@stephenw10

It seemed like performance was less than what it is with the new driver. I didn't do any formal test, but subjectively, the newer driver seems to provide better performance.

Ah, yeah the fact aq0 exists is good. 😀

It may not support the auto-link detection for assigning, not all NICs do.

Try running: ifconfig -vvvm aq0

That will show you what speed it's linked at and what it's reporting it's capable of.

Steve

@stephenw10 Thank you very much for your response, I will keep this in mind, much appreciated.

@stephenw10 said in Original Intel EXPI9404PTG2L20 EXPI9404PT PRO/1000 PT vs Fujitsu PRO/1000 PT Quad Port Gigabit Network Card PCI-E D2745-A11 low profile:

You are basically looking at this:
https://www.intel.co.uk/content/www/uk/en/products/compare.html?productIds=41280,20720

PCIe 1 vs 2

Some virtualisation support in the 82580.

But 82571 is an em(4) NIC but 82580 is an igb(4) muti-queue NIC. If you have multiple CPU cores that is going to load them far more efficiently.

I'd get the newer card given a choice.

Steve

Dear Steve!

Thank You so much for detailed answering.

I keep in memory about 82580 igb(4) multi-Queue, but anyway asking people’s with better knowledge. ;)

Historically there have been some issues with raid controllers. YMMV.

But for me the most compelling reason is if you ever need to recover from a hardware failure you can move the drives into something else and they will boot without needing the exact same controller with the same firmware on it etc.

Steve

@stephenw10 yes, it is not an actual BMP parser, I just used ImageMagick's (known) offset when creating a 1bit bitmap.
something like

convert -pointsize 11 -size 128x64 -font "Arial" -background white -fill black label:"some text\nand some more text" -monochrome text.bmp

and then

perl /path/to/bmp2lcd text.bmp

should work

edit: another thing I noticed: the LCD needs local modem control lines (I can't imagine why, since only three wires are connected to it), like so:

stty -F/dev/ttyS1 clocal speed 115200

@jc1976 said in discrete ethernet adapter efficiency vs onboard nic:

LOL this is incredible stuff!

Aha, 😉

Just read it carefully and you'll see that it's logical and a really good description.
Experiment with the settings (not in production) the loader.conf.local file will be your good friend

+++edit:

this is just a sample schema, a lot of things in it are no longer relevant (bc. : FB12.2-STABLE).......

hw.pci.realloc_bars=1
net.inet6.ip6.auto_linklocal=0
net.isr.maxthreads=-1
net.isr.bindthreads=1
kern.ipc.nmbclusters=1000000
net.inet.tcp.tso=0
net.inet.tcp.lro=0
dev.igb.0.fc=0
dev.igb.1.fc=0
dev.igb.2.fc=0
dev.igb.3.fc=0
dev.igb.4.fc=0
dev.igb.5.fc=0
dev.igb.6.fc=0
dev.igb.7.fc=0
dev.igb.0.eee_disabled=1
dev.igb.1.eee_disabled=1
dev.igb.2.eee_disabled=1
dev.igb.3.eee_disabled=1
dev.igb.4.eee_disabled=1
dev.igb.5.eee_disabled=1
dev.igb.6.eee_disabled=1
dev.igb.7.eee_disabled=1
legal.intel_igb.license_ack=1
hw.igb.rx_process_limit=-1
hw.igb.tx_process_limit=-1
hw.igb.rxd=2048
hw.igb.txd=2048
hw.igb.max_interrupt_rate=128000
net.pf.states_hashsize=1048576
net.pf.source_nodes_hashsize=524288
net.inet.tcp.syncache.hashsize=2048
net.inet.tcp.syncache.bucketlimit=100
net.inet.tcp.syncache.cachelimit=65536

If you try to push the traffic at 4x the speed and encrypt it then, yes, you will need more CPU. 😉

100Mbps IPSec is not that hard to achieve (depending on traffic type, latency etc!). The 3100 will pass 300-350Mbps over IPSec as long as you're using an encryption supported by it's crypto hardware (AES-CBC).
Yeah I'm sure that HP would do it but you'd probably want to run pfSense virtualized to make more efficient use of the hardware with other things.

Steve

Yeah not a listed device in pfSense or FreeBSD even in the latest code. Some devices can be used using a generic driver if they report as a cdce device but you should see it appear as ue0.
Try connecting it and then checking the system log for the new entries it generates.

However it looks like it's probably unsupported. Try a different USB Ethernet device or just use one NIC and VLANs since USB Ethernet is better avoided anyway.

Steve

Exactly like the 2100 it doesn't have a modem built in. The SIM slot is required to use a modem in the m.2/mPCIe slot but does not indicate radio hardware is included.

Steve

Not as far as I know. You could open a feature request, I don't see one yet.

Installing the driver via the kmod pkg is not hard. Though I have nothing to test it with.

Steve

@bingo600

Got a bit interested here ...

According to the Intel i350 reference design here (last entry / bottom):
Document: 323852
https://www.intel.com/content/www/us/en/products/details/ethernet/gigabit-controllers/i350-controllers/docs.html?s=Newest&p=2

Page 4 - SMBus interface
U7 (The i350 controller chip) , is connected directly to J42 (PCIE slot) , with just a couple of 10K pullup's.

So B5 & B6 aren't connected directly to a config eeprom , but to the i350.

On Page 7 - Support circuits.
We see that U7 (i350) has connection to a SPI Flash , and an SPI EEPROM.
I'd assume the SPI Flash holds Boot Code & Other stuff.

And according to - The i350 Datasheet - Page 12
Document : 333171
https://www.intel.com/content/www/us/en/products/details/ethernet/gigabit-controllers/i350-controllers/docs.html?s=Newest&p=2

The EEPROM holds the size of the SPI FLASH flashzise , and prob. other goodies.

So i assume that : What is read via the PCie SMBus , and confuses the bootcycle is residing in either the SPI EEPROM or the SPI FLASH that is connected to the i350.

And since it is the i350 that is responsible for SMBus communication, there's probably no "Easy fix" like lifting an I2C eeprom leg or two , in order to avoid unmasking PB5 & PB6.

Fun stuff ....

Hmm ... In the DS pg. 99 an OEM VPD area is described
1684f166-8b81-4b31-87b3-cefc42c961ac-image.png

Maybe setting the word at 0x2F to 0xFFFF , would skip the Dell OEM information , and stop confusing the PC ??

/Bingo

Probably not in pfSense, it looks like a driver issue.

Unless there is some other sysctl that can alter that behaviour. I don't see anything listed though.

Steve

Create the file /boot/loader.conf.local (if it doesn't exist) then add to it the line:

amdsmn_load="YES"

Steve

Run top -aSH at the command line and see what's using the CPU per core.

It might be stuck at a power saving CPU frequency or something similar.

Steve

@stephenw10 Thanks for the information. I tried that and it ended up making the performance worse, down into the low 700s on download. I tried a couple hacky things including trying to add someone elses BSD patch for muti-threaded PPPoE and that blew the whole thing up. I'm going to table this for the time being and I was able to improve the performance to ~890/930 by using OpenWRT with some tweaking. Thanks for all your help!