[Last Call] Learn how to a build a cloud-first strategyRegister Now

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 530
  • Last Modified:

How to manipulate floating point number in ARM 7 ?

How do I manipulate floating point numbers in ARM 7 based target ?
Our embedded application is based on ARM 7 Microcontroller.  This microcontroller does not have floating point unit.  
We are using Keil's compiler, assembler, linker, and Real-Time Operating System.

I'm getting unsigned32 minutes over modbus.  I need to convert them to hours and send over USB to PC application.

i.e.
Given:  5351 minutes
My embedded application needs to:  5351 / 60 = 89.18333...
Then, I would send 891 or 892 to PC application.  They would divide by 10 to display 891 or 892 hours.

What are my options ?

1)  Does Keil provide any floating point library ?
2)  Floating point math is usually eliminated in embedded systems.  Instead, fixed-point math is used.  Does Keil provide fixed point math library?
3)  Use shift operations to achieve my goal?

What would be the best way to convert minutes to hours in my embedded application?
0
naseeam
Asked:
naseeam
  • 2
1 Solution
 
jkrCommented:
Just use your compiler's built-in support: http://www.keil.com/support/docs/1879.htm

QUESTION

We are changing software that currently uses integer math to use floating-point math. Does the Keil C166 Compiler support floating-point math and are there include files or libraries we need to declare in order for it to compile successfully with float variables?
ANSWER

The C166 compiler supports both single-precision (32-bit) and double-precision (64-bit) floating-point. To use either, all you have to do is start declaring and using floating-point variables just like in ANSI C.

By default, all float and double variables are stored in 32-bit IEEE format. This gives a precision of 7 decimal digits. When you require higher precision (than 7 digits) you may use the C166 FLOAT64 directive to enable 64-bit doubles. Enable this option in ┬ÁVision under Project - Options for Target - C166 - Double-precision floating-point. If you enable FLOAT64, all double variables are stored in 64-bit IEEE format. This provides a precision of 15 decimal digits.

IOW, even though your CPU does not have a FPU, the compiler supplied runtime will take care of that.

See also http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0471j/pge1358786964007.html ("1.8 ARM and Thumb floating-point build options (ARMv7 and later)")
0
 
naseeamAuthor Commented:
Thank you very much for great answer.   If I use Keil's floating point library, I'll get floating point hours.  I need integer hours because I need to send them to PC application as integer.
0
 
jkrCommented:
Well, then cast the result back to integer, e.g.

int nMin = 5351;
float fMin = (float) nMin;

float fHrs = fMin / 60.0;

int nHrs = (int) fHrs * 10.0;

Open in new window


But: If that is all your calculation requirement, you can elominate the whole floating point thing with a simple trick: First multiply the minutes by 10 and then divide by 60, the results will be the same, e.g.

int nMin = 5351;

int nHrs = (nMin * 10) / 60; // or: int nHrs = nMin / 6;

Open in new window

0

Featured Post

Independent Software Vendors: We Want Your Opinion

We value your feedback.

Take our survey and automatically be enter to win anyone of the following:
Yeti Cooler, Amazon eGift Card, and Movie eGift Card!

  • 2
Tackle projects and never again get stuck behind a technical roadblock.
Join Now