Nice heatsink on that puppy. We had to throw some cooling fans on our X540-T2's. Literal finger burn if you touch and it was melting some nearby 3d printed components. Hopefully these are significantly more efficient.

Heard about this a couple months back & it sounded like a huge leap forward. Comparing it to the AQC113 card I bought a month ago suggests it's a pretty incremental improvement though. Both X1, both fanless, both low single digit watt.

Cheaper means more mobos will have 10gig I guess

Hardware CRC at least but still leaves hardware TSO and LRO to be desired least for these chip's intended usage.

Havent used realtek NICs for some time on my hardware but are their drivers still very CPU-bound? I recall them doing a lot more work via CPU than say, a comparable Intel NIC would.

I don't understand how their benchmark is getting close to 20 Gbps. From what I understand a PCIe Gen4 x1 slot should be capping around 16 Gbps when doing full duplex or 8Gpbs in each direction.

Edit: Maybe they are testing within one system?

> We tried PCIe card NIC to the onboard NIC, and onboard NIC to onboard NIC.

Finally a good use of those x1 slots. Only sucky part is I cant find a relatively cheap silent switch that does 10G copper. Have a Linksys with four SFP+ slots which was very cheap new.

Really wish they tested the power usage and not just 2-4W. Previously 10Gbps were using 5-7W. Newer Intel 10Gbps Managed Sub 4W but are expensive.

Worth mentioning this 10GbE on Paper also support 5Gbps and 2.5Gbps as well. This is going to be some very nice upgrade for those who have previously laid some decent CAT6A cables.