Sweet :) That sounds more like it. --Bill
On 8/12/05, Matthew Lenz <[EMAIL PROTECTED]> wrote: > client: iperf -P 2 -w 128k -c server > server: iperf -w 128k -s > > yeilded 940 Mbit/sec > > ----- Original Message ----- > From: "Bill Marquette" <[EMAIL PROTECTED]> > To: "Matthew Lenz" <[EMAIL PROTECTED]> > Cc: <[email protected]> > Sent: Friday, August 12, 2005 4:10 PM > Subject: Re: [pfSense-discussion] iperf > > > I usually use: > client: iperf -P 2 -w 128k -c server > server: iperf -w 128k -s > > And I'd recommend using: > http://dast.nlanr.net/Projects/Iperf/iperfdocs_1.7.0.html > > Also, I'm not sure FreeBSD uses polling mode for the em driver by > default. Are all your NICs on the same IRQ, if not can you set them > to the same IRQ? At least in OpenBSD same IRQ improves performance > somewhat due to how interrupt handling works (if you're in the > interrupt handler for IRQ x, loop through all devices on IRQ x and > process anything they need done). > > --Bill > > On 8/12/05, Matthew Lenz <[EMAIL PROTECTED]> wrote: > > What are some good command line settings to get an idea of what kind of > > throughput pfsense has? currently I'm just doing: > > > > host1: iperf -c host2 > > host2: iperf -s > > > > and getting about 613 Mbits/sec. Where host1 is on the LAN net and host2 > > is on the OPT1 net (in other words the route through the firewall). If > > I run: > > > > host3: ipeft -c host2 > > > > I get about 935 Mbits/sec. Where host2 and host3 are both on OPT1 net > > (and don't go through the firewall). > > > > Currently all LAN and OPT1 net only have a single rule which passes all > > traffic. > > > > For the curious I've got redundant supermicro 5014C-MFB systems with > > 3ghz HT cpu's, 1gig dual channel memory, 2 onboard Gig ethernet and a > > PCI-X 4 port Intel Gig ethernet card. The switches are Dell 5324 24 > > Gig. > > > > >
