On 11/27/2012 10:50 AM, Richard Cochran wrote: > On Tue, Nov 13, 2012 at 12:33:03AM +0400, Andrey Wagin wrote: > >> I found that this test returns 372293 req/sec without a problematic patch >> and only 334911 req/sec with this patch. A degradation is about 10%. > Wow, that seems a little high. Are you sure? > >> commit 1d1a79b5b94b0aa84e1e78dd9acdcffb12274848 >> Author: Jacob Keller <[email protected]> >> Date: Tue May 22 06:18:08 2012 +0000 >> >> ixgbe: Check PTP Rx timestamps via BPF filter > This is a git commit... > >> Ask me, if you will need more information. Sorry, if it is a wrong alarm. > ... but what two kernels did you test, exactly? > > Anyhow, unless you have enabled PTP time stamping, it is hard to see > how that patch could cause such a large performance hit. > > Thanks, > Richard
Actually the reason for the slowdown is pretty obvious. The problem is that in ixgbe_ptp_rx_hwtstamp the read to TSYNCRXCTL is now always happening if ixgbe is built with PTP enabled. The memory mapped I/O read will be quite expensive. I believe Jacob has a patch in the works to disable the call to ixgbe_ptp_rx_hwstamp until the ioctl is called to enable HW timestamping on receive. Thanks, Alex ------------------------------------------------------------------------------ Monitor your physical, virtual and cloud infrastructure from a single web console. Get in-depth insight into apps, servers, databases, vmware, SAP, cloud infrastructure, etc. Download 30-day Free Trial. Pricing starts from $795 for 25 servers or applications! http://p.sf.net/sfu/zoho_dev2dev_nov _______________________________________________ E1000-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/e1000-devel To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
