[G4] Broadband Speed Liar's Poker

Joseph B. Gurman gurman at gsfc.nasa.gov
Sun Dec 14 07:47:53 PST 2003


     Doug McNutt wrote:

>  You aren't the only one.  We need a Truth in Speed law. Well, our 
>US government will never get it right anyway, but we do need some 
>conformity to dimensioning standards.

     Funny comment, considering you cite a NIST site below....

     NIST, of course, like all top-down standards-making bodies, has 
decided in vacuo that kilo, mega, giga, &c. refers\ only to a factor 
of a thousand, and some completely artificial terms need to be made 
up for powers of 1024.

    When a K or k immediately precedes a b or B, we know it means 
either 1000 or 1024 times a bit or byte, and we also know the 
difference is negligible. One might argue that by the time we get to 
300 Gbyte hard drives, the difference can be significant, but it's 
always much less than the difference between claimed and formatted 
capacities.

>
>First of all K doesn't mean kilo or a factor of 1000. It is rather a 
>measure of temperature, degrees Kelvin. k is the abbreviation for 
>kilo meaning 1000 base 10. editors computer types have used K to 
>mean a factor of 1024 but that's approved by no one but some editors 
>and it's not clear how it applies to frequency anyway.
>
><http://physics.nist.gov/cgi-bin/cuu/Info/Units/>
><http://whatis.techtarget.com/definition/0,,sid9_gci825099,00.html>
><http://physics.nist.gov/cuu/Units/binary.html>
><http://www.unc.edu/~rowlett/units/prefixes.html>
>
>And then there's the question of just how many bits there are in a 
>byte.  It's usually 8 but if you're doing things like PPP or a 56 kb 
>modem, it might be 10 to account for a start and a stop bit.

     Well, no, bytes always have 8 bits (except in France, but that's 
an off-color joke, and why they're calls octets there). Your screed 
about baud, below, is correct. Once upon a time, some machines had 
word sizes of 7 or 9 or whatever bits, but they weren't bytes.

>
>And then there's the baud concept. Baud is the frequency in symbols 
>per second. The unknown is just how many bits are in a symbol.
>
>You're right. Liars are the rule. Look at the size of disks as 
>reported by the salesman (base 10) and by the the software 
>formatters (base 1024).

     Apple has always (1) assumed mega, giga, &c. are powers of 1024, 
and (2) stated disk usage stats in both Gbyte and bytes, complete 
with commas separating sets of three digits. Thus, my hard drive says 
it has 8.66 GB used (9,301,409,792 bytes).

>
>The chances of getting a straight answer are the same as the chances 
>that your senator will personally answer an e-mail message.

     From whom would you chose a straight answer? A federal agency 
charged with standardizing weights and measures, but overstepping 
when it decides how we will define shorthand for large numbers? The 
computer industry? Common usage?

     Returning to the original question, the only true measure of 
download speed is made with a large file and a stopwatch.

						Joe Gurman
-- 
"I love deadlines. I love the whooshing sound they make as they go by."
                                                             - Douglas 
Adams, 1952 - 2001

Joseph B. Gurman, NASA Goddard Space Flight Center, Solar Physics
Branch, Greenbelt MD 20771 USA



More information about the G4 mailing list