Here are the SI fractional prefixes:

prefix decimal jargon usage

milli- 1000^-1 (seldom used in jargon)

micro- 1000^-2 small or human-scale (see *micro-**)
nano- 1000^-3 even smaller (see nano-)
pico- 1000^-4 even smaller yet (see pico-)
femto- 1000^-5 (not used in jargon—yet)
atto- 1000^-6 (not used in jargon—yet)
zepto- 1000^-7 (not used in jargon—yet)
yocto- 1000^-8 (not used in jargon—yet)
*

*The prefixes zetta-, yotta-, zepto-, and yocto- have been included in
these tables purely for completeness and giggle value; they were adopted in
1990 by the 19th Conference Generale des Poids et
Mesures. The binary peta- and exa- loadings, though well
established, are not in jargon use either — yet. The prefix milli-,
denoting multiplication by 1/1000, has
always been rare in jargon (there is, however, a standard joke about the
millihelen — notionally, the
amount of beauty required to launch one ship). See the entries on
micro-, pico-, and
nano- for more information on connotative jargon use
of these terms. ‘Femto’ and ‘atto’ (which,
interestingly, derive not from Greek but from Danish) have not yet acquired
jargon loadings, though it is easy to predict what those will be once
computing technology enters the required realms of magnitude (however, see
attoparsec).*

*There are, of course, some standard unit prefixes for powers of 10.
In the following table, the ‘prefix’ column is the
international standard prefix for the appropriate power of ten; the
‘binary’ column lists jargon abbreviations and words for the
corresponding power of 2. The B-suffixed forms are commonly used for byte
quantities; the words ‘meg’ and ‘gig’ are nouns
that may (but do not always) pluralize with ‘s’.*

prefix decimal binary pronunciation}

kilo- k K, KB, *kay*

mega- M M, MB, meg *meg*

giga- G G, GB, gig *gig*,*jig*

*Confusingly, hackers often use K or M as though they were suffix or
numeric multipliers rather than a prefix; thus “2K dollars”,
“2M of disk space”. This is also true (though less commonly)
of G.*

*Note that the formal SI metric prefix for 1000 is ‘k’;
some use this strictly, reserving ‘K’ for multiplication by
1024 (KB is thus ‘kilobytes’).*

*K, M, and G used alone refer to quantities of bytes; thus, 64G is 64
gigabytes and ‘a K’ is a kilobyte (compare mainstream use of
‘a G’ as short for ‘a grand’, that is, $1000).
Whether one pronounces ‘gig’ with hard or soft ‘g’
depends on what one thinks the proper pronunciation of ‘giga-’
is.*

*Confusing 1000 and 1024 (or other powers of 2 and 10 close in
magnitude) — for example, describing a memory in units of 500K or
524K instead of 512K — is a sure sign of the
marketroid. One example of this: it is common to
refer to the capacity of 3.5" floppies as ‘1.44 MB’ In
fact, this is a completely bogus number. The
correct size is 1440 KB, that is, 1440 * 1024 = 1474560 bytes. So the
‘mega’ in ‘1.44 MB’ is compounded of two
‘kilos’, one of which is 1024 and the other of which is 1000.
The correct number of megabytes would of course be 1440 / 1024 = 1.40625.
Alas, this fine point is probably lost on the world forever. [1993 update:
hacker Morgan Burke has proposed, to general approval on Usenet, the
following additional prefixes:*

groucho | 10^-30 |

harpo | 10^-27 |

harpi | 10^27 |

grouchi | 10^30 |

*We observe that this would leave the prefixes zeppo-, gummo-, and
chico- available for future expansion. Sadly, there is little immediate
prospect that Mr. Burke's eminently sensible proposal will be
ratified.]*