CCS/TMS320F28069: Using Math Tables in CLA and FPU

Andrew,

Good catch on that special note in the UG about 2806x devices. Looks like you will have to go with the Flash->RAM approach for this device.

Anyway, the short answer to your question is this - the tables are available in the CLA1_math_library project (in the various files named xxxTable.asm), so when the CLA1_math_library_fpu32.lib is included in your project, the Linker will find the tables in the library and store them in the Flash location specified in the Linked command file, and at run time, they will be loaded to the RAM location specified in the Linker command file. As a user, you don't need to do anything more. My guess is if you setup everything correctly (including library, choosing the right build configuration etc.), your project should be working properly and using the CLA tables and generating the right results already.

There is a lot that can be learnt from the examples in C:\ti\c2000\C2000Ware_1_00_06_00\libraries\math\CLAmath\c28\examples. For example, I loaded the cla_asin project. This has various build configurations for various devices, and within a device, for FLASH, RAM etc. I chose the f2806x_FLASH build configuration. Then if you expand f2806x in the project view, you'll see the linker command file f2806x_cla_c_lnk.cmd. Here, you'll see that if the FLASH build configuration is being used, CLA1mathTables must be loaded into FLASHB (origin 0x3F0000) and run from CLARAM1 (origin 0x008C00).

You'll also see from Project Properties -> C2000 Linker -> Advanced Options -> Command File Preprocessing -> --define that CLA_MATH_TABLES_IN_ROM is not defined, which means the tables are not in ROM for this device (or that option is not being used).

Now if you build the project, and open cla_asin.map (located when you expand f2806x_FLASH in the project view), you'll see the following:
CLA1mathTables
* 1 003f0000 0000030e RUN ADDR = 00008c00
003f0000 0000018a cla0_math_library_fpu32.lib : CLAasineTable.obj (CLA1mathTables)

The above indicates the tables are loaded in FLASHB (Origin 0x3f0000).

and also further down the .map file:

1 00008c00 _CLAasinHalfPITable
1 00008c04 _CLAasinTable

The above indicates the tables are copied to CLARAM1 (Origin 0x008C00).

Just to experiment further, I made the following change in cla_asin.cla:

__attribute__((interrupt)) void Cla1Task1(void)
{
__mdebugstop();
fResult1 = CLAasin(fVal);
fResult1 = CLAacos(fVal);
}

__attribute__((interrupt)) void Cla1Task2(void)
{
//
// Test inline function
//
__mdebugstop();
fResult2 = CLAasin_inline(fVal);
fResult2 = CLAacos_inline(fVal);
}

Now if you build the cla_asin project and inspect cla_asin.map, you'll see the following:

CLA1mathTables
* 1 003f0000 0000030e RUN ADDR = 00008c00
003f0000 0000018a cla0_math_library_fpu32.lib : CLAasineTable.obj (CLA1mathTables)
003f018a 00000184 : CLAacosineTable.obj (CLA1mathTables)

and also further down the .map file:
1 00008c00 _CLA1mathTablesRunStart
1 00008c00 _CLAasinHalfPITable
1 00008c04 _CLAasinTable
1 00008d8a _CLAacosinHalfPITable
1 00008d8a _CLAasinTableEnd
1 00008d8e _CLAacosinTable
1 00008f0e _CLAacosinTableEnd

I hope this clears up things for you.

Thanks,
Sira