Re: TI-H: heh

To: ti-hardware@lists.ticalc.org
Subject: Re: TI-H: heh
From: Grant Stockly <gussie@alaska.net>
Date: Sat, 28 Nov 1998 21:05:35 -0900
Delivered-To: ti-hardware-outgoing@towerguard.unix.edu.sollentuna.se
In-Reply-To: <912311077.22745.930@excite.com>
Reply-To: ti-hardware@lists.ticalc.org

>You are right. I wasn't inclined to help him because I hate macs. It did
>take a long time because he was forcing Virtual Memory, but he took that off
>and programs ran slower than my computer. The only thing a mac is good for
>is BeOS and maybe even linux. But since x86's run Linux, NT, and BeOS, then
>they are pretty much the way to go.

You havn't seen BeOS run on an intel have you.  :)  No network support yet,
and even on a 166 it skipps when playing 2 movies at once.

>If they weren't, then every linux user
>would be buying PPC's off of surplus or getting them out of the trash and
>using them for a linux server...

They are.  Have you ever seen a PPC amiga?  They are faster than anything
you've ever seen.

Or how about Alphas?  Any serious linux user has some of them.  They are
cheep though, $400 for a stripped 350MHz.

>I personally LOVE pissing off my mac friend
>about the tests they do with PPC's...

THey do all their tests in a realtime mode.

>Steve jobs tried to show off the G3 by
>comparing Mac based programs on a wintel PII and on the G3. It makes you
>wonder when G3 233 has completed something, when the PII 300 is barely 10%
>done...

Look at it this way...  The PPC 750 in realtime mode (not thinking abou the
OS) runs at ~400 BOGOs.  The intel at 300MHz runs at about 190BOGOs.  Get
it?  And windows hates giving realtime access to programs, so most tests
are slow.

>I guess Intel really blows, huh?

Yep.  They want to buy the copper trace technology.  The reason why IBMs
processors are faster than intel's is because intel uses aluminum instead
of copper.  IBM uses copper.  Copper is a much better conductor.

>They should try doing some
>scientific tests, instead of bullshit tests.

They have results of those.  You just need to open your eyes.

>Maybe then I'd consider getting
>a ppc or two. Right now I'm looking to get an old risc computer to play
>with.

RISC doesn't mean it will be faster.  It could be slower.  But in IBMs case
it is faster.  Remeber the copper issue, and the fact their instruction
cycle is 1/1.

Here is an example.  We have 2 RISC uCs.  The PIC and the AVR.  The avr's
instruction cycle is 1/1.  The PICs is 1/4.  Both are RISC, but one is
slower.  It takes a 33MHz PIC to equal the power of a 8MHz AVR.

The instruction cycle of the processors intel makes is 1/3.  IBM is 1/1.

Aluminum and some of the CMOS technology is what limits the speed.  If you
were to output bytes to the bus and then read the state of the bus, the new
bytes that you had just read to the port wouldn't have even been applied.
That is why they do 1/3.  But when you can eliminate that delay, and other
issues, 1/1 is possible.

The AVR can output a byte, and read the port and get the new written byte.
The PIC documentation and past knowlege recomends that you insert a NOP (no
operation) instruction after every I/O operation to make sure everything
went well...and the PIC is 1/4.

Don't get me wrong, I'm not a sterotyped mac user.  I use linux as much as
possible.  If I took 4 monitors that were generic, put linux on a SUN,
Intel, PPC, Alpha, Amiga (I could go on) you wouldn't tell the difference.
It comes down to what you can do on the processor, how fast...

You can even bypas the GUI of the macs.  You can either internally clip a
resistor, or every time the computer starts up hold down control+option+o+f
and it will let you do what ever you want.  Run linux, BSD/Unix, BeOS, or
you can write your own OS...

Grant
References:
Re: TI-H: heh
From: "Dan" <danb@excite.com>