◄Up► ◄Contents► ◄Index► ◄Back►
─────C/C++ Compiler─────────────────────────────────────────────────────────
 
     Syntax:  /AH
 
     The huge-model option is similar to the large-model option. But
     with the huge model, the restrictions on the size of individual
     data items for arrays are relaxed.
 
     Array               Data Item
     (Size in Bytes)     (Size in Bytes)
 
     ≤64K                ≤64K
     >64K; ≤128K         ≤64K
     >128K               ≤64K (and a power of 2)
 
     Some size restrictions apply to elements of a huge array if the
     array is larger than 64K. So that efficient addressing occurs,
     array elements are not permitted to cross segment boundaries.
     Therefore, no array element can be larger than 64K.
 
     For any array larger than 128K, all elements must have a size in
     bytes equal to a power of 2. However, if the array is 128K or
     smaller, its elements can be any size--up to and including 64K.
 
     In huge-model programs, be careful when you use the sizeof
     operator or when you subtract pointers. The C language defines the
     value returned by the sizeof operator to be an unsigned int value.
     But the size in bytes of a huge array can be an unsigned long
     value. Similarly, ptrdiff_t, the type returned by subtracting two
     pointers, is defined as an unsigned int. With huge arrays, the
     difference can exceed the capacity of ptrdiff_t.
 
     The defaults in huge-model programs are far code addressing and
     far data addressing. You can modify these defaults with the __near
     keyword for code and the __near or __far keyword for data.
 
     To maximize efficiency, the compiler tries to allocate static and
     initialized global data in DGROUP. The compiler allocates data
     items larger in size than the data-size threshold (set with /Gt) in
     segments outside of DGROUP. The sum of all the items smaller than
     the threshold must be less than 64K.
     See: ◄Data Size Threshold (/Gt)►
 
     Under the compact, large, or huge memory model, the compiler
     allocates initialized data items as near if they are smaller than
     or equal in size to the threshold value set by the /Gt option. The
     /Gx option extends this initialized data allocation rule to data
     that is uninitialized and data that is marked as extern. Near
     data offers two benefits:
 
        ■ The compiler can generate more efficient code to reference
          data it knows is near.
 
        ■ You can achieve multiple instances of a single windows
          application when all data is near.
 
     See: ◄Overriding Default Addressing►
          ◄Assume That Data Is Near Option (/Gx)►
 
     Using the /AH option causes CL to emit both the _M_I86 and the
     _M_I86HM preprocessor identifiers.
                                    -♦-