|
The Question is:
Here's a performance tuning question:
Quantum is the amount of time a process is allowed to run on the CPU. Is there
a way to measure the number of times quantum is reached? It seems that if you
have very CPU intensive processes running that reach quantum regularly, you
could increase the v
alue of quantum, thus eliminating excessive overhead caused by loading and
unloading the context of each process as it enters and leaves (then possibly
re-enters) the CPU.
Are there any popular theories on valuing quantum or methods to determine
whether a system has improved/declined in performance when
increasing/decreasing quantum?
The Answer is :
OpenVMS process scheduling deals (automatically) with quantum end
processing, and the scheduler will cause the current process to
continue (without swapping the process context) when it is the best
candidate.
|
