Out of curiosity, how would on board pci-e be comparatively to dual gig on board?

@covert: Is my speed limit simply not enough resources in the system ? I find it hard to belive the dd-wrt router with a 216Mhz CPU and 32Meg ram can out perform it. In some work loads a faster CPU may actually be a disadvantage.  Here's why. Interrupts are a high overhead activity: not only does the previous context have to be saved and restored but the cache can be trashed by high interrupt rates. You may have an application running. That application has high locality of memory reference so that most of its memory accesses are satisfied from the cache. An interrupt happens and for a while most of the memory references for the interrupt handler can't be satisfied from the cache but from much slower main memory. By the time the interrupt handler returns there may be very little of the application's memory in the cache so when the application runs again the cache gets refiled with application data and code. If there is a heavy network load and the CPU is fast enough that it generally processes only one frame per interrupt then the cache related overhead will be considerably higher than it would be on a CPU that was somewhat slower and consequently processed more frames per interrupt. Some NICs have a mechanism allowing the device driver to specify that interrupt requests should be delayed some short time to increase the likelihood that there will be multiple frames available for processing by the interrupt handler. You don't say what NIC is in your system so I can't comment on whether it has that interrupt delay capability. Some people have found that enabling polling (from the web GUI, System -> Advanced, check the box Use device polling in the Miscellaneous section) produces a considerable improvement in throughput. (This comes with a cost of a somewhat increased latency but the increase should generally be no more than about 1 mS and will often be much less).

It just keeps going up and down…up and down.  Thankfully I found the receipt and will probably return it since nobody seems to have a solution.

You where tryng that with a software raid.  I would actually get a raid controller that is support by freebsd and connect the disks to that controller.  It would be much faster at that point.  Also if a drive should fail the controller is going to be making a beeping sound and you connect to the bios of the controller and see exactly which disk has failed. Software RAID is much sower in comparision to hardware RAID. RC

Well. I ended up going back to linux and maxing out at around 80 megabits. Unfortunately i started having a very similar problem to what I originally had where certain sites and stuff wouldn't work. It was related the MTU. I originally had it set at 1492 and later changing it to 1454 fixed the issue but then it started having problems even at 1454. Certain sites and stuff wouldn't load. I need the connection to be reliable and linux was slightly lower latency so I switched back to linux.

thanks Cry Havok i will do it at night and post the result (i hope it will recognize it) rog

@jahonix: @kc8apf: I recently bought a Lanner FW-7520 for my house. Which version did you choose: 600MHz or 1GHz and how much did it cost? There are actually 3 versions: FW-7520A - 600Mhz Celeron M FW-7520B - 1Ghz Celeron M with no cache FW-7520C - 1Ghz Celeron M with cache I bought the FW-7520B directly from Lanner USA.  The cost as of March 29, 2009 was $430 + tax + S/H.  I had tried to get the FW-7520C originally, but it wasn't immediately available.  Being an OEM device, they wanted to sell to me in quantity and give me a price break.  I told them it was for a custom application and they happily sold me quantity 1.

I must agree that the provider is giving you such high bandwidth w/ such a setup is to prevent people from aggregating the bandwidth. You can use separate gateways for each wan.  Several years ago, I've seen it done on a clarckconnect setup, until they started charging a subscription for such features.  They just package a bunch of other people's "hard work/ingenuity" into a very nice gui and charge quite a bit for.  I believe it's based on Centos/Redhat/Linux.  It's quite reliable, I've been told.  You might wanna check it out.

It will only be used as a Firewall/Router… maybe with a WiFi card in it. No VPN or anything like that.

The Hardware Compatibility List can be found here, from the front page of the pfSense web site.  It looks like the NFE driver may support the RTL8211B chipset. Your other solution would be to get a mini-itx board with a 4x PCI-e slot and a dual port Intel Server grade NIC.  That should allow you to "make do" with a lower spec CPU, possibly an Atom based board.

http://www.newegg.com/Product/Product.aspx?Item=N82E16820141419 Been up for several months now…

If It would help the switch I'm getting is this one. Cisco Small Business SGE2000P http://www.newegg.com/Product/Product.aspx?Item=N82E16833124213 Seeing how It does support 802.3ad It should then support Link aggregation (Or bonding ;))

I like the 5501 for the head office, but an Alix box might be a better fit for the branch offices. It's close to the specs of the 5501 and cheaper than the 4801. http://www.pcengines.ch/alix.htm There are some rough numbers in this post- http://forum.pfsense.org/index.php/topic,14581.0.html Not scientific, but they give you an idea how the crypto cards help. The Geode boards (5501 and Alix) have a built-in crypto chip. As detailed in the referenced post, the driver got incorporated into 1.2.3 builds, but actually didn't help at all and rendered a hifn card useless to boot. I noted this in a support ticket and suggested making glxsb a module instead of in the kernel. I never heard back, and haven't tested recent snapshots. The glxsb driver may be working in the 7.2 based builds. I would suggest you do some tests before deploying. If you need higher IPSec throughput, you may want to stick with 1.2.2 and a hifn.

In order I'd look at: PSU Cooling (CPU and case) Motherboard CPU

Race condition?

so you got your SGS 1660 from the same ebay seller i did.. hes a awesome guy… i was the original buyer of the first one that had PFSense loaded on it.. i have a total of 4 of the boxes... i asked him if he could try pfsense and send me the dmesg of it. note: there are extra headers on the motherboard for the ps/2 keyboard and mouse as well as the VGA header to connect to a monitor as well. the chipset  max's out at 2G.. i got 1G in my box 2x1G dimms works great in this box. i have the docs here on motherboard info. my box's have been ROCK solid.

Thanks for the update. Having (briefly) looked at the hardware profiles of the different Huawei HSDPA modems, it looks like most of them are reporting the same hardware ID and profiles. So, my guess is that if it works with the 220, since the 220 seems to be somewhat of a "reference" platform for Huawei, it will work with other modem models too. I've successfully gotten the E181 (T-Mobile webConnect) to work on both Linux and Mac.

Do not play. Just set up the port on the switch to auto-negotiate. This is the only correct setting unless you want to set up speed/duplex on pfSense.

Maybe you're right ;) I hadn't heard of a lot of problems with the Marvell NICs and I've used them myself without issue, so I assumed the problem was the Realteks Maybe OP does need to replace them all.