[ydl-gen] Cell Processor - correction
Derick Centeno
aguilarojo at verizon.net
Tue Oct 2 07:56:53 MDT 2007
Hi Thomas,
A short response to your questions, in order are:
Yes. Thinking. No. IBM.
Now that we are all clear, I'll elaborate just a wee bit because the
references I'm going to refer you to are quite a forest of
information -- and there are no machetes (shortcuts).
Difficulty:
The essential problem programmers face (together with nearly everyone
else) has been described within Philosophy and Psychology as a
Paradigm problem. In this particular discussion this essentially
means that a structured set of ideas are written into code for
machines to implement. The instructions or code presented to the
machines may be inventive, but they are nonetheless structured and
follow a series of collectively agreed upon assumptions. It is
debatable whether these assumptions are easily changed prior to
implementing the instructions in building either the software or the
hardware. The different skills and services involved do influence
one another in powerful and surprising ways; when an idea is
implemented it always requires study and review regarding it's
effectiveness and value from the perspectives of the various engaged
disciplines. Simply stated sometimes there exists a lag where
apparently endless discussions addressing apparent minutiae occurs
and then just as suddenly there are spurts or bursts of insightful
and astounding comprehension.
A brief consideration of what went into IBM's Cell: 30 years of
research of world class scientists, which includes the research and
utilization of supercooled metals which contributes to removing
friction (expressed as heat) such that electrons move through
circuitry effortlessly; the implementation of advanced mathematics
brought via the insight gained from exploring Benoit Mandelbrot's (an
IBM Fellow Emeritus) fractals. These contributions, and more, are
what IBM's hardware/software engineers brought forth as the Cell.
The software engineers and programmers, external to IBM are still
learning what this thing can do.
What is already clear however is that although all the already
commonly available code prepared for Intel and other commonly
available (or even earlier generation PowerPC) processors can be
executed upon recompilation within the Cell -- none of it approaches
using a fraction of the capacity of any one of the available PPC
cores or SPUs, nor does anything currently coded exist to fully
implement the capacity available in the Cell.
This is the "sticking point" because now the problem is not the Cell;
it is rather what to do with it. This is something akin to a 17 year
old having a Lamborghini for $20. The difference is that while s/he
can work out her/his impulses via the PS3 Game OS s/he can also be on
the same "floor" -- together with the best programmers on the planet
-- in learning how to implement the Cell well. Now that's a level
"playing field"! Consider the 17 year old who opened up the iPhone,
he demonstrated the opportunity for a young mind free of learned
structured assumptions has to do fresh programming/engineering that
had not been conceived of. This is both a real problem and
opportunity. In our time titles nor laurels are as important as
solid study skills, disciplined determination and insight dedicated
to constructive learning.
Although I'm discussing just the field of computing, the overall
process and problem of a paradigm was elaborated and explored in more
detail by Thomas Kuhn in his work "The Structure of Scientific
Revolutions" (ISBN-13: 978-0226458083).
Tracking processors:
The question you raise is really a very interesting one because it
reflects the traditional view or approach of current professionals in
nearly every field. It is because of this collective mindset that
discovering anything really new or proceeding in a new direction in
software (or any other endeavor -- take your pick: Politics,
Religion, Whatever) is very much a challenge. Upon serious and deep
analysis nearly any traditional approach (in nearly every field) is a
modified rehash of whatever has been attempted before in one or other
culture or even period of history. This is exactly where (and very
few human societies are ready for this) what is "tried and true" --
fails. Human Societies are terrible at adapting to new situations
quickly or well, many choose denial as the preferred method in
confronting oncoming realities.
However, the opportunity remains for those who will resist succumbing
to denial or the collective mindset surrounding their environment, to
apply constructive effort towards developing a thorough foundation so
that constructive new developments and contributions can be explored,
tested and implemented. Engaging upon such a path is difficult for
anyone because it is much easier to cooperate and get paid, by the
commonly accepted and established ways of participating with the
"tried and true" way of doing things. Remember that Da Vinci engaged
in all the madness of his time never allowing any hint or indication
of the personal research or effort he was engaged upon deeply. What
is very interesting in our time is that open source, as a concept,
may counter the need or the impulse to respond as Da Vinci did to a
dangerous and still mostly superstitious world.
The opportunity is very real, but seeing it and implementing it are
challenges which not many will, or can accept. To help out just a
bit, the cores or SPUs are aware of each other and can be instructed
to work in tandem or independently; the one SPU which Sony reserves
and restricts YDL (or any Linux) from accessing doesn't affect the
very real problem that programmers don't know enough how to take
advantage of the available SPUs anyway. Not yet.
As I said, the problem is Thinking. There is a nice little book
written by Martin Heidegger entitled "What is Called
Thinking?" (ISBN-13: 978-0060905286) which could be useful in
preparing to address the task of approaching a new direction
intelligently. A 17 year old wouldn't need that, we however do. We
need to rediscover what beginning anew and moving forward actually
means. The youngster by her/his very nature is ready to explore and
investigate everything and so s/he does. Our advantage however is
what the youngster has in energy; we have, or should have, in ...
Of course, how that sentence is completed by a person addresses
another layer of the challenge.
More technically relevant references are available within these lists
from within the archives:
From: lscharf at vt.edu
Subject: Re: [ydl-gen] So, PS3
Date: November 20, 2006 10:14:37 AM EST
To: yellowdog-general at lists.terrasoftsolutions.com
also
From: bmueller at terrasoftsolutions.com
Subject: Re: [ydl-gen] So, PS3
Date: November 20, 2006 12:27:49 PM EST
To: yellowdog-general at lists.terrasoftsolutions.com
and
From: jonathan at newmedio.com
Subject: [ydl-gen] Programming the PS3 Introduction
Date: January 4, 2007 8:57:16 AM EST
To: yellowdog-general at lists.terrasoftsolutions.com
Explanation:
Type the subject title as listed here into Google. A list of all
participants discussing that subject will appear. The dates listed
above will help you identify the specific comment together with the
links those persons provided to other references which could help you
further.
Good Luck....
On Oct 1, 2007, at 4:31 PM, Thomas A. McGonagle wrote:
> Hello All,
> Through the mainstream press, I have often heard just how hard it
> is for game developers to develop their games for the PS3's Cell
> Processor.
>
> Have any Cell Processor programmers found it terribly hard? What
> causes the difficulty? Do you need to keep track of which processor ?
>
> Can anyone recommend a resource to learn more about the Cell
> processor? I am particularly interested in why only 6 PPUs are
> available to Yellow Dog, and not all 8.
>
> Thank you very much for the help, and have a nice day!
> -Tom
================================
Life only demands from you the strength that you possess. Only one
feat is possible; not to run away.
-- Dag Hammarskjold (July 29, 1905 - September 18, 1961)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.terrasoftsolutions.com/pipermail/yellowdog-general/attachments/20071002/709b3668/attachment.html
More information about the yellowdog-general
mailing list