Advantages of Parallel Hz

M

MooseFET

Hi:

Below is an example of "parallel Hz"

http://img56.imageshack.us/img56/2427/clocksignalexample8is.gif

If each clock signal is 1 Hz, and you have a billion of them,
staggered such that every 1ns part of the CPU can start, and finish,
an instruction - making the effective 'clock rate' 1 GHz.

With a billion CPUs, the leakage current would kill you. If you want
real processing speed at low power, you should look at using 3 phase
clocks. There are several advantages to this. You only have to swap
two lines of the 3 phase clock to invert the order. This means that
the processor can back step. It doesn't really make a general purpose
computer but it would be very handy if you were playing jeopardy.
 
D

David L. Jones

Maybe you should google for "pipelining"

Okay. According to my research [on google] pipelining doesn't have
much to do with "parallel Hz".

In addition, pipelining uses buffers and has significant latency. Not
something I am found of.

My dream PC does not have any buffers or latency.

My dream PC uses RAM chips -- instead of magnetic discs -- in to store
information. It is entirely chip-based.

This PC is built in such a way that it freshly generates the correct
electric signals ["on the fly"] instead of playing them back from its
ROM chips.

There are sets of instructions stored in ROMs. In the case of most PC,
these instructions load before the CPU "knows" it has a hard drive or
other peripheral devices. However, in my dream PC, those instructions
be generated in real-time instead of storing them.

I am aware that EEPROM is reliable, low power, customizable, reprogram-
able, cheap and proven. But just out of personal preference, my dream
PC is hard-wired in such a way that it does not need any ROM.

Other specs are below. The stuff below also do not need any ROM memory
because they are physically-built to generate the signals which
correspond to the following.

OS: Windows 98SE
Browser: Mozilla Suite 1.8b

No fans, no discs, no moving parts, no ROM [except for the CD/DVD
recording/playing and re-writing].

IOW, my dream PC would work perfectly but would not need any moving
parts, discs, or fans. The "HDD" would consist of silicon RAM chips in
place of disc-platters and electric parts in place of magnetic parts.
No moving parts, no noise, no fans, no magnets, no hazardous heat.

You can buy and/or build a PC like that right now.
Solid state IDE hard drives are available, and there are plenty of low
power industrial embedded PC's around that don't need fans and can run
on couple of watts. You could run 98SE on almost any low power
embedded hardware platform.
If you don't want the IDE drive you could even use DiskOnChip.

Dave.
 
R

Radium

You can buy and/or build a PC like that right now.
Solid state IDE hard drives are available, and there are plenty of low
power industrial embedded PC's around that don't need fans and can run
on couple of watts. You could run 98SE on almost any low power
embedded hardware platform.
If you don't want the IDE drive you could even use DiskOnChip.

Could I plug in my SB16 ISA card into this PC? I really like my old
SB16 card because of its FM synth.
 
R

Radium

I hear that if you have nine women working in parallel, you can get a
baby in one month, too.

You are confusing *clock-rate* and *processor speed*. Although
related, they are two different things.
 
D

David L. Jones

Could I plug in my SB16 ISA card into this PC? I really like my old
SB16 card because of its FM synth.

If it has an ISA bus slot, then yes.
PC-104 format is popular with low power embedded PC's, but ISA and
other connector and form factors are available.
You can get PC-104 format SB compatible sound cards.

You can plug a solid state IDE hard drive into ANY PC. If that PC is
low enough power not to need a fan, then you have your "dream" machine
with no moving parts or noise.

Dave.
 
M

MooseFET

The issue is not processor speed, but clock rate.


Noboby gives a darn about clock rate. It is work done per Watt-second
that is the real issue. The 3 phase clock is the lowest number that
has an unambiguous order and thus is the lowest power one with the
property of a reversible order. With a single clock or a two phase
one, the forwards order is always assumed. This greatly limits what
the processor is able to do.

Consider a simple case like this:

You have a list of values that you need to fit a curve to. You know
that a program like this:

for I=1 to 10
print F(I)
next I

would print:

1 17 33 105 117 119 67 52 37 23

Given this you can burn a lot of CPU time figuring out exactly what
F(I) is. With a 3 phase clock, you only need the list, the program to
call the F(I) and a copy of the compiler. You simply invert the C and
B clock lines and then look at source code for F(I) to find out what
the function is. A great deal of energy is saved by doing this.
 
T

The little lost angel


erm, didn't you go through this last year too?
The benefit of using a billion 1 Hz clock signals to make a clock rate
of 1Ghz is that such a system would not get as hot as system running
one 1 GHz clock signal . While the overall amount of heat generated by
both systems maybe around the same, the system running a billion 1 Hz
clock signals will have less heat per area than the system running one
1 billion Hz clock signal. Hence, the former system is far less
vulnerable to thermal damage than the latter.

Erm, wouldn't this simply be achievable by making a physically bigger
1Ghz chip? A single fast chip is also much more likely to perform
better for the same amount of heat since most general computing
problems are not readily parallelized, at least as far as I understand
it.
Let's say two CPUs of different frequencies have been running at the
same voltages and amperages and for the same amount of time. The CPU
with a higher-frequency will be hotter than the CPU with a lower-
frequency.

erm, DUH. Quite obviously something like a P4 running at 3Ghz will be
hotter than a P4 running at 2Ghz if both used the same voltage. The P4
3Ghz will also do more work than the P4 2Ghz, so what's your point
here?
This design would go great for any application that cannot be
efficiently parallelized [in terms of bits]. Examples of such are
arithmetics and Boolean logic.

How so? Perhas you can illustrate with an example of 1 single 40x
speed chip versus a 10 1x speed "parallel hz" chip?

As far as I can see it, if my next instruction was waiting for the
result of the preceding boolean logic result, my single 40x chip will
get me to the next result 10x faster than your 4x //Hz chip.

Since you mention the application cannot be efficiently parallelized,
I take it to mean the instructions are highly dependable on each
other's results. Thus the faster each instruction gets finished, the
faster the next can go.

Not much point having 3,999,999 //Hz units waiting 1sec for the result
of one instruction compared to a 4Ghz chip that could spit out the
next 1million instructions within the same time.

But of course I hardly qualify as a chip architect so feel free to
point out where I'm mistaken :p
 
T

The little lost angel

This PC is built in such a way that it freshly generates the correct
electric signals ["on the fly"] instead of playing them back from its
ROM chips.

And how does it know how to generate the correct signals?
There are sets of instructions stored in ROMs. In the case of most PC,
these instructions load before the CPU "knows" it has a hard drive or
other peripheral devices. However, in my dream PC, those instructions
be generated in real-time instead of storing them.

Isn't it more efficient to calculate static results beforehand and
store them for use, than to waste time generating the same
instructions all the time?

Other specs are below. The stuff below also do not need any ROM memory
because they are physically-built to generate the signals which
correspond to the following.

OS: Windows 98SE
Browser: Mozilla Suite 1.8b

In other words, your system is not upgradable and has to live with
whatever bugs there are for the entire "useful" life of the system
since everything's hardwired?
To put it simply, what I am describing is a PC that does not need to
store any information because all of the signal codings for the info
is generated in real-time.

A PC that does not store any information... what good is it for?
Most importantly, though, my dream PC uses parallel-Hz and is
massively-serial!!

And so ?
 
R

Robert Redelmeier

In comp.sys.ibm.pc.hardware.chips The little lost angel said:
A PC that does not store any information... what good is it for?

Not that I wish to support the OP in any way, but a stateless
PC is a terminal, now AKA "Thin Client" or "Internet Appliance".
Information is stored on a server elsewhere, accessible everywhere.

I like Xterminals, but I'm a dinosaur and remember the Real
Thing: terminals that did X and you could log into the networked
machine. Sort of like a VT220 doing graphics. They ran BOOTP
(iso DHCP) and TFTP to boot.

Now you'd want boot from flash and DHCP. The minicomp would be
a small box like a SohO router with SVGA out (only 2D required),
10/100baseT or wireless, a wall-wart for power, and USB or PS/2
for kbd/mse. Very tidy, very neat and very cheap. Add monitor,
kbd, mouse and network to run.

Hardware specs very similar to a SoHo router:
486-class CPU, 64 MB RAM, 64 MB FLASH. Tight Linux SW.

Onboard SSH would be a must, but a key design decision would be
whether to incorporate a browser client to the local X server. Doing
so would usually improve performance and always cut X-traffic. But
this would jeopardize making the box a stateless appliance. That
might depend on whether the box was tethered to a LAN server, or
expected to work standalone.

With software, you can do much the same thing to much more powerful
desktop and laptop machines using something like a Knoppix boot CD.

-- Robert
 
R

Radium

This PC is built in such a way that it freshly generates the correct
electric signals ["on the fly"] instead of playing them back from its
ROM chips.

And how does it know how to generate the correct signals?

It's built that way.

How does SB16's FM synth "know" how to generate its FM signals? Much
in the same way.
Isn't it more efficient to calculate static results beforehand and
store them for use, than to waste time generating the same
instructions all the time?

Yes. However, due to my personal opinions, I prefer to use the least
amount of non-RAM storage/memory necessary.
In other words, your system is not upgradable and has to live with
whatever bugs there are for the entire "useful" life of the system
since everything's hardwired?

Well, the system could be made without bugs. Right?

Moreover, upgrades are possible and stored in the RAM chips I
described above.

Now, if those chips had their info erased [like formatting an HDD],
then the upgrades would have to be performed again -- which is an easy
task.
A PC that does not store any information... what good is it for?

It does store info in the RAM chips.

My point is, parallel-Hz and "parallel bits" are not to be confused
with each other.
 
G

Gavin Scott

Not that I wish to support the OP in any way, but a stateless
PC is a terminal, now AKA "Thin Client" or "Internet Appliance".
Information is stored on a server elsewhere, accessible everywhere.
With software, you can do much the same thing to much more powerful
desktop and laptop machines using something like a Knoppix boot CD.

And such an "Internet Terminal" system can be ideal for one's parents
or kids, or anyone else who wants a fool-proof easy-to-use computer
setup.

These days not only is the web almost the only application on the
Internet, but it's almost the only computing application. Give
someone a slightly older PC with no hard drive and a Knoppix or
Ubunto live Linux CD and they may be able to accomplish everything
they want to do through a web browser with no local persistent
store of any kind.

Something like Google's web application suite (gmail etc.) would
even let you do normal office application functions as well.

User gets confused? Makes a mistake? Gets a trojan/virus? Just
power cycle the box and the problem is gone. If it fails to boot
to the user's familliar starting state then it's a hardware problem.

"Push the reset button and try again" may be all the tech support
the user ever needs (though this training may turn out to be ill-
advised if you ever let them near a "real" computer :)

G.
 
R

Roger_Nickel

And such an "Internet Terminal" system can be ideal for one's parents
or kids, or anyone else who wants a fool-proof easy-to-use computer
setup.

These days not only is the web almost the only application on the
Internet, but it's almost the only computing application. Give
someone a slightly older PC with no hard drive and a Knoppix or
Ubunto live Linux CD and they may be able to accomplish everything
they want to do through a web browser with no local persistent
store of any kind.

Something like Google's web application suite (gmail etc.) would
even let you do normal office application functions as well.

User gets confused? Makes a mistake? Gets a trojan/virus? Just
power cycle the box and the problem is gone. If it fails to boot
to the user's familliar starting state then it's a hardware problem.

"Push the reset button and try again" may be all the tech support
the user ever needs (though this training may turn out to be ill-
advised if you ever let them near a "real" computer :)

G.

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.
 
R

Radium

Note: I already posted my response to the message below. However,
Google -- being the sick@$$ piece of f--k they are -- did not display
it.

Erm, wouldn't this simply be achievable by making a physically bigger
1Ghz chip? A single fast chip is also much more likely to perform
better for the same amount of heat since most general computing
problems are not readily parallelized, at least as far as I understand
it.

Whether the computation is parallelized or not is irreverent.
erm, DUH. Quite obviously something like a P4 running at 3Ghz will be
hotter than a P4 running at 2Ghz if both used the same voltage. The P4
3Ghz will also do more work than the P4 2Ghz, so what's your point
here?

My parallel-Hz is an efficient means of keeping high-frequency serial
processors cool without needing any cooling equipment.
This design would go great for any application that cannot be
efficiently parallelized [in terms of bits]. Examples of such are
arithmetics and Boolean logic.
How so? Perhas you can illustrate with an example of 1 single 40x
speed chip versus a 10 1x speed "parallel hz" chip?

As far as I can see it, if my next instruction was waiting for the
result of the preceding boolean logic result, my single 40x chip will
get me to the next result 10x faster than your 4x //Hz chip.

Your 40x chip maybe faster but it runs a significant risk of
overheating unless cooled by a fan.
Since you mention the application cannot be efficiently parallelized,
I take it to mean the instructions are highly dependable on each
other's results. Thus the faster each instruction gets finished, the
faster the next can go.

Sometimes yes. Sometimes no. In any case, the purpose of parallel-Hz
is to allow serial processors to operate at super-high frequencies
without needing any cooling equipment. Parallel processors [using 1
bit per line] don't need to use parallel-Hz because they don't get as
hot at the same frequencies that would fry a serial processor. This is
partly due to the fact that parallel processors don't have as many bit
transitions as serial processors.
Not much point having 3,999,999 //Hz units waiting 1sec for the result
of one instruction compared to a 4Ghz chip that could spit out the
next 1million instructions within the same time.

Well, if you want a massively-serial PC running at super-high clock
rates and at the lowest gear [1 bit per cycle] without fans, then not
using parallel-Hz can mean disaster for that PC as the high-frequency
signals vaporize the circuits.

Once again, I will point out that there is a significant different
between parallel-bits and "parallel Hz".

A parallel processor uses parallel-bits. A serial processor doesn't.
"Parallel Hz", OTOH, can be used on both serial as well as parallel
processors. However, parallel-Hz doesn't provide any benefit for
parallel processors. For serial processors, parallel-Hz prevents
dangerous increases in temperature that result from high-frequency
signals.
 
D

Donald

Radium said:
My parallel-Hz is an efficient means of keeping high-frequency serial
processors cool without needing any cooling equipment.

I am sorry, but I missed the beginning of this thread and Its no longer
on my server.

But, If this is your "parallel-Hz" design, why not just patent it and
have the market decide if its any good ??

It seems you are either preaching to the choir or it doesn't really work.

I am not trying to shoot down your ideas, I just don't see anyone doing
a real design with this idea.

donald
 
K

kony

I am not trying to shoot down your ideas, I just don't see anyone doing
a real design with this idea.

donald

You have realized the crux of the problem. Radium ignores
purpose, that any thought would need a useful gain to be
more than a wasteful folly.

Radium could have followed through with some research and
pitching these ideas to those who have an ability to
seriously contemplate, and possibly implement the ideas, but
instead is only wasting everyone's time with a random
thought then proclaiming "why don't we do this" as if
rejecting modern technology, finding it a problem that needs
resolved, but wouldn't be when technology allowed, was a
reasonable use of time.

Radium lacks a method to achieve any goals, just random
thoughts which more than anything, demonstrate an inability
to use contemporary technology as well as everyone else.
Anyone could claim "oh but what if this or that was better
than it is", but those who actually drive innovation and
progress, do so leveraging the tech that IS available to do
so instead of only finding fault in it.

The funny part is when basic concepts ARE already possible
today, anyone with a thick enough wallet can build a system
without fans or mechanical hard drives. Radiun pitches
this as if it is some new thought but without the key
details of implementation others already follow, if/when
they find it a reasonable alternative, which isn't very
often because when all is said and done, mechanical failures
are managable and can be planned for, and beyond hard
drives, can be, through competent system design, assumed to
be outside the viable lifespan of the system.

Unfortunately, everyone and their brother fancies themselves
to be competent system integrators, but when it comes down
to finer details, suddenly cost often matters more than
lifespan, then only LATER does one claim "it's a problem".
The problem was usually following time-tested strategies
instead of seeing a fault and making random theories instead
of comparing how the system(s) with faults deviated from
those that didn't have same faults.

It is a wilderness out there, plenty of shady businesses
selling parts in the computer biz that are not suitable for
longer term use. Experience and discrimination, not random
rejection of contemorary tech, allows most people to find a
happy medium, but Radium has yet to find that medium and
instead of accepting that Radium has more to learn, Radium
instead drifts off on tangents which are essentially excuses
for why Radium can't manage to use contemporary tech.
 
J

jasen

Given this you can burn a lot of CPU time figuring out exactly what
F(I) is. With a 3 phase clock, you only need the list, the program to
call the F(I) and a copy of the compiler. You simply invert the C and
B clock lines and then look at source code for F(I) to find out what
the function is. A great deal of energy is saved by doing this.

LOL!

maybe that'd be possible ifthe computer only did reversible operations,
but a computer like that is either useless or trivial.

What you propose is like trying to run a printer in reverse to recycle
paper.

Bye.
Jasen
 
I

invalid

Radium said:
This PC is built in such a way that it freshly generates the correct
electric signals ["on the fly"] instead of playing them back from its
ROM chips.

There are sets of instructions stored in ROMs. In the case of most PC,
these instructions load before the CPU "knows" it has a hard drive or
other peripheral devices. However, in my dream PC, those instructions
be generated in real-time instead of storing them.

To put it simply, what I am describing is a PC that does not need to
store any information because all of the signal codings for the info
is generated in real-time.

Generated from what? Pixie dust?

Please explain *in detail* how the first three such instructions
are generated.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top