khaw@parcplace.com (Mike Khaw) (11/10/89)
Some time ago, an issue of the Sun S/W Bulletin published a list of
mappings between hostids and model-types (e.g., 12xxxxxx is a 3/50).
However, since then, newer models have come out and I no longer get the
Sun SWB.
So, does anyone have a more complete/correct list than the following:
hostid model
------ -----
11001* nominal 3/160
11002* nominal 3/75
11003* nominal 3/140
11004* nominal 3/160
11005* nominal 3/180
12* 3/50
13000* nominal 3/280
13004* nominal 3/260
14* 3/110
17* 3/60
21* 4/??
22001* 4/110
23* 4/330 (?)
3* 386i
4200* 3/80 (?)
51* SparcStation 1
The ones marked "nominal" apparently have different enclosures, but
interchangeable CPU boards, so the hostid may not match what the badge on
the box says. The ones marked "(?)" are based on single samples.
What about the gaps: were there ever 15* and 16* hostids in the Sun-3
line? I'm also not sure why my list has both 11001* and 11004* as 3/160.
Thanks,
Mike Khawdupuy@cs.columbia.edu (11/16/89)
/bin/hostid format: 8 hex digits: A M NNNNNN
A = architecture:
0=sun2, 1=sun3, 2=sun4, 3=sun386i, 4=sun3x, 5=sun4c
M = machinetype:
sun2 1=multibus, 2=vmebus
sun3 1=160, 2=50, 3=260, 4=110, 7=60, 8=Eurocard
sun4 1=260, 2=110, 3=330, 4=460
sun3x 1=470, 2=80
sun4c 1=60
NNNNN = cpu serial
system serial number format: 3 digits, letter, 4 digits: Y WW L NNNN
Y = year: 4 = 84, 5 = 85, ...
WW = week: 23 = 23rd week of year
L = letter, C through F (possibly manufacturing site?)
NNNN = number of unit
The complete dope on the hostid # can be found in the include file
/usr/include/sun{2,3,4,386}/cpu.h.
inet: dupuy@cs.columbia.edu
uucp: ...!rutgers!cs.columbia.edu!dupuy