The binary system is based on 2s not tens.
1,2,4,8,16,32,64,128,256,512,1024,2048 etc are binary numbers not rounded
numbers.  Binary is the most efficient use in hardware design and logic.
Special hardware was developed for the base 10 system because base 10 is
what the general public uses.

People who deal in binary and bytes understand that the prefixes in 1 000s
or 1/1 000s know the prefixes do represent exact binary numbers.  The
standard prefixes are for ease of use.

The SI prefix definitions remain unchanged.  Mega still means millions etc.
regardless of the unit.  For example megapixels still means millions of
pixels.  Megasbits still means millions of bits etc.  In dealing in the
context of pure binary, the prefixes do not mean exact binary numbers.

Stan Doore

----- Original Message ----- 
From: "Bill Potts" <[EMAIL PROTECTED]>
To: "U.S. Metric Association" <[EMAIL PROTECTED]>
Sent: Sunday, December 28, 2003 4:48 PM
Subject: [USMA:28018] RE: Moral Issue?...


Marcus Berger wrote:
"For instance, I'd much rather see 10-bit, 100-bit buses than the current
16, 32, 64, etc...  Nothing, *technically* would make such construction
wrong or flawed IMHO.  It's just a pity that someone "decided" to call 8
bits a byte, as opposed to 10 being a bite."

We've been over this ground before, Marcus. A 10-bit bus wouldn't make a
computer any less binary.

The range of memory that would be addressable over a 10-bit bus would be
2^10. Each of the memory elements thus addressable could have any number of
bits. For consistency with your approach, each element might contain 10
bits. Again, the largest binary number that could be stored in that memory
element would be 2^10-1. The size of the largest decimal number would be
dependent on how one structured bit groups for expressing decimal digits. In
fact, for the storage of decimal numbers, a bit group containing a multiple
of 4 bits would work better. A 12-bit group would be good for decimal
numbers from 0 to 999 (10^3-1). However, used in binary fashion, it could
accommodate numbers from 0 to 4095 (2^12-1).

As a 4-bit group, used for decimal digits, would only use 10 of the 16
possible combinations, it would only be 62.5% efficient (as would any
multiple of a 4-bit group). Used for binary numbers, it's 100% efficient (as
is any number of bits).

Bill Potts, CMS
Roseville, CA
http://metric1.org [SI Navigator]


Reply via email to