Shared memory oddity

Why would this happen?

I had a server where postgres appeared to be running very slowly. Following
some advice I found on-line, I boosted the shared memory max for that
machine and told postgres to take 15200 buffers (about 128 meg of shared
memory). For a while this worked great, but today, the system runs very
slowly again.

free and shared memory look like this:

[slane(at)iep slane]$ free
total used free shared buffers cached
Mem: 513940 511652 2288 295248 38100 371896
-/+ buffers/cache: 101656 412284
Swap: 0 0 0
[slane(at)iep slane]$ ipcs

------ Shared Memory Segments --------
key shmid owner perms bytes nattch status

------ Semaphore Arrays --------
key semid owner perms nsems status

------ Message Queues --------
key msqid owner perms used-bytes messages

[slane(at)iep slane]$ cat /proc/sys/kernel.shmmax
cat: /proc/sys/kernel.shmmax: No such file or directory
[slane(at)iep slane]$ cat /proc/sys/kernel/shmmax
160000000

Postgres used to have a big section of shared memory in use. Today, nothing.

postgres 7.1, running on LinuxPPC on a PowerMac G4.

Any ideas?

-- sgl