There's a bit of a conceptual problem with trying to <i>measure</i> entropy. That's because entropy is based on probabilities, and probabilities are <i>subjective</i>. What are "the" odds that my brother will pass his SAT exam? My brother can make the most confident prediction (lowest entropy[1]). But the reader of this comment can do no better [2] than take the generic SAT success rate (higher entropy). And some peasant in Tuvalu might as well say 1/2 (maximum entropy).<p>So entropy is subjective. It's a measure of how much <i>new information</i> you expect to get from a message. The more you know already, the lower the entropy.<p>EDIT: An awesome quote from Jaynes.<p>"Probability assignments are not assertions of any physical property; they are a description of a <i>state of knowledge</i>. It is therefore illogical to speak of "verifying" the probabilities by performing experiments on the physical world. That would be like trying to verify a boy's love for his dog by performing experiments on the dog"<p>[1] Not necessarily true, see Wiki or David MacKay's wonderful book for details. <a href="http://www.inference.phy.cam.ac.uk/mackay/itila/" rel="nofollow">http://www.inference.phy.cam.ac.uk/mackay/itila/</a><p>[2] Also not strictly true! You might argue that his passing the SAT should correlate with my passing it, and estimate what my odds would be/were from my comment history.
I wondered how it would compare if you measure this from the gate level rather than the transistor level. To check this, I looked online for a Verilog file describing a 6502 core. This site has one:<p><a href="http://www.birdcomputer.ca/Projects/Prj6502/bc6502_page.html" rel="nofollow">http://www.birdcomputer.ca/Projects/Prj6502/bc6502_page.html</a><p>The zipped size of that file is 38 kilobytes, which is actually larger than the approximately 7 kilobytes guessed in the article for transistor level information. That is probably due to the verilog code being more verbose; you would think zipping it would make up the balance but apparently not. However, even though 38K >> 7K, I think the fact that the two are at least in the same ballpark is encouraging.