From: "joe" <[EMAIL PROTECTED]>

> >Joe, you hit upon the reason I stick around and am considered a PITA
> >in some circles, particularly kernel circles. I keep reminding them
> >that it is possible to build a kernel such as NT has that demonstrates
> >remarkably good near-enough-to-real-time performance.
> >
> 
> nt? realtime? doesn't sound like any nt I've ever seen...

You have to set it up and use it right. Setting priorities properly
makes a whale of a difference. Even the GUI is reasonably fast for
updating. But if you leave that task lower priority than the task
that processes the cue lists or the video clips the screen update
gets done secondary to the more important things. NT has a real time
priority class. And it means it. This is as high or higher priority than
anything in the OS. Like I mentioned, we get outstanding performance.

> >which corroborated the general Linux "sluggish response" feel that
> >makes it a poor gaming platform.
> >
> 
> I've certainly found linux to be an acceptable game platform
> I'm not seeing the "sluggishness" you report.

X-Windows updates seem laggy to me. And some command responses
seem to lag more than on NT. NT's overall user interface makes it
feel slower than it is.

> In the 2.4 kernel, scalability and networking improved a great
> deal - back in the summer of 2000, linux started smashing
> specweb records with beta versions of the 2.4 kernel. Linux
> still holds the records for 1, 2 and 4 way web server results,
> last time I checked anyway -

I ain't doing web servers. We're not processing hundreds of requests
at a time. It's one thing at a time done REALLY fast that is the
ballgame. (NT's web server is somewhat <chuckle> slow. But it is
nowhere near the production code. It's only use is documentation
and (if the user is a silly masochist) the email.))

> There are still weaknesses for instance in disk I/O performance,
> but the block I/O in 2.5 has been rewritten and is more efficient
> and scalable. General latency is much improved as well, such that
> the "low latency" patches are no longer required for good gaming
> performance - the 2.5 kernel tends to be low latency by nature.

For the show control disk access is minimal in general. For the
video the disk access is viciously fast. (Two of the Spencer
Technology machines brought a fiber channel raid to its knees
trying to play 8 video clips at once at 7 to 10 MB/second quality.)

> >I dearly wish I had a practical alternative to NT/2K/XP.
> >
> 
> Practical alternative? sheesh, do we ever live in different worlds.
> I see windows nt/2k/xp every day - and I provide the tech support
> for my dear wife, who uses w2k.

And you think NT is practical? Right now I'm seeing the latest update
introduce instability into the email system. Outlook Express blows
itself away as I type in answers to email. It appears mshtml.dll is
screwed up. That's where it blows up. And the machines that do blow
up have the newest versions. (And because I use it for communicating
with my customers it is FAR easier to use it for general email than
try to fumble from system to system across standards boundaries for
email and development.)

> >Oh God how I wish! 
> >
> I use linux all day every day - and unlike you, I have no wish
> to find an alternative, since Linux gives me far less grief than
> windows ever did - (Well, there are problems, but they mostly
> boil down to market share and it's effects)

It is not ready for service as a clip server with special effects for
broadcast video; and it does not seem to be ready for the level of
show control work we do.

> >But Linux ain't there. 
> >
> 
> Not where?

Suitability for the kind of work that earns me my income. It may
never be with GPL in the way.

> I'll grant that microsoft has taken some baby steps towards
> better reliability and managability, but windows has still got a
> long way to go before I'd consider using it instead of Linux.

Enh, the video clip servers are up 24 hours a day and are used on
live video. We ever so strongly suggest to the customers that this
be the ONLY use to which the machines are subjected. Stability is
generally excellent. (But then, we never install bleeding edge
code that has not been vetted by Matrox. That helps.) And for the
Show Control that is pretty much the same rules of the game and
we get excellent uptimes in solid hardware. (Oddly, it is often
the cheaper hardware that performs most solidly. Go figure.)

{^_^}



-- 
Phoebe-list mailing list
[EMAIL PROTECTED]
https://listman.redhat.com/mailman/listinfo/phoebe-list

Reply via email to