[LINUX:20527] RE: Ynt: Re: mail server kurmak istiyorum

---------

New Message Reply About this list Date view Thread view Subject view Author view

From: Andreas Mueller (amu@tr.debian.net)
Date: Fri 13 Oct 2000 - 16:45:01 EEST


hi,

> > You forget the most importent factor, how many user use the
> > System/Services at the same time ?
> >
> Bandwidth'e ve zamana gore degisen seyler bunlar..
>

maybe there's a missunderstanding you can add 60.000 users a
a intel based system, but the most importend question is
how many users can connect to your system at the same moment/ at the same
second.

"Newest" AMD 64Bit Prozessors MircoProzessor send their datas to the
SystemController with 64bit Data + 8bit ECC, 13Bit SADDOUT , 13Bit
SADDIN, with 200MHz and max. 1.6Gb/sec Datas
( just a clever Idea to double to "Mhz" and double the Bit-rate from 32
to 64 to get double bandwidth )
So now the System Cotroller send his Datas to PCI-Bus ( 32 Bit ) Memory
( 64Bit ) that means this is a socalled "Bottleneck" to the PCI-Bus
you get Datas from the MircoProzessor with 1.6Gb/sec but you Bus can
handle only "800MB/sec". If you now using ISA, the datas are again bisects
to half, from PCI-Bus to Periferal Bus Controller, to the ISA-Bus ( 16Bit
) Also importent are the channels. PC's have 1 Channel ( INPUT,OUTPUT )
A RISC System has 1-30, with a Bandwidth of 3.2GB/sec 100Mhz ) of _EVERY_
channel. Most Intel Pentiums are still 32Bit, with 100Mhz ( 133 ) that means
Bandwidth of max. 800MB/sec. Thats still hight for a PC !
PC-Industrie is still working at a 64Bit Architecture, to make this
big performance little smaller. It takes some more Years.

The Differences between CISC ( PC's ) and RISC ( IBM,SUN )

CISC and RISC (Complex and Reduced Instruction set Computer,
respectively) are dominant processor architecture paradigms. Computers
of the two types are differentiated by the nature of the data
processing instructions sets interpreted by their central processing
units (CPUs). They both have advantages and drawbacks, which are
detailed below. To improve performance, CISC systems try to reduce the number
of instructions programms must call. To do this, they have large sets of
microcode instructions that cover a broad range of tasks. A single
microcode instruction, in turn, when translated in the CPU, may become
several tasks the processor performs. As a consequence, instructions
are of variable length and often require more than one clock cycle to
complete.

RISC systems, on the other hand, seek to improve performance by
reducing the number of clock cycles required to perform tasks. They
have small sets of simplified instructions, doing away with microcodes
altogether in most cases. While this means that tasks require more
instructions, instructions are all of the same length and usually
require only one clock cycle to complete. Because of this, RISC
systems are capable of processing instructions in parallel in a
process called pipelining. The CPU works on more than one instruction
at once by starting the second instruction before it completes the
first one. This greatly increases throughput and makes RISC systems
substantially faster than their CISC counterparts. RISC systems do
have a few disadvantages, however. Notably, because programs usually
have more instructions, compilers and applications written in assembly
language are more difficult to build for RISC systems.

Also see
http://www.inf.fh-dortmund.de/person/prof/si/risc/intro_to_risc/irt0_index.html
http://ciips.ee.uwa.edu.au/~morris/Courses/CA406/RISC_intro.html

If you're more interested in a discussion about Systemarchitecture
you're welcome to email me. I guess some people starting to disturb :)

: so here some jave-code,
import java.util.*;

public class test {

        public static void main( String args[] ) {
                int threadCount = 1;
                Hashtable ht = new Hashtable();
                for(int i=1; i<=40000; i++ ){
                        System.out.println("Creating thread # "+i);
                        ht.put(Integer.toString(i), new myThread(i));
                }
                for(int i=1; i<=40000; i++ ){
                        System.out.println("Starting thread # "+i);
                        ((myThread)ht.get(Integer.toString(i))).start();
                }
                //while(true){
                        Thread[] tarray = new
Thread[Thread.activeCount()+10];
                        int ThreadCount = Thread.enumerate( tarray );
                        System.out.println( "\n\n\n Active Thread
count: " + ThreadCount+ "\n\n\n" );
                //}
        }
}

amu

-- 

"The BeOS takes the best features from the major operating systems. It's got the power and flexibility of Unix, the interface and ease of use of the MacOS, and Minesweeper from Windows." ---------------------------------------------------------------------- Public-gpg-Key: http://tr.debian.net/amu/amu.key

Listeden cikmak icin: unsub linux mesajini listeci@bilkent.edu.tr adresine gonderiniz. Lutfen Listeci icin MIME / HTML / Turkce Aksan kullanmayin. Listeci arayuzu: http://listweb.bilkent.edu.tr/yardim/bilkent/linux.html Liste arsivinin adresi: http://listweb.bilkent.edu.tr/


New Message Reply About this list Date view Thread view Subject view Author view

---------

Bu arsiv hypermail 2b29 tarafindan uretilmistir.