The sky_model simulator will run on any modern computer.
All programs are written in ANSI C and one can build it on any
operating system with a standards-compliant C compiler. Apart E5-2640 2.5 GHz server running CentOS 6.3 (equivalent to Red Hat
6.3). Time values shown in the figure is an average over three runs.
On the left panel of Fig. 3 is the total time elapsed from the start to
the end of the program. This includes the amount of time spent
by the processor to execute instructions, data read/write to the
disk, I/O wait, etc. The CPU time (yellow curves) is the amount
of time spent in user-mode code (outside the kernel). This is the
actual CPU time used in executing the process. Other processes
and the time spent in waiting are not included in this figure. A
comparison of the CPU and wall-clock times makes it clear that the
program is not spending much time waiting for external operations
to complete. The system time, plotted on the right panel, is the
amount of CPU time spent in the kernel within the process (system
calls, such as memory allocation, file read/write execution, etc.). It
is also clear from the right panel that time spent on the system
(kernel) processes is only a small fraction of the total time. Given
the computationally intensive nature of the code, these numbers
indicate that our code is optimized with minimal I/O bottlenecks.