Hello, I just created a 2.6.17.6 kernel (patch as of yesterday), and tried again, with much the same results. So, I suspect it's somewhere in the guts of the 2.6.17 kernel - switching back to 2.6.16 also brings performance back to acceptable levels (already included in the report).
Should I report this bug upstream? Best, --Toni++ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]