[OT] really?
Derick Centeno
aguilarojo at verizon.net
Thu Jun 9 12:02:27 MDT 2005
Thanks to Owen for this link regarding the current position of YDL
regarding this announcement:
http://lwn.net/Articles/138807/
========================================================================
=====================================
The following was posted by Andres Tello Abrego <criptos at allox.com>
under the subject "Intel on Mac". I presented it here because he
raised interesting questions which should be part of this discussion:
For me, this are very bad news... I really love the PowerPC
achitecture... way to much better than intel. x86 is CISC with
something internal that appers to be risc...
PowerPC architecture, is beautiful.. a lot of register, non destructive
assambler instructions, ortogonal register handling...
My Ibook g3 900 with 640mb ram at ram, kick ass a new dell pIV 2.8 ghz
512 mb ram, at cpu intensive task, sometimes my ibook is faster, the
hardware behaves better...
The reason I switched to mac, was to use linux at powerpc...
Now mac is going intel, what would be the real diference between a
ibook intel and a IBM/HP/DELL laptop? Even hp have a cooler hardware as
the tablet pc, something apple invented years ago with the newton...
and died...
For me, powerpc, was cost/effective reason to buy an ibook instead a
x86 system based...
I hope we will need to wait. They said they have been unable to make g5
laptops... what technology for building the chips are they using?
========================================================================
==================================
My own views rely upon the article reported in the June 2005 issue of
Laptop, which appears on p. 28, entitled: "The Rise of Dual Core".
The article is written for non-engineers, but it does explain several
salient problems in CPU design very well. Although the context of the
article does discuss Intel design exclusively a few members of this
list are very astute and comprehend the engineering argument in great
detail and how these design problems affect non-Intel architectures --
including the PowerPC. For the rest, it is enough to remember that one
of the reasons the G5 is limited in regards to being released as a CPU
used in a laptop is because removing heat from it remains a challenge
although researchers at MIT had developed a solution which was
reportedly presented to Apple which then discussed the use of specially
designed "ribbons" as a method of cooling and redistributing the CPU's
heat. I don't recall more than that unfortunately.
In any case, the solution was obviously has yet to be implemented for
technical reasons. Meanwhile Intel's efforts has led to the very real
possibility of having as many as 15 processors on one chip and at least
one manufacturer is seriously looking into this. Considering that the
PowerPC as sold by Apple is limited to 2 whereas IBM sells their own
version of the chip in 4, 8 and higher CPU modules on their PowerPC
servers; Intel's dual core development for the home pc market is
definitely a winning argument especially as viruses will not be able to
handle all the various internal checks between the cores and from the
cores to the OS, especially OS X which is perhaps closer to a fault
tolerant OS for the home as an OS can get.
Finally, true fault tolerance and redundancy on a chip! This has been
a dream for many professionals especially as this nullifies hackers,
etc. when the proper settings and recommendations are employed by
management and astute system administrators.
Mac OS X and beyond on Intel could allow for products such as
StarOffice to finally be available for the OS, instead of just it's
poor cousin, OpenOffice.org. And what about the likes of AutoCAD, and
many others. If the processor is standard and truly as powerful as
described, we may just be on the verge of a resurrection of a turf
battle between OS's which this time is way more even and far more
clearly evaluated by the public.
I have every confidence that the Mac OS is superior not because of
faith or mere hope, but because of engineering and I believe that as
Intel becomes the new chip in the heart of a Mac others will discover
more of what makes the Mac OS so compelling.
Regarding the actual differences in powers of calculation of extremely
large and small values important in science and some unique business
applications there is a nice program which appeared originally in
"Applied Numerical Methods in C" by Shoichiro Nakamura which identifies
something he names as the machine epsilon. The machine epsilon is the
discrimination value which that CPU cannot surpass. He also published
a chart in the text on some machines he ran the program on revealing
their machine epsilon; I further embellished that program to include
tables of comparison on other systems available to me.
If one was to compare the values of the machine epsilon across the
various architectures one would clearly see that the machine epsilon of
Intel compared to PowerPC CPUs were closer. The table and the program
were posted on the YDL FAQ pages but for technical reasons it is no
longer there; so if there is an interest I'll publish an output of the
program for others who may be interested to review on the YDL
discussion board if I can find an appropriate space. Prof. Nakamura's
program originally evaluates single precision values only, mine
evaluates single and double precision. However, as any C programmer
knows writing such a thing is pretty simple and so publishing the table
I was able to generate should be enough to have a sense of the
different CPUs.
So welcome all to the new challenges ahead...
Best wishes...
On Jun 7, 2005, at 2:37 PM, rev rob wrote:
> So, does all this talk of apple using intel processors really mean one
> could, in the future, install Mac os x or whatever new Mac os on a
> dell or gateway or whateves?
> OR is there going to be a new architecture for apple hardware unlike
> the architecture for windo$e boxes?
> I'm sure this has been answered already but I'm trying to read
> between the lines to sort this out, as it were.
> Thanks for your time
> rob
> _______________________________________________
> yellowdog-general mailing list
> yellowdog-general at lists.terrasoftsolutions.com
> http://lists.terrasoftsolutions.com/mailman/listinfo/yellowdog-general
> HINT: to Google archives, try '<keywords> site:terrasoftsolutions.com'
>
More information about the yellowdog-general
mailing list