{ MA>generate an interrupt (int $70?) 1000 times a second. To speed up the system timer by a factor of 2, call "goosetimer(2)": } procedure goosetimer(goose: byte); { Speed up system timer } var gooseword: word; begin gooseword := $ffff div goose; { Number of oscillations between ticks } port[$43] := $36; { Set timer at new speed } port[$40] := lo(gooseword); port[$40] := hi(gooseword); end; { In that procedure, you are telling the timer chip how many chip oscillations to wait before generating a "tick". The default is to generate a "tick" every 65536 cycles, or 18.2 times per second (which also works out to 65536 "ticks" per hour). The drawback here is that the system clock will speed up accordingly, as will all functions dependent on the reception of "ticks". So a useful approach is to reprogram Int $08 such that additional "ticks" are not passed along to the standard system functions. The way to accomplish that is to generate an "End of Interrupt" command on all the additional "ticks" instead of invoking the "original" Int 08h ISR. The "End of Interrupt" command is a command that, much like the STI instruction, enables subsequent hardware interrupts to be processed. Code is in order: } var goosefactor, ticklooper, cloktick: byte; oldtimint: procedure; { GOOSEFACTOR: the multiplication factor for the system speed TICKLOOPER: loops from 0 to GOOSEFACTOR - 1, resetting itself to 0 when it gets to GOOSEFACTOR -- used internally to determine which "ticks" get passed along CLOKTICK: counts "ticks"; used in standardizing program timing OLDTIMINT: points to "original" Int 08h ISR } {--------------------------------------------------------------------------} procedure tickwait(time2wait: byte); { delay until counter reaches } begin { certain value } repeat until cloktick >= time2wait; cloktick := 0; { reset counter } end; {--------------------------------------------------------------------------} procedure newtimint; interrupt; { new timer interrupt } begin if ticklooper > 0 then { "suppress" this "tick" } port[$20] := $20 { "End-of-Interrupt" command } else begin asm pushf; end; { call old timer interrupt } oldtimint; end; inc(cloktick); { update "tick" counter } inc(ticklooper); if ticklooper = goosefactor then ticklooper := 0; end; {--------------------------------------------------------------------------} procedure initnewtimint(goose: byte); { set up new timer interrupt } var gooseword: word; begin goosetimer(goose); { speed up timer } goosefactor := goose; { record speed increase } getintvec($08, @oldtimint); { record location of old interrupt } setintvec($08, @newtimint); { install new interrupt procedure } cloktick := 0; { set counter to 0 } ticklooper := 0; { set "extra tick" determiner to 0 } end; {--------------------------------------------------------------------------} procedure setoldtimint; { reset old timer } begin setintvec($08, @oldtimint); { original interrupt } goosetimer(1); { original system speed } end; {--------------------------------------------------------------------------} { To start new system timing, it's "initnewtimint"; to turn it off, it's "setoldtimint". Most of the rest of that code is "internal": you don't need to worry about it. The one procedure I haven't explained is "tickwait": it is used to standardize timing in programs from machine to machine by waiting for a given number of "ticks" to have gone by before continuing. (Unlike "Delay", "tickwait" monitors an interrupt-driven counter, meaning the number of "ticks" is advanced even while other routines are executing.) For example, this loop will print a new line every 18 "ticks", which are coming at twice the normal speed: } initnewtimint(2); cloktick := 0; while not keypressed do begin writeln('18 more ticks have gone by'); tickwait(18); end; setoldtimint;