Advantages of Parallel Hz

[Guy Macon]

Couldn't circuitry be designed in such a way that the leakage current
will travel in a path out of the system, instead of toward vital
circuits?

This troll was clearly compiled with inferior tools.
My guess is that it was created with Visual Troll++
and the Troll Foundation Class, or possibly TurboTroll
2000.

These trolling tools are quite limited, and there is
a severe garbage-collection related performance hit
when you try to optimizing the output of VT++ for
kook-trolling.

IMO, you will get better results using GTC; the Gnu
Troller Collection. GTC is the gold standard for
creating trolls. It is also open Source, fully
reentrant, and compliant with the Triple Troll,
Troll-On-Troll and YATC protocols.

 

[MooseFET]

Couldn't circuitry be designed in such a way that the leakage current
will travel in a path out of the system, instead of toward vital
circuits?

No, the leakage current creates a power loss in the transistors that
are needed for operation. Some of these transistors must have a
voltage drop on them or your circuit can't change any logic levels.
Any transistor with a drop on it will leak and thus generate heat.
Since your idea increases the number of transistors needed, the heat
will be a huge problem.

 

[Radium]

No, the leakage current creates a power loss in the transistors that
are needed for operation. Some of these transistors must have a
voltage drop on them or your circuit can't change any logic levels.
Any transistor with a drop on it will leak and thus generate heat.
Since your idea increases the number of transistors needed, the heat
will be a huge problem.

Okay. Thanks for clearing this up.

 

[Donald]

Okay. Thanks for clearing this up.

So I guess this means you won't patent this idea ??

Shame, with the state of the patent office, I am sure you would get it.

donald

 

[Bob Myers]

No offense but I would appreciate serious responses to my questions.
Please leave out the dark humor.

You have yet to ask a serious question.

Bob M.

 

[Radium]

You have yet to ask a serious question.

How does SB16 ISA's FM synth freshly generate its instructions?

I want my CPU to freshly generate instructions in a similar manner
instead of playing them back from ROM.

 

[The little lost angel]

The RAM chips store what would normally be stored in the magnetic
platters of an HDD.

In other words, you're describing a PC without HDD, aka a PC with just
solid state storage. Which others have mentioned, already exists with
systems running off solid state storage using existing technology such
as gigabyte sizes CF storage. They even have RAID controllers that run
off such RAM.

As I said before, parallel-Hz allows massively-serial devices to
operate at extremely high clock rates without needing any cooling
equipment.

It doesn't. As I understand it, heat is generated by transistors
mainly during switching, i.e. when it's doing something.

Simplistically, a 4Ghz CPU generates 4 giga unit of heat when
operating at 4Ghz. A 1Hz CPU of similar output efficiency generates
1unit of heat. But you need 4 billion 1Hz CPU, that translate to a
total of 4 giga unit of heat as well. There's no heat advantage to
your system.

Furthermore, pins, packaging and wiring takes up space. Your 4 Billion
1Hz CPU will be extremely space and wiring inefficient compared to a
single 4Ghz cpu.

The latency to first instruction out is also magnitudes times faster
than your 'parallel Hz' system. A modern 4Ghz CPU can always go full
speed for 1us to produce results and shutdowns/downclock to conserve
power. But your 1Hz system cannot be faster than 1 sec. So a 4Ghz CPU
can have the same heat efficiency AND faster output. Your 4Ghz
Parallel-Hz system is not comparable.

 

[Nick Maclaren]

|>
|> >As I said before, parallel-Hz allows massively-serial devices to
|> >operate at extremely high clock rates without needing any cooling
|> >equipment.
|>
|> It doesn't. As I understand it, heat is generated by transistors
|> mainly during switching, i.e. when it's doing something.

That was once largely true, but isn't any longer. Look up "passive
leakage" - it is now comparable to active leakage.


Regards,
Nick Maclaren.

 

[Don Bowey]

How does SB16 ISA's FM synth freshly generate its instructions?

I want my CPU to freshly generate instructions in a similar manner
instead of playing them back from ROM.

You have yet to tell us how you plan to get your system to operate at a rate
higher than one Hz.

An idea (or daydream) is not a design. All of your questions are a waste of
time until you provide some serious answers.

So far, you continue to show yourself as a troll.

 

[MooseFET]

How does SB16 ISA's FM synth freshly generate its instructions?

It doesn't. It does as it is told. It has built in hardware the
performs certain actions when told to. Even if it is claimed to
"generate instructions" it must be doing so based on some higher level
of instructions it is given or contains.

Consider an interpretive language such as BASIC. You can run a BASIC
program on your PC. At no time did your PC contain the instructions
for the specific program you ran and yet it did its job. The same can
be said to be true of things like sound and video cards. They are
given some collection of bits as input and make sound and light in
responce to them. The bits they are passed are not a full explanation
of the details of how to do those actions just the command that they
be done. Internal hardware inside the card fills in the details.

I want my CPU to freshly generate instructions in a similar manner
instead of playing them back from ROM.

Well I want to rich and good looking. Lets see whos wish comes true.

 

[MooseFET]

On May 6, 3:37 am, [email protected] (The little
lost angel) wrote:
[...]

It doesn't. As I understand it, heat is generated by transistors
mainly during switching, i.e. when it's doing something.

This is true at 4GHz. If you run as slow as 1Hz, the "leakage" sets
the lower limit on the power.

See the bit on "power estimation" in:
http://direct.xilinx.com/bvdocs/publications/ds055.pdf

If you put zero in as the frequency, you will see some power remains.
This part has enough hardware in it to make an extremely dumb
processor.

[....]

The latency to first instruction out is also magnitudes times faster
than your 'parallel Hz' system. A modern 4Ghz CPU can always go full
speed for 1us to produce results and shutdowns/downclock to conserve
power. But your 1Hz system cannot be faster than 1 sec. So a 4Ghz CPU
can have the same heat efficiency AND faster output. Your 4Ghz
Parallel-Hz system is not comparable.

Actually his idea may be worse than that. If an instruction needs the
value produced by a previous one, it can't be started until the other
is done. Consider the statement:

IF (A + B) * C > D THEN E = 1

(A + B) takes a cyle
* C takes a cycle

D takes a cycle

E = 1 takes a cycle

Thats 4 seconds to get the answer.

 

[Quadibloc]

Those mentifex devices are massively-parallel. As I said, my "parallel
Hz" design is intended for applications that are serial.

..
In that case, your design won't help.

It's true that somewhere, a device starts an instruction, and then one
nanosecond later, another device can finish an instruction. But they
won't be the _same_ instruction.

So, if something depends on the answer of an instruction that finishes
at a certain time, it needs to get that answer when _that_ instruction
finishes, not another one.

Having a time offset between processors in a massively parallel system
- which is all that your system remains, despite the offset - just
means that they're not synchronized, so if you need to take the result
from one instruction finishing, and give it to _many_ of the other
processors, some time may be wasted for some of them to become
available.

John Savard

 

[kony]

How does SB16 ISA's FM synth freshly generate its instructions?

Because there's a CPU and code telling it what to do, which
DOESN'T freshly generate.

I want my CPU to freshly generate instructions in a similar manner
instead of playing them back from ROM.

OK, then stop using your present system and do that.

 

[Mark]

You have yet to tell us how you plan to get your system to operate at a rate
higher than one Hz.

An idea (or daydream) is not a design. All of your questions are a waste of
time until you provide some serious answers.

So far, you continue to show yourself as a troll.

Not a troll, just stupid. A troll is someone who actually knows what
they are saying is incorrect. A stupid person simply isn't smart enough
to understand that what they are saying is incorrect.

 

[Guy Macon]

Not a troll, just stupid. A troll is someone who actually knows what
they are saying is incorrect. A stupid person simply isn't smart enough
to understand that what they are saying is incorrect.

From the Jargon file:

-----------------------------------------------------------------------

Troll

1. v.,n. [From the Usenet group alt.folklore.urban] To utter a posting
on Usenet designed to attract predictable responses or flames; or, the
post itself. Derives from the phrase "trolling for newbies" which in
turn comes from mainstream "trolling", a style of fishing in which one
trails bait through a likely spot hoping for a bite.
The well-constructed troll is a post that induces lots of newbies and
flamers to make themselves look even more clueless than they already do,
while subtly conveying to the more savvy and experienced that it is in
fact a deliberate troll. If you don't fall for the joke, you get to be
in on it. See also YHBT.

2. n. An individual who chronically trolls in sense 1; regularly posts
specious arguments, flames or personal attacks to a newsgroup,
discussion list, or in email for no other purpose than to annoy someone
or disrupt a discussion. Trolls are recognizable by the fact that they
have no real interest in learning about the topic at hand - they simply
want to utter flame bait. Like the ugly creatures they are named after,
they exhibit no redeeming characteristics, and as such, they are
recognized as a lower form of life on the net, as in, "Oh, ignore him,
he's just a troll."

Some people claim that the troll (sense 1) is properly a narrower
category than flame bait, that a troll is categorized by containing
some assertion that is wrong but not overtly controversial.

The use of `troll' in either sense is a live metaphor that readily
produces elaborations and combining forms. For example, one not
infrequently sees the warning "Do not feed the troll" as part of
a followup to troll postings.

See also Kook.

-----------------------------------------------------------------------

Kook

[Usenet; originally and more formally, `net.kook'] Term used to
describe a regular poster who continually posts messages with
no apparent grounding in reality. Different from a troll, which
implies a sort of sly wink on the part of a poster who knows
better, kooks really believe what they write, to the extent that
they believe anything.

The kook trademark is paranoia and grandiosity. Kooks will often
build up elaborate imaginary support structures, fake corporations
and the like, and continue to act as if those things are real even
after their falsity has been documented in public.

While they may appear harmless, and are usually filtered out by
the other regular participants in a newsgroup of mailing list,
they can still cause problems because the necessity for these
measures is not immediately apparent to newcomers; there are
several instances on record, for example, of journalists writing
stories with quotes from kooks who caught them unaware.

An entertaining web page chronicaling the activities of many
notable kooks can be found at http://www.crank.net/index.html


-----------------------------------------------------------------------

 

[Robert Redelmeier]

http://www.ltsp.org/
Linux terminal server project. Seems like a step backwards but
certainly solves many of the usual support problems. Etherboot
or Intel PXE for the terminals is a possibility if the hardware
supports it or could boot off a usb disk or flash card.

Thanks. Very nice. There are some decent thinclients available.
Mostly looks like for internet cafes, classrooms, workrooms.

A bigger market might be "standalone" thin clients. Home internet
appliances with built-in browsers, but otherwise no programmability
(state retained). Certainly more than the 32 MB LTSP minimum, but
probably not more than 128 MB RAM. US$50 plus monitor (or S-video
out for exhibitionists/remotecontrol-hogs who want to surf on the TV!)

Such a device would be attractive to non-computer experts
(nothing to go wrong) or as second PCs in a household.

-- Robert

 

[kony]

Not a troll, just stupid. A troll is someone who actually knows what
they are saying is incorrect. A stupid person simply isn't smart enough
to understand that what they are saying is incorrect.

They become a troll after it is pointed out over and over,
which also happened last year when Radium posted same thing.

 

[krw]

On May 6, 3:37 am, [email protected] (The little
lost angel) wrote:
[...]

It doesn't. As I understand it, heat is generated by transistors
mainly during switching, i.e. when it's doing something.

This is true at 4GHz. If you run as slow as 1Hz, the "leakage" sets
the lower limit on the power.

With newer processes the lower limit may approach 50% of the total
power.

See the bit on "power estimation" in:
http://direct.xilinx.com/bvdocs/publications/ds055.pdf

If you put zero in as the frequency, you will see some power remains.
This part has enough hardware in it to make an extremely dumb
processor.

No dumber than the one proposing it.

Actually his idea may be worse than that. If an instruction needs the
value produced by a previous one, it can't be started until the other
is done. Consider the statement:

IF (A + B) * C > D THEN E = 1

(A + B) takes a cyle
* C takes a cycle
E = 1 takes a cycle

Not necessarily. The "E = 1" step can be done in zero cycles. It's
just a mux of '1' and whatever was there. There are several ways of
handling this in zero cycles. Fer instance one could do it by
assigning one register ('1') or another (the previous 'E') during
rename.

Thats 4 seconds to get the answer.

That't "That's" (before the apostrophe police get here). ;-)

 

[The little lost angel]

Actually his idea may be worse than that. If an instruction needs the
value produced by a previous one, it can't be started until the other
is done. Consider the statement:

IF (A + B) * C > D THEN E = 1

(A + B) takes a cyle
* C takes a cycle
E = 1 takes a cycle

Thats 4 seconds to get the answer.

I mentioned this earlier to him albeit without detailed example. But
Radium seems to conveniently overlooked it so I figured he's probably
not going to answer even if I repeated it a second time. Or maybe he
didn't understand the implications of my statement but let's see if he
does with your detailed example pPp

 

[Wilco Dijkstra]

Multiplies usually take more than one cycle.

The "THEN" (a conditional branch) also takes a cycle (unless you do
E = 1 as a conditional move).

Not necessarily. The "E = 1" step can be done in zero cycles. It's
just a mux of '1' and whatever was there. There are several ways of
handling this in zero cycles. Fer instance one could do it by
assigning one register ('1') or another (the previous 'E') during
rename.

So E = 1; F = 2; G = 3; H = 4; I = 5; also takes zero cycles in total???

Moves don't take zero cycles. Although they are pretty simple, they
need resources such as a constant decoder and a register port to
write the result to. So moves take 1 cycle just like any other basic
ALU instructions.

It's 5 with the branch (1 cycle if not-taken), and if you assume all variables were
allocated to registers and a multiply take just one cycle. But remember that the
OP was proposing a simple bit-serial CPU, so if registers are 32 bits wide, and
the multiply is done bit-serially too, it would take 1152 seconds to complete!

Wilco