In fact, it turns out I used to have a 7345.  Here is my disktab entry for
it:

max345:Maxtor 7345S:\
	:ty=winchester:dt=SCSI:se#512:nt#64:ns#32:nc#329:rm#3600:\
	:pc#673792:oc#0:tc=unknown:\
	:pd#673792:od#0:\
	:pe#673792:oe#0:be#8192:fe#1024:te=4.2BSD:

Two things to note:  this uses ficticious geometry (which will have the effect
of leading the filesystem code to believe that many cylinders are the same
cylinder and thus seeks won't cost any time), and it sets up the whole disk
to be one big partition on sd1e.  If you want multiple partitions, you can use
max345:Maxtor 7345S:\
	:ty=winchester:dt=SCSI:se#512:nt#64:ns#32:nc#329:rm#3600:\
	:pc#673792:oc#0:tc=unknown:\
	:pd#673792:od#0:\
	:pe#336896:oe#0:be#8192:fe#1024:te=4.2BSD:
	:pf#336896:of#336896:bf#8192:ff#1024:tf=4.2BSD:

(I *think* I split that in half, check my arithmetic.)  Note that you can
have an "e" partition on multiple disk drives (to correct what seemed to
be a misconception in an earlier message; this is not DOS, and sd0e has
nothing to do with sd1e (or on my system, with sd2e or sd3e either :-).

A version of the first which correctly reflects the reported geometry (if
not the actual geometry) would be

max345:Maxtor 7345S:\
	:ty=winchester:dt=SCSI:se#512:nt#4:ns#76:nc#2220:rm#3600:\
	:pc#674880:oc#0:tc=unknown:\
	:pd#674880:od#0:\
	:pe#674880:oe#0:be#8192:fe#1024:te=4.2BSD:

Note that you get a few extra usable blocks out of that, assuming the drive
isn't overreporting its capacity by claiming 76 sectors on *all* tracks.
When the system boots, the kernel also reports the drive's total reported
capacity in sectors, check to see that is more than 674880 sectors.

Note that I haven't every *used* the last version there, but it should be
correct.  The first disktab ran quite happily, if possibly suboptimally,
for quite some time.