

   ͻ
                                                                            
           HOW THE INTEL 387(TM) DX AND THE WEITEK 1167 ARE DIFFERENT       
                                                                            
   ͼ




       The two major difference between the Intel 387DX and the Weitek 1167
       are the number of programs that support each math coprocessor and the
       way each of them performs.  The following sections highlight these
       differences.


   SOFTWARE SUPPORT
   

       At this writing, over 300 programs support the 387DX and other Intel
       Math Coprocessors.  Most compilers and assemblers (tools programmers
       use to create new applications) for the 386DX microprocessor support
       387DX.  The 387DX functions the same whether the 386DX is in real mode
       (which DOS uses), protected mode (used by OS/2 and UNIX), or virtual 86
       mode (for multi-tasking).   This make it a snap to produce more
       powerful, speedy programs which use the Intel Math CoProcessor.  The
       387DX is compatible with programs written for the Intel 8087 and 80287
       math coprocessors too!

       The Weitek 1167 requires the 386DX to run in protected mode.  Since DOS
       programs run real mode, it is impossible to use the 1167 without a
       special program called a DOS extender.  At this time there are very few
       programs under a DOS extender.


   PERFORMANCE
   

       The Intel 387DX supports basic math (addition, subtraction,
       multiplication and division), exponential, logarithmic and
       trigonometric functions, and it can perform these operations much more
       quickly than the 386DX.  Since all math functions can be processed by
       the 387DX, this leaves the 386DX free to perform other system
       operations until the math coprocessor returns its result.  This
       coprocessing is what gives the Intel 387DX its overall advantage.

       The Weitek 1167 performs basic math functions faster than the 387DX,
       but basic math is all this math coprocessor can perform.  It cannot
       process higher math functions like the 387DX.  This means the 386DX
       must perform the higher math functions.  Because the 386DX uses
       software subroutines to perform these complex tasks, the calculations
       take longer and system operations must either wait or interrupt the
       calculation in progress.







   

   2/90                        Intel FaxBack # 3004                     Page 1


