.CO.UK] > Sent: Friday, February 22, 2002 10:04 AM > To: RemoveMEPICLISTTakeThisOuTMITVMA.MIT.EDU > Subject: Re: [PIC]:MPLAB floating point in C > > > On 22-Feb-02 Rich Mcelroy wrote: > > I'm using Bytecraft's compiler for MPLAB. it does basic > math, but the > > results I'm getting indicate that it is droping anything after the > > integer on my calculations. I'm trying to calculate a percentage: > > > > Percent = A/B * 100; something like that. I have been > getting a zero > > for percent everytime. > > What are the types of A & B please ? > > If they are integers, then the result of the division will be > an integer. > > I would try > > (100.0 * A) / B > > That should "promote" the result to be a floating point type. > See Page 44 of the Second edition of K&R. > > Peter. > > -- > http://www.piclist.com hint: The list server can filter out > subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics ***************************************************************** This email has been checked by the altohiway e-Sweeper Service ***************************************************************** -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Bytecraft doesn't recognize the type defs - double or float, but thank you for the advice -----Original Message----- From: Kevin Blain [spamBeGonekevinbspamBeGoneWOODANDDOUGLAS.CO.UK] Sent: Friday, February 22, 2002 10:13 AM To: TakeThisOuTPICLISTEraseMEspam_OUTMITVMA.MIT.EDU Subject: Re: [PIC]:MPLAB floating point in C That's why then Do somehing like Percent = (double) A / (double) B Cast K&R page 45 {Quote hidden}> -----Original Message----- > From: pic microcontroller discussion list > [RemoveMEPICLISTTakeThisOuTmitvma.mit.edu] On Behalf Of Rich Mcelroy > Sent: 22 February 2002 16:30 > To: PICLISTEraseME.....mitvma.mit.edu > Subject: Re: [PIC]:MPLAB floating point in C > > > A is a signed long, and B is an unsigned long. two large values. > > -----Original Message----- > From: Peter Onion [EraseMEponionSRD.BT.CO.UK] > Sent: Friday, February 22, 2002 10:04 AM > To: RemoveMEPICLISTEraseMEEraseMEMITVMA.MIT.EDU > Subject: Re: [PIC]:MPLAB floating point in C > > > On 22-Feb-02 Rich Mcelroy wrote: > > I'm using Bytecraft's compiler for MPLAB. it does basic > math, but the > > results I'm getting indicate that it is droping anything after the > > integer on my calculations. I'm trying to calculate a percentage: > > > > Percent = A/B * 100; something like that. I have been > getting a zero > > for percent everytime. > > What are the types of A & B please ? > > If they are integers, then the result of the division will be > an integer. > > I would try > > (100.0 * A) / B > > That should "promote" the result to be a floating point type. > See Page 44 of the Second edition of K&R. > > Peter. > > -- > http://www.piclist.com hint: The list server can filter out > subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics ***************************************************************** This email has been checked by the altohiway e-Sweeper Service ***************************************************************** -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Are A and B integers? If so the reason you are getting zero is because the compiler is doing integer math. You have to use "casting". Try doing it this way instead: ((float)A)/B * 100.0. While this doesn't look like it is any different from what you are doing it most certainly is. The (float) tells the compiler to treat A as a float (even though it is an integer) and the compiler converts the A to a float. Then it sees the division, since A is a float the division must be done with two floats, so the compiter converts B to a float as well, and then does the division, with the result being a float. The 100.0 isn't really necessary but it does give a hint to what you are doing if someone else is reading your code. Please note that I don't have that compiler, I'm simply speaking about standard ANSI C, I'm sure that compiler has these features. TTYL > {Original Message removed}

> Does anyone have math libraries for using floating point variables in c. I > am working on a PIC and I have seen microchip's AN575. That would be very > useful if the application I was working with was in asm. I have looked all > over for help on doing this in c. Maybe someone here can help. Thanks You can mix assembler and C, the processor won't know. At worst you might have to write a small assembler wrapper for the canned routines to implement whatever the C calling conventions are for your compiler. You could also ask your compiler vendor about floating point math libraries. ******************************************************************** Olin Lathrop, embedded systems consultant in Littleton Massachusetts (978) 742-9014, RemoveMEolinspam_OUTKILLspamembedinc.com, http://www.embedinc.com -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

> That's probably the problem. Dunno about Bytecraft, but the compiler > probably isn't quite intiutive enough to know that you want to perform an > operation with two longs and get a float result. It's probably going to > do a long mult, then try to map the result into a float variable... the > results aren't going to be pretty, I've found. But that is exactly what the compiler is required to do, if it implements C. If you want a fractional result, cast one of the operands to float (or multiply by 1.0) ((float)A)/B or 1.0*A/B etc. Wouter van Ooijen -- Van Ooijen Technische Informatica: http://www.voti.nl Jal compiler, Wisp programmer, WLoader bootloader, PICs kopen -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

If you are not already using floating point in your program, using it now is likely to increase your code size a lot. If you have it, maybe that is okay. But this is why C gets a "bad rap" over assembly. It is usually important to consider the result of your C statements before accepting a quick solution. I suggest "rolling your own" function in C to handle the overflow/underflow that results from integer math. David RemoveMEdavidTakeThisOuTspamKosmosIsland.com On Fri, 22 Feb 2002 10:29:35 -0600 Rich Mcelroy <EraseMErmcelroyspamspamBeGoneCSTSURVEY.COM> wrote: > A is a signed long, and B is an unsigned long. two large values. > > {Original Message removed}

> Bytecraft doesn't recognize the type defs - double or float, but thank you > for the advice I don't like compilers on this tiny resource-limited systems, but I don't know what amazes me more: That somebody tries to pass off crap like that as a commercial product or that somebody else actually bought it. There has got to be a "real" compiler out there for PICs. ******************************************************************** Olin Lathrop, embedded systems consultant in Littleton Massachusetts (978) 742-9014, RemoveMEolinKILLspamembedinc.com, http://www.embedinc.com -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

On Fri, 22 Feb 2002, Olin Lathrop wrote: > > Bytecraft doesn't recognize the type defs - double or float, but thank you > > for the advice > > I don't like compilers on this tiny resource-limited systems, but I don't > know what amazes me more: That somebody tries to pass off crap like that as > a commercial product or that somebody else actually bought it. There has > got to be a "real" compiler out there for PICs. Geez, not again. The compiler I use, while not 100.000 (or 99.999...) percent ANSI compliant, is a "real" compiler, as evidenced by the wide variety of "real" code it has produced over the past year and a half for "real" products using "real" PICs. I'm sure there are even better ones available. So far I have not found a single instance of something that can't be done, and done well, with the compiler I use -- the one that shortens my development time and allows me to complete projects in far less time than I would without it. Don't just assume that because someone else uses different tools than your particular favorites that it just won't work or will somehow produce vastly inferior results. Some people don't like assemblers, some don't like C, some don't like BASIC. In each case, obviously, someone has found a set of circumstances for which the language or the means of translating it to machine code is a good fit. Dale -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.