Date: Wed, 23 Aug 1995 18:06:55 +1000 From: Bruce Evans <bde@zeta.org.au> To: dyson@freefall.FreeBSD.org, hackers@freebsd.org Subject: Re: Here are my benchmark results, so far Message-ID: <199508230806.SAA10195@godzilla.zeta.org.au>
next in thread | raw e-mail | index | archive | help
>NOTE: The very high random seeks and the sequential output perf are probably >due to the asynchronous writes that Linux uses. Note also that Linux appears >to use much more CPU for I/O operations. (It is probable, however, that the >FreeBSD version of bonnie does not properly measure the overhead of IDE PIO >operations.) If MULTI-BLOCK I/O is enabled, FreeBSD can easily surpass >the write performance of the Linux results above, but >the MULTI-BLOCK I/O mode has not been enabled for this Linux test. Bonnie's estimates of the CPU overheads are bogus for almost all systems. They are especially bogus for FreeBSD because the interrupt overhead is carefully separated from the interrupt overhead so it that it doesn't add to the (average) system time even for interrupts that happen to occur during syscalls. IDE PIO overhead for FreeBSD can be estimated very accurately by running benchmarks on unloaded systems systems and watching the output from `systat -vmstat'. The interrupt overhead gives the PIO overhead and the user+sys overhead gives the other overhead. There is no application interface for reading the interrupt overhead. Busmastering SCSI overhead is harder to measure. It slows down everything but there is no way of seeing the slowdown while the system is idle. Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199508230806.SAA10195>