Underutilisation of CPU --- am I PCI bus bandwidth limited?

2004-10-24 Thread lukem . freebsd
I posted this to freebsd-performance, but have as yet not satisfactorily answered the question. Since it is primarily network related, I'm reposting it here. I have been doing some benchmarking as a part of some driver development work, and have encountered a phenomenon I can't explain. I am us

poor performance of nge interface

2004-10-25 Thread lukem . freebsd
I notice that there are some poorly defined default values: dev/nge/if_nge.c: CSR_WRITE_4(sc, NGE_IHR, 0x01); dev/nge/if_nge_reg.h: #define NGE_RX_LIST_CNT 128 #define NGE_TX_LIST_CNT 128 net/if.h: #defineIFQ_MAXLEN 50 netinet/ip_input.c: static int ipqmaxlen = IFQ_MA

em pause frames

2005-02-24 Thread lukem . freebsd
Is there an easy way to enable or disable pause frames and autonegotiation of flow control for the em driver? (other than changing em_fc_full to em_fc_none and recompiling). Something similar to linux's "ethtool" command, or an ifconfig parameter is the type of thing I'm after. (And yes, I've

Re: changes to make ethernet packets able to be unaligned...

2005-03-20 Thread lukem . freebsd
Why not just fix the protocols which do unaligned accesses? Is speedup of doing non-byte-wide reads when manipulating packet headers really larger than the cost of aligning everything? I doubt it. Your average tcp packet shoule only cause a few unaligned accesses anyway, so we are putting in a

Re: Network performance in a dual CPU system

2006-02-12 Thread lukem . freebsd
On Sat, 11 Feb 2006, dima wrote: There are several software (FreeBSD specific) options though: 1. You should surely try polling(4). 50kpps mean 5 interrupts and the same amount of context switches, which are quite expensive. While this was true in the 80's, it is blatantly wrong for any

Re: [fbsd] Re: [fbsd] Network performance in a dual CPU system

2006-04-30 Thread lukem . freebsd
On Thu, 27 Apr 2006, Robert Watson wrote: Yes -- basically, what this setting does is turn a deferred dispatch of the protocol level processing into a direct function invocation. This reminds me of a problem I saw about a year ago, where the number of entries in the DMA ring was much greater (

Trouble booting NFS root

2006-05-31 Thread lukem . freebsd
I'm trying to set up a small cluster of diskless boxes using FreeBSD 6.1. So Far, PXE loads pxeboot which loads the kernel. But the kernel stops in the middle of booting for no apparent reason. The last thing on the console is "Timecounters tick every 1.000 msec", which usually occurs just bef

Re: Trouble booting NFS root

2006-06-01 Thread lukem . freebsd
make sure you have this: -rw-r--r-- 1 root wheel diskless_root/etc/rc diskless_root/etc/rc: /bin/sh #which points your rc file to shell so that you are your own rc! While that is a useful tip, the NFS trace shows that the kernel doesn't even try to read any files. The last thing that gets re

Re: Trouble booting NFS root

2006-06-04 Thread lukem . freebsd
On Thu, 1 Jun 2006 [EMAIL PROTECTED] wrote: I'm trying to set up a small cluster of diskless boxes using FreeBSD 6.1. So Far, PXE loads pxeboot which loads the kernel. But the kernel stops in the middle of booting for no apparent reason. The last thing on the console is "Timecounters tick eve

Re: Plans to port OpenBSD trunk(4)?

2006-08-03 Thread lukem . freebsd
On Thu, 3 Aug 2006, Antony Mawer wrote: Is there any interest or plans underway to port the trunk(4) feature from OpenBSD? OpenBSD's trunk(4) appears to be exactly what I'm looking for, but there doesn't appear to be anything I can find on a port to FreeBSD. http://www.openbsd.org/cgi-bin/m