While trace flag -T8038 has been noted, I think it deserves more information.

Hp has a support Document ID: c02110402 that explains it a bit more than the KB article

DESCRIPTION

ProLiant servers that have multiple processor cores and that are running Microsoft SQL Server 2005 SP3 or SQL Server 2008 may stop responding when under a heavy I/O processing load. If Automatic Server Recovery (ASR) is enabled on these servers, the server may reboot when the server stops responding.

SQL Server 2005 SP3 and SQL Server 2008 use mmtimer (Multimedia) timer rather than the RDTSC timer, which changes the clock granularity (to 1ms). This change may result in clock-drift or an unresponsive server if the server uses enhanced power management technologies that change CPU frequencies.

Note: This issue may occur with any ProLiant server that supports power management technologies. Refer to the Affected Platforms included at the bottom of this advisory; however, other platforms may also be affected.

SCOPE

Any ProLiant server that supports power management technologies is running SQL Server 2005 SP3 or SQL Server 2008 on Windows Server 2003.