12.14 Memory Models

The compiler supports several memory models. Command-line options are available for selecting the optimum memory model for your application, based on the specific device that you are using and the type of memory usage.

Table 12-1. Memory Model Command-Line Options
Option	Memory Definition	Description
`-msmall-data`	Up to 6 KB of data memory¹. The default is device dependent².	Permits use of PIC18 like instructions for accessing data memory.
`-msmall-scalar`	Up to 6 KB of data memory¹. This is the default.	Permits use of PIC18 like instructions for accessing scalars in data memory.
`-mlarge-data`	Greater than 6 KB of data  memory¹. The default is device dependent².	Uses indirection for data references.
`-msmall-code`	Up to 32 kWords of program memory. This is the default.	Function pointers will not go through a jump table. Function calls use `RCALL` instruction.
`-mlarge-code`	Greater than 32 kWords of  program memory.	Function pointers might go through a jump table. Function calls use `CALL` instruction.
`-mconst-in-data`	Constants located in data  memory.	Values copied from program memory by startup code.
`-mconst-in-code`	Constants located in program memory. This is the default.	Values are accessed via Program Space Visibility (PSV) data window.
`-mconst-in-auxflash`	Constants in auxiliary Flash.	Values are accessed via Program Space visibility window.
Note: For most devices 6K of RAM is the near data space, but for some devices it is 4K of RAM. For devices that have all of their data memory in the near space, the memory model is “small data” “small scalar” so that all memory will be allocated in the near space. For all other devices the default memory model is “large data” “small scalar”. This will have the effect of allowing the tool chain to place aggregate objects, such as arrays and structure, into the far memory space. This can be over-ridden by explicitly selecting “small data” in the compiler options.

The command-line options apply globally to the modules being compiled. Individual variables and functions can be declared as near, far or in eds to better control the code generation. For information on setting individual variable or function attributes, see Variable Attributes and Function Specifiers.