Specifying Memory Models to use with Systems Based on Intel® 64 Architecture

Additional Notes on Memory Models and on Large Data Objects

• When you specify the medium or large memory models, you must also specify the -shared-intel compiler option to ensure that the correct dynamic versions of the Intel run-time libraries are used.

• When you build shared objects (.so), Position-Independent Code (PIC) is specified (that is, -fpic is added by the compiler driver) so that a single .so can support all three memory models. However, code that is to be placed in a static library, or linked statically, must be built with the proper memory model specified. Note that there is a performance impact to specifying the medium or large memory models.

• The use of the memory model (medium, large) option and the -shared-intel option is required as a by-product of the code models stipulated in the 64-bit Application Binary Interface (ABI), which is written specifically for processors with the 64-bit memory extensions. Both the compiler and the GNU linker (ld) are responsible for generating the proper code and necessary relocations on this platform according to the chosen memory model.

• The 2GB restriction on Intel® 64 architecture involves not only arrays greater than 2GB, but also COMMON blocks and local data with a total size greater than 2GB. The Compiler Options reference contains additional discussion of the supported memory models and offers details about the 2GB restrictions for each model. See the -mcmodel options page.

• If, during linking, you fail to use the appropriate memory model and dynamic library options, an error message in this format occurs:

<some lib.a library>(some .o): In Function <function>:
  : relocation truncated to fit: R_X86_64_PC32 <some symbol>