Double precision addition in MSP430F5438A with Code Composer studio.

Pugazhendi Muthuswamy

Other Parts Discussed in Thread: MSP430F5438A

Hi,

I am using MSP430F5438A. The double allots 32 bit only at CCS. How to set it as 64 bits. I am reading from a devices, which produces 8 bytes for double. It should be added and write to another device. The struct and uions are given below. I am using CCS 5.1. Looking forward of reply. Thanks.

typedef union regType_t{
    unsigned char cha[8];
    unsigned int i;
    double d;
    float f;
    long l;
    char c;
}regTy;

EXTERN regTy storeSlaveRegValues[2];
EXTERN regTy calculateSlaveRegValues;

calculateSlaveRegValues.d = storeSlaveRegValues[0].d + storeSlaveRegValues[1].d;

over 13 years ago

0 Leo Hendrawan over 13 years ago

TI__Guru 56650 points

Hi,

according to the CCS compiler documentation, in EABI format, the double data type should be implemented as 64 bits.

To change the output format from COFF to EABI, you can do the following:

0 Pugazhendi Muthuswamy over 13 years ago in reply to Leo Hendrawan

Prodigy 50 points

Dear Leo,

Thanks for the reply.

However, if I chage the option to eabi, the following error occurs.

>> ERROR: mklib: could not unpack rtssrc.zip
warning #10207-D: automatic RTS selection: resolving index library "libc.a" to
   "C:\ti\ccsv5\tools\compiler\msp430\lib\rts430x_lc_sd_eabi.lib", but
   "C:\ti\ccsv5\tools\compiler\msp430\lib\rts430x_lc_sd_eabi.lib" was not
   found
warning #10062-D: entry-point symbol "_c_int00" undefined

undefined      first referenced
symbol            in file
---------      ----------------
__TI_zero_init
__mspabi_slli ./evec001.obj

error #10234-D: unresolved symbols remain

warning #10202-D: no suitable entry-point found; setting to 0
>> Compilation failure
error #10010: errors encountered during linking; "ConVer012.out" not built
gmake: *** [ConVer012.out] Error 1
gmake: Target `all' not remade because of errors.

Can you please help.

0 Leo Hendrawan over 13 years ago in reply to Pugazhendi Muthuswamy

TI__Guru 56650 points

Hi,

i am not sure about the error given above, but i think the compiler doesn't manage to find the rts (run-time support) library. Could it be that somehow your CCS installation get corrupted? (Maybe try to reinstall CCS).

0 Pugazhendi Muthuswamy over 13 years ago in reply to Leo Hendrawan

Prodigy 50 points

Dear Leo,

I can able to compile the rtslib. Its Cygwin problem. Eventhough, the long double or double uses 4 bytes. (32 bit). See the attached JPG.

0 Pugazhendi Muthuswamy over 13 years ago in reply to Pugazhendi Muthuswamy

Prodigy 50 points

0 Leo Hendrawan over 13 years ago in reply to Pugazhendi Muthuswamy

TI__Guru 56650 points

Hi,

i think this is only a debugging display issue, the variable is already allocated as 64 bit. I made the following test:

As you can see in the debug memory window above, the var1 variable declared as double has 4*2=8 byte long memory allocation.

0 Pugazhendi Muthuswamy over 13 years ago in reply to Leo Hendrawan

Prodigy 50 points

Thanks Leo.

0 Chris Norris over 12 years ago in reply to Leo Hendrawan

Intellectual 935 points

Good morning,

I have changed my project to use EABI so I can use 8-byte double precision in my program. Running CCS 5.2.1.00018, I am seeing the debug display issue where 8-byte double precision values are not displayed properly.

Is there a fix planned for this from TI?

Thanks,
Chris Norris

**Attention** This is a public forum

MSP low-power microcontrollers

MSP low-power microcontroller forum

Double precision addition in MSP430F5438A with Code Composer studio.