Take a look in the measurement/ and examples/ directories. They have some useful tools.
} I was wondering how one can test on the spot if the rtlinux just installed } is indeed working properly, and what kind of benchmarks one can give. } Also, what kind of relation is there between the scheduling precision, the } hardware setup, and the linux configuration. } } I had a hard time to find proper tests (any concise info, collection of } performance-tests, etc, would be still welcome!). I gathered from the } mailing lists and various places the following two methods: } } 1. } cd /usr/rtlinux/ } ./regression.sh } less (or cat/more, whatever) regression.log } ... } } For me, this test was almost useless, 99% of the logfile is such as: } } 1 Warning: loading regression/fifo_module.o will taint the kernel: no license } 2 Warning: loading regression/fifo_module.o will taint the kernel: no license } 4 Warning: loading regression/fifo_module.o will taint the kernel: no license } 8 Warning: loading regression/fifo_module.o will taint the kernel: no license } 16 Warning: loading regression/fifo_module.o will taint the kernel: no license } 32 Warning: loading regression/fifo_module.o will taint the kernel: no license } 64 Warning: loading regression/fifo_module.o will taint the kernel: no license } 128 Warning: loading regression/fifo_module.o will taint the kernel: no license } } However, the file also tells the following: } } } system (/sbin/rmmod thread_time): Success } Warning: loading regression/thread_time.o will taint the kernel: no license } poll (&fifo_poll, 1, 1600): Success } } OR sometimes (if i am lucky?): } } system (/sbin/rmmod thread_time): Success } Warning: loading regression/thread_time.o will taint the kernel: no license } avg_drift: 3.06 us worst_drift: 15.04 us } } } *** Q: Could anyone shed some light on what exactly these mean, and why is } that I don't see most of the time the "avg_drift" line? } } Average drift of the start of periodically emitted signals? Monotonic } drift? Or the avg value of the (absolute-value of the) deviation } between the real signal and theoretical time of emission? } } 2. The other method: } } cd /usr/rtlinux/examples/measurements } insmod rt_process.o } ./monitor > /tmp/rtlinux.stat } ...let it run for a few days... } <CNTRL-C> } } Look at /tmp/rtlinux.stat and get some statistics. } } *** Q: Again, a bit of an explanation on what is in the file would be } appreciated. } } The min column is always 0,sometimes 32, no other value. The max } value is between 0 and "infinity". The median is usually around } 5000, but values as big as 10^6 occur. } } *** Q: Any other good method for testing the performance of a system? } Scripts would be welcome. } } Cheers, } Gaspar } } + ------------------------------------------------------------------------ + } Gaspar A. Bakos } Predoctoral Fellow, Solar and Stellar Physics Division } Harvard-Smithsonian Center for Astrophysics } 60 Garden Street, Cambridge, MA 02138 (USA) } Phone(office): 617-495-7410 Fax: 617-495-7049 } email: [EMAIL PROTECTED] or [EMAIL PROTECTED] } homepage: http://cfa-www.harvard.edu/~gbakos } + ------------------------------------------------------------------------ + } } _______________________________________________ } Rtl mailing list } [EMAIL PROTECTED] } http://www2.fsmlabs.com/mailman/listinfo.cgi/rtl _______________________________________________ Rtl mailing list [EMAIL PROTECTED] http://www2.fsmlabs.com/mailman/listinfo.cgi/rtl
