> hmm, that's interesting, could you find out what was in the blocks
> that you were able to cluster?  I'd guess it's inode data, but it
> could be something else.
I've killed my dev environment, so it'll be next week b/4 I can answer
this one.

> it's kind of disappointing that there was no measurable improvement
> in performance, though.
I'd expect CPU time to dominate this particular benchmark.

> could you try experimenting with ccd or
> raidframe and see if it helps noticably in that context?
That'd be hard for me to do: I don't have 2 similar SCSI disks.
The 2 spares I do have were minted about 10 years apart from each
other, which'd give some...interesting results.

> it'll probably
> help if you use a machine with a slower CPU as well.
I might be able to get it going on an hpcmips machine
Well, if that's got a physically indexed cache.

> my point with trying
> to see a performance improvement is that if we think there should be
> a performance improvement but there isn't one, then maybe something
> isn't working correctly.
Understood.  But I was thinking/writing in terms of wall-clock time,
not disk throughput.  I took a look at # of Megabytes transferred vs
the time the disk was busy for both compiles.  There's a 5% increase
in disk bandwidth, which is in keeping w/ the improvement in # of
transactions.

> -Chuck
Chris.