[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [f-cpu] F-CPU architecture...



hi,

woodelf wrote:

Take great care with negative numbers like 2^8, 2^16 , 2^32 ... .

if this is the only problem, i can live with it.
Try computing with one's complement ...
you will find that no representation is ideal.

and as well as advantages as compared to other number systems and as in hardware designs . You still have people doing computer science with may other ways of looking at computing and that may be the way of future with the F-cpu.

so ... what ?


You have a clean piece of paper to design from.

It is not "clean" anymore. countless people have scribbled over it during the last 7 years.

A few simple new ideas
may be what it takes to change computer algorithms used and may give
a better design than copying blindly just what is done.

Today, in "Computer Science" , a lot of things are copied because those are the best working methods. Or at least because they exist. Not everybody can spend years creating a whole new compiler for example. I'm also currently investigating the CRC algorithms, and few methods are optimal. And as long as the maths are unchallenged, i don't see why the already optimal algo would change.

Looking at the old F-cpu design

which ?

is the model of the cpu design well understood?
Can you list the advantages and disadvantages of the model compared to
other design models. This may be the time to try different models
of operation. Remember RISC machines calculate effective addresses
well,they don't often do other operations better other than a side
effect of the design.

now, take some weeks to read everything at http://f-cpu.seul.org/mail/ Maybe you will understand my despair and why F-CPU has not yet "succeeded".

There are countless people with revolutionary ideas and unbeatable arguments
about why everybody should follow. I am among those wide open mouths
and this behaviour has taken me to my current, difficult position as one of the project's critical men.


But almost nobody does what one says.
Michael Riepe is one of the few counter-examples.

you will maybe soon realize that in the F-CPU project, advances
don't come with ideas, but work and compromise : this is how FC0
was born. It did not come only with "A few simple new ideas"
but also ways to make them look insignificant and yet easily implementable.
And because many people had one's own "groundbraking ideas"
that are exclusive of the others, the only way to make the project
advance is to make something everybody can understand and do.

So please understand that " This may be the time to try different models
of operation." is NOT the way for FC0. Maybe FC1, IF we ever
finish FC0. FC0's architecture and instruction set was frozen in 2001.
Any "feature request" is not possible because it can break the architecture's
fragile balance and complex compromises.


And most people miss the real point : FC0 is not groundbreaking because
it is "new", "better than others", "faster" or "sexier". It is because IT IS FREE.
All the rest comes from this. I don't care if FC0 runs at 10MHz or 1MHz,
if it RUNs and is FREE (as in freedom). Freedom has a cost, which
means here a hard work that not many people can or want to do.


Another point is that there is already a lot of work done.
F-CPU has spent a lot of time in the middle of nowhere
because each "new great idea" required everything to be
rebuilt from zero, which is definitely stupid and a loss of work
and waste of time.

But nobody here is forced to "follow the line of the party".
We are all free. anybody can create a "alternative project", who cares
(this has already happened, so what ?). Everybody is free to play with
one's ideas (this is what i do with the VSP). This is a much faster
way to test them and see what goes wrong.

I don't want to sound harsh or impolite, i just want to reinject some
sense of reality in the *whole* mailing list. I feel like the same situation is coming back
again and again, we're discussing "features" which lead to nowhere, while "implementation"
always lacks. And we divert development effort in fruitless babbling and ranting.


@+

Ben alias woodelf

YG

*************************************************************
To unsubscribe, send an e-mail to majordomo@xxxxxxxx with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/