[ydl-gen] Linux yellow dog manual for dummies

[Rob Sanders]

It is an interesting article, and reminds me of some discussions I've  
had with 'younger'
programmers at work.  One of my coworkers is an absolutely awesome  
web developer, using
the newer frameworks (Groovy on Grails) to develop from scratch a  
replacement GUI for out
linux security tools (commercial plug - www.trustedcs.com/ 
securityblanket) in 8 months, having
no previous experience with the product.  Seeing what he has been  
able to do using the right
tools has been amazing.  He enjoys the whole meta-programming way of  
doing things.  But when
I've discussed some of the things I've done (some cell code, device  
driver work, cross-platform porting,
C/C++/Fortran, bit level manipulation, etc) he honestly confesses  
that he is baffled and lost trying to
do those things.  They just aren't taught much it seems.

I've also been amused that several of the programming tricks  
*required* in cell based work were things
I did 'back in the day' trying to get every cycle I could beg,  
borrow, or steal out of my old Apple II+.  When
a no-op costs two clock cycles, you learn to count every one of  
them.  Anyone else remember the extra
16 K of memory (or there abouts) you could get by bank-switching the  
language card?  Or programming
overlays?  Manually figuring out how to bit pack you data so you  
would actually be able to fit everthing
into your limited memory space?  Again, skills that (outside of the  
embedded world perhaps) just aren't
taught much anymore.  <sigh>  I'm not saying that everyone needs to  
be able to demonstrate mastery
of these skills, but I am a *firm* believer that the concepts should  
be taught and coded at least once,
if only so the developing code can understand what the wiz-bang  
compiler is doing on their behalf.

Ok, I will now step down from one of my many soapboxes.....

-Rob

On Dec 27, 2009, at 11:23 AM, Derick Centeno wrote:

> I thought you, Rob and interested others would find this article
> interesting as it follows along the lines of what you stated earlier
> with some different details.
>
> First the article:
> http://arstechnica.com/hardware/news/2009/11/end-of-the-line-for- 
> ibms-cell.ars
>
> You'll note that future development and computers (including laptops -
> Yippee!) will follow along an integrated or morphed design path which
> implements both AMD/Intel with Cell/PowerPC aspects of handling  
> complex
> processing.  This could well mean systems which look more like the
> system which is within the Toshiba Qosimo (referring to what Toshiba
> calls it's SPEC engine).
>
> I recall discussing details regarding how the Cell's technology was
> morphed into the SPEC engine with some astute people at the YDL Board
> awhile ago.  IBM's decision therefore kills the Cell as Cell, but not
> what was derived by learning and working with it.
>
> It is interesting that this movement forward comprises methods
> previously exclusive to disparate architecture families (Intel v.
> PowerPC); everything running today will need to be rewritten for those
> newer systems as they will be unlike what has come before. This also
> means something rather challenging which I hinted at previously, the
> skills needed to program the new systems will require people familiar
> programming for PowerPC/Cell and Intel systems.  A skill which remains
> rare and becoming rarer still.  However those skills are transferable
> and advantageous in leaping forward into seriously working with  
> this new
> hybrid.  The new hybrid will be further advanced than the system
> within the Qosimo.
>
> As the hardware moving forward has changed so must Linux and  
> commercial
> operating systems.  YDL won't remain what it is; it will have to morph
> into something else to meet what is coming.  That choice, if any is
> made, is up to Fixstars.  I hope that they choose well so that they  
> are
> right in the thick of it.
>
> The best to all...
>
> On Sat, 26 Dec 2009 12:46:22 -0800
> Warren Nagourney <warren at phys.washington.edu> wrote:
>
>> That's very interesting, Rob. Were the apps compiled using the same
>> compilers (with the same degree of optimization) in both cases? Were
>> the time differences actual CPU time of just elapsed time? ...
>>
>
>
> =========
>
> Refranes/Popular sayings:
> The Taino say:No hay mal que por bien no venga.
> There is no evil out of which good cannot blossom.
> _______________________________________________
> yellowdog-general mailing list - yellowdog-general at lists.fixstars.com
> Unsuscribe info: http://lists.fixstars.com/mailman/listinfo/ 
> yellowdog-general
> HINT: to Google archives, try  '&lt;keywords> site:us.fixstars.com'