$FG,5$$TX+CX,"MultiTasking"$$FG$ $WW,1$TempleOS is a preemptive multitasking operating system. However, you can turn-off preemption on a task by task basis and the task will not yield ctrl unless it volunteers with $LK,"Yield",A="MN:Yield"$(), $LK,"Sleep",A="MN:Sleep"$() or a call to I/O which yields ctrl. Use the $LK,"Preempt",A="MN:Preempt"$() cmd to turn on or off preemption. This is useful when you have multiple tasks modifying data structures, simultaneously. Interrupts still occur when preemption is off. Turning-off preemption does not protect against multicore simultaneous access. $BK,1$Preemption is turned-off by default when a task is $LK,"Spawn",A="MN:Spawn"$()ed.$BK,0$ The time-slice rate is set by $LK,"JIFFY_FREQ",A="MN:JIFFY_FREQ"$. TempleOS does not remove tasks from $FG,2$the queue$FG$ when they are waiting. They swap-in, check whatever condition they are waiting for and swap-out. This keeps it simple, but would cause problems with too many tasks. TempleOS is intended for non-server, home computers with a maximum of around ten, or so, tasks. See $LK,"Scheduler",A="FL:::/Kernel/Sched.CPP,1"$. TempleOS has no task priority scheme. Disk requests are not broken into pieces for sharing. Therefore, if one task makes a big request, the drive is locked onto that task until it is done. During that time, other tasks can do CPU-only operations, however. $FG,5$$TX+CX,"Interrupts"$$FG$ You can turn on and off interrupts in your user programs. This is part of the mission statement, giving complete freedom to programmers. Use $FG,2$CLI$FG$ or $FG,2$STI$FG$. TempleOS currently doesn't use interrupt besides the timer and keyboard and sometimes mouse interrupt. Since the interrupt flag is part of your task context and since you are voluntarily swapped-out if you call $LK,"GetChar",A="MN:GetChar"$() or whatever, you won't even notice if you turn-off interrupts at the cmd prompt because you are being swapped-out and interrupts are being turned-on by the next task. I saw no reason to make the hard drive or CD-ROM code interrupt driven -- it just creates potential compatibility issues. On a home system, you do one thing at a time, mostly. TempleOS can do multiple things at once, but not as efficiently as a server operating system. It would bogg down with a hundred tasks asking for service. It is much simpler, though, the way it is. The disk code does not break-up large requests in pieces, if you are curious why one task must wait until another finishes an operation. See $LK,"::/Demo/Lectures/Spinning.CPP"$.