[hetman]

The AntSDR is a pretty solid piece of kit. It's compatible with the ADALM-PLUTO, so the available documentation is pretty great, and it's got both a compact and very solid construction (unlike a lot of the other offerings in the market).

The one thing to keep in mind (my experience is with the E200 though the E310 hardware architecture seems pretty similar from what I can see) is that the sampling rate is somewhat CPU bound by the onboard CPU. In spite of having a GigE interface, your sample rate will be limited to somewhere on the order of 110 MB/s which will give you a combined (Rx and Tx) sample rate of under 15 mega samples per second. I have heard of some people try to overclock these to boost the sample rate a bit. If that is all you need I could definitely recommend it though.

[anyfoo]

But the AD936x is directly connected to the programmable logic (FPGA) of the Zynq, not the PS (CPU), so you can either do the bulk of your processing (e.g. filtering, demodulation, possibly higher parts of a protocol stack) in the FPGA, or, if you absolutely have to, let the FPGA pipe out raw IQ sample packets onto the Ethernet, because IIRC the Ethernet controller itself can also be routed into the FPGA using MDIO, no?

I haven’t tried this yet (despite owning an AntSDR E200), but that’s what the schematics of the AntSDR and my prior experience with Zynq suggests.

[hetman]

You're right in that only the libiio firmware relies on the CPU to forward the samples out to the ethernet interface. The UHD firmware on the other hand does not seem to occupy the CPU to do this, however I have still had trouble getting transfer speeds to go much faster than they do with libiio (and I've had a lot of other problems with UHD personally). I'm not certain what could cause this limitation though.

In theory the hardware probably could do what you're describing if someone wanted to hack the FPGA internals themselves. I'd love to hear if you ever have any success getting things running better.