Hello all,

I am testing an E810-XVV-2 twin port community card and I am dealing with actually dangerous efficiency in comparison with an outdated (and sadly out of date) X520.

In the meanwhile I want simply 10Gbps hyperlinks.

Below Home windows 10 64 bit 22h2 (desktop)  I wish to use the 2 ports of the board to learn from 12 gigabit cameras related to an Aruba 2730F swap. Every port of the board learn 6 cameras (about 5Gbps). Notice: these cameras sends frames utilizing 8K UDP telegrams (i.e. lots of small telegrams). 

I normally doesn’t tweak an excessive amount of the driving force settings: I simply allow the jumbo frames, disable all of the CRC offloading choices and enhance the variety of obtain buffers. All the opposite choices, specifically the RSS and the variety of obtain queues, are stored to their default worth.

Sadly this board appear to disregard what it’s set as “variety of obtain queues” and at all times use 1. As you possibly can see the default is 16

however after I verify within the powershell the precise quantity is 1.

After I begin the software program to seize from the digital camera is see that each one the load is deal with by a single CPU thread.

I’ve the identical behaviour in two totally different PCs with totally different motherboards and processors: a single i7 processor MB and a twin Xeon processors MB.

Any trace to get this setting to a price totally different from 1: easy a single CPU thread just isn’t in a position to deal with all of the interrupts for the incoming telegrams. 

Regards

Giovanni