ANONYMOUS wrote:
> As this example should be idle from 247(including)... 566(including) which is exactly 320usecs and then at 567 it should declare it has finished. However the sample solution includes from (246...565) as idle then uses 566 to declare it has finished as shown in the second picture
I'm pretty sure that this has been discussed previously, but I can't find it.
The 'belief' is that '246 idle' should not be reported, but it happens because the final reporting/debugging code at the end of each microsecond just prints 'idle' if the CPU is now not occupied. Thus, it reports that multiple things are happening at 246. The 'idle period' genuinely commences at 247.
> Additionally is there any leniency for marking when your time is off by 1 second?
Certainly; given that we're seeing multiple interpretations of when the clock should be incremented and that reported, there's different solutions that meet the 'overall requirement' of executing all processes and I/O in the correct order. The final time in the measurements line is more a reflection of whether code is queuing and selecting processes in the defined order, and will tend to be off by a lot (a lot more than 1usec) if the ordering is wrong.
In effect, your final time should be within a range of the sample solution's time
T.
(don't quote me, but) if a whole sequence of commands involves a total of
S system-calls, and each is coded to take an extra 1usec (worst case??), then we'll be looking for a total time of
T +/- S usecs. The automatic marking script looks for final times within such a range.
