Date: Sun, 23 Apr 2006 19:27:35 -0700 From: Don Dugger <dugger@hotlz.com> To: freebsd-chat@freebsd.org Cc: Benjamin Lutz <benlutz@datacomm.ch>, Dan Strick <strick@covad.net> Subject: Re: Why is not more FreeBSD software written in C++? Message-ID: <444C3797.7010909@hotlz.com> In-Reply-To: <200604232232.32653.benlutz@datacomm.ch> References: <200604231525.k3NFP64X003155@mist.nodomain> <200604232232.32653.benlutz@datacomm.ch>
next in thread | previous in thread | raw e-mail | index | archive | help
For fun I run the tests on a SUN box I have here it's an old Sparc 200MHz Running Solaris 8. and I used the gcc 3.4.2 compiler. ---------- time ./test_putchar >/dev/null real 0m0.45s user 0m0.44s sys 0m0.01s time ./test_write >/dev/null real 0m0.66s user 0m0.38s sys 0m0.28s time ./test_iostream >/dev/null real 0m3.28s user 0m3.12s sys 0m0.04s time ./test_string >/dev/null real 0m1.50s user 0m1.48s sys 0m0.02s ---------- Don 8) Benjamin Lutz wrote: > Hello Dan, > > On Sunday 23 April 2006 17:25, Dan Strick wrote: > >> My results show something quite different. Even though the buffer >> size used by Benjamin's write() test (61 bytes) is much larger than >> the one I used (only 24 bytes), the write() test to /dev/null still >> takes about 4 times longer than the putchar() test. As one might >> expect, the write() test spends most of its time in the kernel doing >> all those write() system calls. The putchar() test (i.e. stdio) >> is probably using a 1K byte buffer. >> > > Your results are intriguing. I ran the tests again on a FreeBSD-6 machine (A > Sempron 3000+), and I see the same thing, test_putchar runs faster and with > much lower cpu load than test_write. > > I'm not sure what to make of this. I would have said that the reason that > write() is faster is because there's much fewer kernel calls necessary. I'm > beginning to think though that FreeBSD's write(2) implementation is not as > good as it could be. My Linux Mac Mini using a notebook HD actually runs > test_write 10% faster than the FreeBSD Sempron 3000+ using proper SATA disks. > > Or maybe Linux just cheats by doing some smart delaying/caching. > > Thinking about it some more, I think you're right with your assumption that > test_putchar is faster because stdio does some internal buffering. I ran > ktrace on test_putchar (the version with 500'000 lines), and it calls write() > 4562 times on said FreeBSD system, each time with 0x1000 as last argument, ie > it uses 4KB buffers. To confirm that buffering to be the reason, I increased > the size of each line from 61 bytes to 16001 bytes, at which point the > write() version showed a clear performance advantage on FreeBSD as well. > > I had not thought about this possibility when I made my original statement. > Given that, it makes sense that putchar is quite fast. > > It still puzzles me though why the Mini (it's only a 1.25GHz G4!) beats the > Sempron though. > > >> I stand by my suggestion that using putchar() might not be slower. >> On my system it was much faster. >> > > Ok, I accept that then. If we think of a case where buffering must not happen, > and thus a putchar() must be followed by a fflush(), would you agree that > just using write() is faster? > > Cheers > Benjamin >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?444C3797.7010909>