Conservation of Energy

analogue filter (low pass). Measure the energy in and the energy out
(power in and out). We should be able to ccalculate where all the
power dissipates in the circuit.

For a digital filter we have only numbers (as pointed out in another
thread). However, the laws of thermodynamics still apply from input to
output. For example if you have to do more number crunching you must
draw more power from the supply and the DSP gets hotter. Energy
balance must still apply surely.


Hardy

On Wed, 30 Sep 2009 17:50:47 -0700, HardySpicer wrote:

> analogue filter (low pass). Measure the energy in and the energy out
> (power in and out). We should be able to ccalculate where all the power
> dissipates in the circuit.
> 
> For a digital filter we have only numbers (as pointed out in another
> thread). However, the laws of thermodynamics still apply from input to
> output. For example if you have to do more number crunching you must
> draw more power from the supply and the DSP gets hotter. Energy balance
> must still apply surely.

But what's the point?  When we speak of the "energy" in a digital signal 
it's not the reality as is the case for a real filter.  It's a metaphor 
for the RMS intensity of the signal.  It'd a damn useful metaphor, but 
it's not the real thing.

-- 
www.wescottdesign.com

Tim Wescott <tim@seemywebsite.com> wrote:
 
> But what's the point?  When we speak of the "energy" in a digital signal 
> it's not the reality as is the case for a real filter.  It's a metaphor 
> for the RMS intensity of the signal.  It'd a damn useful metaphor, but 
> it's not the real thing.

I think I agree, but remember that in many systems energy is quantized.
Only specific values are allowed.  That isn't so different from
sampling, though as far as I understand quantum mechanics time
usually isn't quantized.  Then again, maybe we just don't understand
it up to that point yet.

-- glen

On Thu, 01 Oct 2009 05:52:13 +0000, glen herrmannsfeldt wrote:

> Tim Wescott <tim@seemywebsite.com> wrote:
>  
>> But what's the point?  When we speak of the "energy" in a digital
>> signal it's not the reality as is the case for a real filter.  It's a
>> metaphor for the RMS intensity of the signal.  It'd a damn useful
>> metaphor, but it's not the real thing.
> 
> I think I agree, but remember that in many systems energy is quantized.
> Only specific values are allowed.  That isn't so different from
> sampling, though as far as I understand quantum mechanics time usually
> isn't quantized.  Then again, maybe we just don't understand it up to
> that point yet.
> 
> -- glen

Quantized, yes.  Evenly quantized, no.

And when I store a "high energy" signal in an EPROM it doesn't get any 
hotter than if I store a "low energy" signal.

It just ain't real energy.

-- 
http://www.wescottdesign.com

>analogue filter (low pass). Measure the energy in and the energy out
>(power in and out). We should be able to ccalculate where all the
>power dissipates in the circuit.
>
>For a digital filter we have only numbers (as pointed out in another
>thread). However, the laws of thermodynamics still apply from input to
>output. For example if you have to do more number crunching you must
>draw more power from the supply and the DSP gets hotter. Energy
>balance must still apply surely.
>

The DSP also puts off heat when you crunch numbers to simulate an
oscillator or other unstable system.  Is energy being created?  I hope
you're not claiming heat is where the numbers "go"...

On Sep 30, 11:58&#4294967295;pm, "Michael Plante" <michael.pla...@gmail.com>
wrote:
> >analogue filter (low pass). Measure the energy in and the energy out
> >(power in and out). We should be able to ccalculate where all the
> >power dissipates in the circuit.
>
> >For a digital filter we have only numbers (as pointed out in another
> >thread). However, the laws of thermodynamics still apply from input to
> >output. For example if you have to do more number crunching you must
> >draw more power from the supply and the DSP gets hotter. Energy
> >balance must still apply surely.
>
> The DSP also puts off heat when you crunch numbers to simulate an
> oscillator or other unstable system. &#4294967295;Is energy being created? &#4294967295;I hope
> you're not claiming heat is where the numbers "go"...

Well the energy is being drawn from the supply. Most things end up as
heat in the end...
What goes in must come out! (somewhere!)

Hardy

>>analogue filter (low pass). Measure the energy in and the energy out
>>(power in and out). We should be able to ccalculate where all the
>>power dissipates in the circuit.
>>
>>For a digital filter we have only numbers (as pointed out in another
>>thread). However, the laws of thermodynamics still apply from input to
>>output. For example if you have to do more number crunching you must
>>draw more power from the supply and the DSP gets hotter. Energy
>>balance must still apply surely.
>>
>
>The DSP also puts off heat when you crunch numbers to simulate an
>oscillator or other unstable system.  Is energy being created?  I hope
>you're not claiming heat is where the numbers "go"...
>
>

I think Hardy's point is only that if you do more computations in DSP, it
will need more power. That's why the order of computation is so important
in DSP algorithms. But this is independent of the values of the numbers
themselves. For example, a convolution operation of two arrays will take
the same amount of energy irrespective of the values in those arrays
(assuming the numbers take the same amount of memory all the time).

Nikhil

Nikhil A D wrote:
>>> analogue filter (low pass). Measure the energy in and the energy out
>>> (power in and out). We should be able to ccalculate where all the
>>> power dissipates in the circuit.
>>>
>>> For a digital filter we have only numbers (as pointed out in another
>>> thread). However, the laws of thermodynamics still apply from input to
>>> output. For example if you have to do more number crunching you must
>>> draw more power from the supply and the DSP gets hotter. Energy
>>> balance must still apply surely.
>>>
>> The DSP also puts off heat when you crunch numbers to simulate an
>> oscillator or other unstable system.  Is energy being created?  I hope
>> you're not claiming heat is where the numbers "go"...
>>
>>
> 
> I think Hardy's point is only that if you do more computations in DSP, it
> will need more power. That's why the order of computation is so important
> in DSP algorithms. But this is independent of the values of the numbers
> themselves. For example, a convolution operation of two arrays will take
> the same amount of energy irrespective of the values in those arrays
> (assuming the numbers take the same amount of memory all the time).
> 
> Nikhil


News of a certain recording was recently posted to the sursound mailing 
list - a surround recording of a Shuttle launch:

http://www.digido.com/we-have-lift-off.html

Here the live sound source is described as "an 8.3 GWatt 
amplifier/loudspeaker with zero percent distortion and response down to 
DC". All of which has now been, um, reduced to a 24bit soundfile. So 
where is the "energy"?  It no longer exists; it has become 
~virtualized~. In principle you can take those samples and write all the 
values out on paper, using a pencil. The only energy involved is the 
calories consumed by the poor soul writing it out.  Forget about the 
existence of a DSP. This would have been a recording on magnetic tape 
not so many years ago. Data is just that - data. I could play that 
soundfile back over a tinpot hifi system, and the "energy" is merely 
that of my 10W amplifier. When dynamic data gets virtualized, all you 
can then say about it is that it has a bandwidth and a dynamic range - a 
smallest non-zero value and a maximum value. You can find the average 
and peak RMS power (which is relative to the maximum available in that 
data representation). You can plot the relative energy at various 
frequencies. The same data may represent a shuttle launch or the 
proverbial knat peeing in a millpond. The "energy" in a running dsp is 
not that of the data, but that of the power supply.

If you play back that shuttle recoding as recommended, and get up to a 
peak SPL near the 120dB mark, you may just about ~reproduce~ the 
perceived energy of the original. But you cannot say it "is" the energy 
of the original, all 8.3GWatts of it!

Richard Dobson

Nikhil A D wrote:
>>> analogue filter (low pass). Measure the energy in and the energy out
>>> (power in and out). We should be able to ccalculate where all the
>>> power dissipates in the circuit.
>>>
>>> For a digital filter we have only numbers (as pointed out in another
>>> thread). However, the laws of thermodynamics still apply from input to
>>> output. For example if you have to do more number crunching you must
>>> draw more power from the supply and the DSP gets hotter. Energy
>>> balance must still apply surely.
>>>
>> The DSP also puts off heat when you crunch numbers to simulate an
>> oscillator or other unstable system.  Is energy being created?  I hope
>> you're not claiming heat is where the numbers "go"...
>>
>>
> 
> I think Hardy's point is only that if you do more computations in DSP, it
> will need more power. That's why the order of computation is so important
> in DSP algorithms. But this is independent of the values of the numbers
> themselves. For example, a convolution operation of two arrays will take
> the same amount of energy irrespective of the values in those arrays
> (assuming the numbers take the same amount of memory all the time).
> 
> Nikhil


News of a certain recording was recently posted to the sursound mailing 
list - a surround recording of a Shuttle launch:

http://www.digido.com/we-have-lift-off.html

Here the live sound source is described as "an 8.3 GWatt 
amplifier/loudspeaker with zero percent distortion and response down to 
DC". All of which has now been, um, reduced to a 24bit soundfile. So 
where is the "energy"?  It no longer exists; it has become 
~virtualized~. In principle you can take those samples and write all the 
values out on paper, using a pencil. The only energy involved is the 
calories consumed by the poor soul writing it out.  Forget about the 
existence of a DSP. This would have been a recording on magnetic tape 
not so many years ago. Data is just that - data. I could play that 
soundfile back over a tinpot hifi system, and the "energy" is merely 
that of my 10W amplifier. When dynamic data gets virtualized, all you 
can then say about it is that it has a bandwidth and a dynamic range - a 
smallest non-zero value and a maximum value. You can find the average 
and peak RMS power (which is relative to the maximum available in that 
data representation). You can plot the relative energy at various 
frequencies. The same data may represent a shuttle launch or the 
proverbial knat peeing in a millpond. The "energy" in a running dsp is 
not that of the data, but that of the power supply.

If you play back that shuttle recoding as recommended, and get up to a 
peak SPL near the 120dB mark, you may just about ~reproduce~ the 
perceived energy of the original. But you cannot say it "is" the energy 
of the original, all 8.3GWatts of it!

Richard Dobson

On Oct 1, 6:28=A0am, Richard Dobson <richarddob...@blueyonder.co.uk>
wrote:
> Nikhil A D wrote:
> >>> analogue filter (low pass). Measure the energy in and the energy out
> >>> (power in and out). We should be able to ccalculate where all the
> >>> power dissipates in the circuit.
>
> >>> For a digital filter we have only numbers (as pointed out in another
> >>> thread). However, the laws of thermodynamics still apply from input t=
o
> >>> output. For example if you have to do more number crunching you must
> >>> draw more power from the supply and the DSP gets hotter. Energy
> >>> balance must still apply surely.
>
> >> The DSP also puts off heat when you crunch numbers to simulate an
> >> oscillator or other unstable system. =A0Is energy being created? =A0I =
hope
> >> you're not claiming heat is where the numbers "go"...
>
> > I think Hardy's point is only that if you do more computations in DSP, =
it
> > will need more power. That's why the order of computation is so importa=
nt
> > in DSP algorithms. But this is independent of the values of the numbers
> > themselves. For example, a convolution operation of two arrays will tak=
e
> > the same amount of energy irrespective of the values in those arrays
> > (assuming the numbers take the same amount of memory all the time).
>
> > Nikhil
>
> News of a certain recording was recently posted to the sursound mailing
> list - a surround recording of a Shuttle launch:
>
> http://www.digido.com/we-have-lift-off.html
>
> Here the live sound source is described as "an 8.3 GWatt
> amplifier/loudspeaker with zero percent distortion and response down to
> DC". All of which has now been, um, reduced to a 24bit soundfile. So
> where is the "energy"? =A0It no longer exists; it has become
> ~virtualized~. In principle you can take those samples and write all the
> values out on paper, using a pencil. The only energy involved is the
> calories consumed by the poor soul writing it out. =A0Forget about the
> existence of a DSP. This would have been a recording on magnetic tape
> not so many years ago. Data is just that - data. I could play that
> soundfile back over a tinpot hifi system, and the "energy" is merely
> that of my 10W amplifier. When dynamic data gets virtualized, all you
> can then say about it is that it has a bandwidth and a dynamic range - a
> smallest non-zero value and a maximum value. You can find the average
> and peak RMS power (which is relative to the maximum available in that
> data representation). You can plot the relative energy at various
> frequencies. The same data may represent a shuttle launch or the
> proverbial knat peeing in a millpond. The "energy" in a running dsp is
> not that of the data, but that of the power supply.
>
> If you play back that shuttle recoding as recommended, and get up to a
> peak SPL near the 120dB mark, you may just about ~reproduce~ the
> perceived energy of the original. But you cannot say it "is" the energy
> of the original, all 8.3GWatts of it!
>
> Richard Dobson- Hide quoted text -
>
> - Show quoted text -

information IS energy

http://en.wikipedia.org/wiki/Entropy_(information_theory)

Mark