some low-memory 2.1.109 compile times

Robert Hamilton (hamilton@iits-129.kdi.com)
Sun, 19 Jul 1998 21:48:55 -0500

This is not expected to prove anything, have no idea
if these results are useful at all, but here are some
numbers the reader may or may not find interesting. In
any case the experiment was painful enough that others
may be reluctant to repeat it, so I report the results.

In each case I rebooted [1] 2.1.109 and mem=8M or =12M option.
Starting with pre-cleaned source, timed 'make zImage' [2]
with the following options and results:

mem = 8 MB, no other activity on box:
freepages = 20 40 60, compile-time = 92 min
freepages = 5 10 11, compile-time = 84 min
mem = 8 MB, toying with X on another vt:
freepages = 20 40 60, compile-time = 113 min
mem = 12 MB, no other activity:
freepages = (default), compile-time = 48 min

The 12 MB was just large enough for little or no
swap activity, and the performance was not noticebly
different from usual with 36 MB.

Just to see if anything interesting happens:
mem = 8 MB, no swap enabled
overcommit = 0, compile fails with out-of-memory enabled
overcommit = 1, box grinds to a halt, but responds to alt-SysRq

[1] A 486 at 50Mhz, with ide drives (hdparm -t gives 1.5 MB/s)
[2] Can't imagine anyone actually dedicating their 8-meg 486's
to kernel builds, but at least they can if they want to.

--


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html