> NT's asynchronous I/O is very useful when dealing with external hardware
> and something approaching real time. I would be most happy if Linux were
> to support asynchronous I/O and a system call like NT's WaitForMultipleObjects().
> For those who have not done this sort of thing on NT, WaitForMultipleObjects
> is like an enhanced select() which can accept files, semaphores, mutexes, the whole
> range of things that a program might want to wait for and wakes the program
> up as soon as one of these is ready. It allows me to write simple code which
> deals efficiently with a long lists of asynchronous events.
\
Everyone talks about how great NT does with Async I/O, but to tell you
the truth, has anyone ever used NT with a Cd-Rom Changer? Has anyone
actually used NT with a CD-Rom? I don't know what is wrong with them,
but watch the entire UI, and all the processes HANG as it reads from a
*(&*& CD. With my pioneer 6 disk changer, the entire system *WAITS*
while NT reads each drive, and windows 95 & 98 are WORSE. If I leave my
cartridge in AT ALL explorer scans _EACH _CD _EVERY _FEW _MINUTES!!!!
This causes NT to just FUBAR ALL the time. It is the single most
annoying thing in the world. With Linux I have never had anything
similar to that, and to tell you the truth I don't know why.
Anyone ever try to use WinAmp under NT? Anyone ever change the process
to 'Realtime' on a P-II 450, with a descent sound card? I have, it
skips every minute or two.... that is with just it running, and the
'blank screen' screen saver running.
I have used x11amp, and a variety of mp3 players under Linux... None of
them take very much CPU time, [3%, maybe 5% of a P-166], and the only
time that I heard it skip with when I accidentally cat'd a text file
into one of the /proc sysctl interfaces as root.. ["... it does not keep
you from doing stupid things..."] *oops*. (BTW: that did bring down
the whole system in a about 45 seconds). But good grief. Even when X
was running, and the mp3 player ran as a user process.. rock solid.
NT has so many holes, and small performance issues that I have just
come to realize that it is not cureable in it's current state. You can
show me all the Lies, Damn Lies, and Statistics you want, but anyone who
is worth thier salt can readily determine that You are better to spend
your software [OS] budget modifiying a particular area of Linux, be it
within the kernel, or userspace, than to spend endless dollars on things
that work for 5 minutes under a techinicans eye in a test laboratory.
Show me an NT box, that has a HUGE web load sustained for more than 24
hours in that state, and show me a Linux box next to it doing the same.
Then wait a month, and then you give me results. The NT box may
outperform the Linux box in a 5 minute test run. You may be able to put
some theoretical, never-happens-in-the-real-world test that shows that
NT can 'outdo' Linux. You may be able to manipulate the numbers to read
whatever you like but remember: All the MS Computers, and all the MS
Engineers could not make HotMail run on windows NT, and so they went
back to Sun's.
js
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/