In article <msbE575oA....@netcom.com>, Mark S. Bilk <m...@netcom.com> wrote:

YOu're not seriously suggesting that anybody that uses a handle should
have their service terminated by their ISP...are you?  If so, every
"root" user in the country would be terminated (by themself).  Then
where would you be?

I'm not jumping to Jim' defense here (so don't accuse me of doing so).
I'm not even condoning the use of handles (I actually don't like them).
I just find the argument without any merit.

Christopher C. Wood (chri...@lexis-nexis.com) wrote:

: We have another champion of the "Nobody will ever need more than 640K
: RAM" school of thought.  Should past trends continue (RAM densities
: quadrupling every four years), the "base" amount of RAM will cross 1GB
: in three more intervals; about 12 years.

I thought is was doubling every two years? <g,d&r>

  How soon before more manufacturors embrace multiple processors as
a standard feature?    The BeBox is already an inexpensive move in
this direction and manages to be pretty effective with only two
PowerPC cpu's and not much cache (if any).

Scott Nudds <af...@james.freenet.hamilton.on.ca> wrote in article
<5dakrk$...@james.freenet.hamilton.on.ca>...

In article <handleym-0502972336420...@handma.apple.com>,

--

Well, I've got Nudds in my "bozo bin" (Its like a kill file for the Gravity
newsreader that I am using) so I cannot respond to his post directly, but I
would like to just make a technical point (that I believe, only Jim here
seems to understand, even if he refuses to put it into plain english for
everyone to read it.)

The MPEG data originates typically from one of 3 sources: (1) live internet
transmission, (2) stored on CDROM, (3) stored on hard drive.

Now with each of these sources, one way or another the data is handled by a
non-display processor running in its own clock domain.  The display controller
or display processor is the intended target for the data, and exists in its
own clock domain.  These clock domains are driven by different clock crystals
and typically at different frequencies.

Now, to transfer data from one clock domain to another, computer architects
came up with the concept of a "data bus".  It turns out that these BUS's are
expensive things that are built into mother board; you probably have one
called the PCI bus (on the other hand, in Nudds case, who knows what obsolete
piece of garbage he has driving his useless Cyrix 5x86.)

Strangely enough the PCI bus attached to the x86 CPU (the investment you can
count on all PC users to have made) is the only generally supported interface
between all three sources above and a potential target display architecture.
Intel's fallacy is in believing that the right answer is just to pump up all
the clock rates and treat the display device generically, by throwing all of
the decompressed pixels at the display.  Chromatic's improvement is send raw
MPEG data (significantly compressed) over the bus to a smart display device
which will do the decompression for you.  This is a very light load on the
x86 which allows for many MPEG streams to be processed simultaneously (we
can do this *TODAY*), something Intel will not be able to claim for many,
many years to come.

The alternative of somehow piping data straight from the CDROM to display
device (whether it handle it as an overlay or not is irrelevant) still
requires the construction of some sort of BUS, some sort of physical wiring
of your CDROM to your display adapter and excludes the other two forms of
MPEG sources.

The alternative of integrating an overlay display processor into a CDROM
drive processor makes no sense since the clock rates of CDROMs and displays
are so radically different as to make design of such a processor practically
infeasible.  It precludes the possibility of doing multiple MPEG streams
executing simultaneously, as well as the other two forms of MPEG sources
described above.

That is just talking technology.  Now if we talk economics, its a no brainer
decision yet again.  If you've got a PC, then already have a x86 processor
and PCI bus.  What's more, you also own a CDROM, hard drive, and potentially
a modem, that doesn't have any provisions for slapping on an extra data bus
or display processor on it anyway.  You would have to somehow convince
consumers that they need a different, more expensive, kind of computer
architecture.

Nudds, your pathetic attempt to show me up on the x86 domain (where my
knowledge and skill over yours is only about 5 times) was bad enough.  Now
you dare challenge me in the multimedia domain?  You don't know what you
are getting yourself into.

--
Paul Hsieh
qed "at" chromatic "dot" com
http://www.geocities.com/SiliconValley/9498
Graphics Programmer
Chromatic Research

<stripped out a few of the newsgroups from this follow-up>

I've noticed that Jim Brooks had pretended to be other users in
various sections of this thread.  Whether these users exist or
not in rl is beside the point.  If I change my header to point
to rcoll...@x86.org, and post an inflamatory message, I imagine
that you would be pretty annoyed.

----------------------------------------------------------
Elda Inc.       - Eastern European Translations

Is it true that Intel subcontracts the motherboard component work for
any new release of a chip?  Example, I know that the P7 Merced is ready
for production and I heard that they (Intel) are waiting for the
components to be built that will work with the faster chip, etc.  So I
was wondering, if this is true, does anyone know who these companies
might be or how I would find out who they were? Any other opinions,
welcome.

On Sun, 09 Feb 1997 17:21:39 -0600, God <gcro...@aoc-resins.com>
wrote:

] I don't see very much difference between using a "handle" and
] "falsifying your headers to intentionally misrepresent your identity."
] There's hardly any difference.  In one case, a handle s perfectly
] normal, commonplace, acceptible, and even condoned (on AOL) in some
] areas. So what' the big difference between using a handle and
] falsifying news headers?  The end results are identicle.
]
-snip-
]
] I'm not jumping to Jim' defense here (so don't accuse me of doing so).
] I'm not even condoning the use of handles (I actually don't like
] Ithem). just find the argument without any merit.

The difference is between somebody with the name Mark Rhodes saying call
me Dusty, and the same guy saying his name is Robert Collins on
slip.net.

Daniel R. Haney (salvar...@std.com) wrote:
:   How soon before more manufacturors embrace multiple processors as
: a standard feature?

Christopher C. Wood (chri...@lexis-nexis.com) wrote:
: In article <5dakrk$...@james.freenet.hamilton.on.ca>, af...@james.freenet.hamilton.on.ca (Scott Nudds) writes:
: |> Ted <ted...@tedros.demon.co.uk> wrote:
: |> : So in two to three years you can expect to have a PC with around 16
: |> : gigabytes of ram 32 megs of cache  credit card 10 gigs of flash HDD
: |> : and a trailblazing 1GHz 3D chip...........anyone like to bet!!!!!

: |>   Any betting person should take you up on this offer if they would
: |> like to make some easy cash at your expense.

: |>   No personal computer will ever have gigabytes of RAM, just as no
: |> automobile has giga-gallon gas tanks.

: We have another champion of the "Nobody will ever need more than 640K
: RAM" school of thought.  Should past trends continue (RAM densities
: quadrupling every four years), the "base" amount of RAM will cross 1GB
: in three more intervals; about 12 years.

I meant dropped NT for MIPS and PowerPC. The MIPS processor is still a
very viable CPU. It's found in the Silicon Graphics range for a start.

In article <5dakrk$...@james.freenet.hamilton.on.ca>,

: Yes, Scott, your remark about Paul's description is, frankly, dumb, very dumb.

  MPEG data can come from a variety of sources, I have seen it come from
CD-ROM, transmitted over a network, and from various professional video
boxes and from a couple of VCR's.

  Whatever the source, it should be very clear that sucking the data off
of an I/O port into a CPU and then depositing it into memory, reading it
back from memory decoding it and writing it back to memory, and then
reading it from memory again and depositing it onto the screen is a
hopelessly inefficient process, that places a severe burden on the PC.

  A much better solution would be to decompress the video stream with a
secondary processor external to the PC, and overlay the result on top of
the existing PC video.  If a CD-ROM is the data source, data should be
taken off of it and sent directly to the MPEG decompression hardware
in exactly the same manner sound from a CD-ROM is now routed directly
to sound cards.

  Anything other approach to displaying real time video on the PC is
<BRAIN DEAD>.

  I think that video data should be <overlayed> onto the SVGA video
signal and should never reside in the SVGA video buffer.

  Yes. Exactly how this is done would depends on the CD-ROM.  Direct
connection to a digital port on the CD-ROM would be the best method, but
most CD-ROM's are now built without these ports due to the <BRAIN DEAD>
decision to suck MPEG data over the IDE bus.

  Brain dead.

  For IDE CD-ROM's without a digital video connector, plug the CD-ROM
into the MPEG card and use an on board CPU to read the data, decompress
it and overlay it onto the existing VGA video signal.

  This would reduce PC-CPU overhead to zero or near zero.

  If PC video presentation from CD-ROM is ever going to succeed, this
will eventually be the technique adopted.  Performance demands it.

  False economy.  The user saves $100 on a new CD-ROM but pays by having
his $2000 PC reduced to $10 XT speed during video playback.

  Do it properly, or don't do it at all.

  If they wish to survive they will have no choice.  People will accept
an application sucking up 100% of system resources only as long as it is
a novelty.

  Exponential growth is impossible to sustain for any appreciable
length of time as a practical matter.  The fact that current growth is
exponential means that in short order we can expect it to abruptly halt.
But this is not the reason PC's will never have gigabytes of RAM.

  The reason is simple.  Somewhere in the 50 to 200 megabyte range, all
applications, (or at least their active portion), will reside in memory.
Doubling memory may allow the entire set of applications to reside in
memory, but the performance gain will be small.  The larger the memory
capacity, the smaller the gain.

  As a result, additional RAM memory will not be added.

  And how many more years before the count of bits is equivalent to the
number of silicon atoms in earths crust?

  Does exponential growth confuse you?

: Kristian Thommesen, science student at the University of Aarhus.

  How sad.

: Scott Nudds wrote
: >   No personal computer will ever have gigabytes of RAM, just as no
: > automobile has giga-gallon gas tanks.

: I might agree with you that there is a practical limit of usefulness
: for that much RAM on the *average* consumer computer, but applications
: such as graphic image editing, or movie editing could easily chew
: up gigabytes of RAM in the future.

  A 24 bit color image, scanned at 1000 DPI over an entire 8.25 * 11 inch
page will only require .27 gigabytes of RAM.

  "Gigabytes" of RAM implies 2 or more gigabytes.  2 gigabytes would
allow a user to store 10 such images completely in RAM.

  There is no practical purpose for doing this.

  Even with disk I/O with a speed of 5 megabytes/sec, loading or saving
these images to disk would take 6.7 min.

  Loading them off of a 20 * CD-ROM would take 10.2 min.

  Transmitting them over a 10 Mbps network connection would take 33 min.

  Transmitting them over a 1 Mbps modem would take 5 hours.

  Transmitting them over a 56k bps modem would take 4 days.

  Isn't it amazing the number of children who don't understand the
limits of exponential growth?

Dear Uncle Jeb:  I have no wish or intent to post to comp.sys.intel.
But postings from there echoed to asm.x86 will be answered if I feel a
response is appropriate.

I have not visited comp.sys.intel, nor have I modified any message
header.

If my posts are appearing there, then I can assure you the person
responsible is a member of the C religion who is seeking to enlist
others to attack me by disrupting yet another sig.

I urge you to identify who the childish individual is, identify him
publicly, and take whatever steps you see fit to prevent this from
happening again.

On 5 Feb 1997 13:51:32 -0500, this dude,

Christopher Michael Jones (cjo...@appendix.cs.uoregon.edu) wrote:
: By the way, comparing cars and computers ain't a good idea.
: Many people have noted that if the automobile industry was like the computer
: industry, you should be able to buy a car for about a dollar and it would
: get billions of miles to the gallon and travel around the world in a few
: seconds.

... and if you turned signaled for a left turn while the radio was on, the car
would instantaneously explode, killing everyone inside.

---N