<p>I am running boinc_4.09_i686-pc-linux-gnu on a dual hyper-threaded P4 Xeon machine running Linux 2.6.8.1. I have configured BOINC to run on 2 out of the 4 logical CPUs.</p>
<p>I have noticed that the BOINC process has a nasty habit of consuming 100% of one CPU under certain circumstances. I attached the debugger to it and took a stack trace, so that I could see what it was doing at the time:</p>

(gdb) bt
#0  0xffffe410 in ?? ()
#1  0x400db03b in _IO_new_do_write () from /lib/tls/libc.so.6
#2  0x400dbac8 in _IO_new_file_overflow () from /lib/tls/libc.so.6
#3  0x400dc6b2 in _IO_new_file_xsputn () from /lib/tls/libc.so.6
#4  0x400bc136 in vfprintf () from /lib/tls/libc.so.6
#5  0x0807bae3 in MIOFILE::printf(char const*, ...) (this=0x4001b000,
    format=0x4019ffdc "

ASIDE: WHERE THE 8^%! DID THE REST OF MY STACK-TRACE GO?????
<p>I think this problem appeared when BOINC failed to download one new work-unit. Only one instance of SETI@Home was running at the time.</p>
<p>Can anyone reproduce this? It does worry me because the BOINC process isn't 'niced', of course. So when it does this, it is effectively wasting my CPU time.</p>
<p>Cheers,
Chris</p>

We're seeing this problem on dual processor G4 Mac's as well. Whenever one processor is idle, the disk I/O starts going nuts.

See: http://setiweb.ssl.berkeley.edu/forum_thread.php?id=4774