Re: POLL: Would you use Ada more if...

[ada@polarhome.com (Anatoly Chernyshev)]

Dmitry A.Kazakov <mailbox@dmitry-kazakov.de> wrote in message news:<akflgg$1hdrbh$1@ID-77047.news.dfncis.de>...
> Robert A Duff wrote:
> 
> > That reminds me of a talk I heard some years ago, given by a guy from
> > NASA.  He was talking about the software for the Apollo program.  They
> > had weight budgets for all the pieces of the rocket -- obviously you
> > want to make the thing as light as possible if you're launching it into
> > space.  So the manager asked the software guy how much the software
> > would weigh.  He said software doesn't weigh anything.  The manager said
> > come on, these weight budgets are very important, even if it's not much,
> > every ounce counts.  The software guy tried in vain to convince the
> > manager that the weight of software really is zero.  Eventually the
> > software guy showed the manager a deck of punch cards, and said, See the
> > holes?  That's the software.  Then the manager understood.
> 
> (:-)) 
> 
> [OT] I have almost forgot physics, but I suppose that the weights of 
> RAM/ROM bits set to 0 and 1 should slightly differ. So same CPU may weigh a 
> pair picogramms (?) more or less depending on the software version.

From the Brillouin's theorem:					
Increase in information is equal to the decrease in entropy:		

	dI = -dS							

As soon as dS is in J/(mole*K), the intrinsic energetic part of S
depends on the fraction of the matter by which 1 bit is coded, and, of
course, on the temperature too.
If the bit is coded by, let's say, the spin state of single electron
then mole fraction is equal to 1/NA (Avogadro's number = 6.02e23) =
1.66E-24	moles
Another example, when the bit is coded by an abacus stone maden of
quartz, then molar fraction is going to be mass of the stone (take it
for 5g)/molar mass of quartz (60) = 8.00E-02	moles

The difference in energy for 0 and 1 states for these devices at	
298 K (1*mole*T) will be respectively 4.95E-22 J and 2.38E+01 J.				
From E=mc**2 we now obtain the mass difference between 0 and 1 bits
for these cases: 5.50E-39 kg and 2.65E-16 kg.
If the length of program is 1Mb which is 8388608 bits (we do not
consider the domination of 1 states over 0 states because they do not
complement each other, since one has to always spent some energy to
keep both states ordered) then the mass difference between bulk
support (CD, abacus, etc) will be 4.61E-32	kg in the first case or
2.22E-09 kg in the second case (if one was dare enough to code 1Mb
program in abacus).

This is just an estimation made for fun, so I may be totally wrong.

Regards,
Anatoly