Re: GB vs. MB

Jason Burrell (
Wed, 27 Nov 1996 11:16:13 -0600 (CST)

On Wed, 27 Nov 1996, Albert Cahalan wrote:

> > I think that linux should report what's true, not what the
> > marketers shove down our throats. A megabyte is 2^20, it
> > should be reported as such.
> People expect a disk MB to be 10^6 now, so maybe Linux should
> do what people expect. With 2^20, it looks like Linux makes
> disks get smaller.

Please, folks, don't do this.

2^20 is the classical definition for the megabyte. Drive manufactuers
decided they would advertise falsely by redefining the megabyte to be
10^6, and then throw in a factor of 2 for compression. There's simply no
other word for "decimal megabytes" than false advertising in the hope of
making a product look better. But this we all know.

As someone that worked in a place where he was told he was calculating the
number of megabytes on a drive wrong (they were thinking in 10^6), I argue
vehemently against having Linux redefine the megabyte to be 10^6. 10^6
makes absolutely no sense at all. It isn't a power of two, isn't evenly
divisible by number of bits or number of K, and throws us "old timers"
off. Might as well say "AOL is the Internet" because AOL says it is.

I suggest we leave 10^6 in the realm of lusers and drive makers.

> I think we should just be glad that hard disk manufacturers
> don't throw in a factor of 4 for compression -- not yet anyway.

No kidding. The factor of 2 for compression is ludicrous enough -- if I'm
using a non-compressed filesystem (read: Not throwing in more processor
overhead to make my data unreadable by other OSes), the value is
completely shot. Tape drives are even sold this way these days.

"The POP3 server service depends on the SMTP server service, which 
failed to start because of the following error:
The operation completed successfully." -Windows NT Server v3.51