[MUD-Dev] Noise: boys will be boys (was Re: [MUD-Dev] Multi-threaded mud server.)

Ola Fosheim Grøstad <olag@ifi.uio.no> Ola Fosheim Grøstad <olag@ifi.uio.no>
Wed May 19 02:16:20 CEST 1999


Chris Gray wrote:
> Sniff! Mine's only 300 Mhz. :-)

ohshudupmineis100mhzandincrediblyfastandiamnotgoingtobuyanewoneunlessitistentimesfasterandhalfthepricecanihaveyoursplease?

>     - I grew up with processors that were about 1000 times slower than
>         the one you mention, and that has built some habits that are
>         hard to get rid of. (I had to think about it for a minute, but
>         1000 *is* the right number!)

Only 1000???  Hardly impressive for a man at your age!!!  Even *I*, a snotty
twentysomething, can beat that!  My CBM64 is probably 1200 times slower than
my P100 is ((32bit/8bit)*(3cycles/1cycle)*100), and that is for simple
integers.  If we are talking floatpoint, expect to see numbers approaching
100000.  Not to mention that it takes two minutes for this multiprocessor
machinegun to load 64kb or so from a diskdrive, for cassette it would be...
forget it. 

More importantantly, I spent my pimplefaced learning years with this
beauty.  She took me with storm and I was faithful!  Tickling her inner core
every night pushing the right buttons with gentle movements.  I still
remember some of the opcodes of love.  And I am still sleeping with her
every night. EAEAEAEAEAEAEAEA  She is waiting for me, can't you hear?  And
she is never going to break away from me.  She is me! 00

>     - I have *BIG* dreams. I have a friend who works for one of the
>         largest ISP's in Canada. I want to try putting my MUD server
>         on a box in one of their network centers. That'll be a big,
>         fast box (likely SMP) with a ton of memory and a great big
>         pipe to connect it to the internet. I figure that the system
>         networking overhead will tie up a CPU or two, and that'll leave
>         one or two for my server process, and it'll need to handle
>         1000 or more active clients to keep the pipe going. Remember
>         that I said this was a dream!

MY goals are B I G G E R.  MUCH BIGGER!  I am going to write a distributed
server that is also a sneaky worm.  It will eat all the spare time on all
the servers on the entire internet. And then some.  My client will be
implemented as a Word macro virus that displays hypnotic patterns
communicating seductive messsages. "play with me, play with me, play with
me".  Then I will be God.

Chris, you really need to understand that your dream will probably take
another 10 years or so to implement, and by that time we probably have a new
programming paradigm, a new architecture, computers will be 1000 times
faster than now, free muds are better than the commercial muds of today and
you will have to rewrite your server in a new multiprocessing language
anyway.  Let it go, there is a better way!  You see, my system will not have
to change much.  Word macros will still exists, and all the servers on the
net will be NT.  Switch to Word NOW and you will have a truly fast MUD in
ten years. I promise!

<lessnoisebutstillnoise>
Seriously, I am wondering if understanding how the processor pipeline works,
caching etc, etc, is getting in the way when I use languages like C and C++
where I _know_ how this will look as machinecode, and what the consequences
are.  I am an addict.  I am addicted to the fictive power that comes with
writing efficient code.

I like to believe that I used to write better code in higher level languages
which were inherently and unpredictably slow.  Then I would only optimize in
O() notation. In C I think on the bitlevel.
</lessnoisebutstillnoise>


--
Am I looking for an excuse for writing tight and messy code?  I guess...




_______________________________________________
MUD-Dev maillist  -  MUD-Dev at kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev




More information about the mud-dev-archive mailing list