2. Wireless protocols are timing sensitive. Can the card load up a program to process signals? Similar to a kernel in a GPU?
2. Wireless protocols are timing sensitive. Can the card load up a program to process signals? Similar to a kernel in a GPU?
1. It most likely can. They are probably hoping to not bother the FCC by treating it as "test equipment" (that's what Ettus Research does), leaving the responsibility of compliance with the end-user.
1b. I doubt they're getting as much as a Watt out of that board (personal estimate would be a peak of 10-50mW, [10-17 dBm]) so in ideal conditions, significantly less power than a cell phone or wifi signal. Still enough to be trouble in some bands, and it doesn't change the regulatory situation.
2. There is an FPGA, so there is a physical possibility. It's not a trivial thing though, and working with FPGAs is even more remote from what most programmers are used to than working with GPUs. If I had to guess, the plan would be to capture/transmit bursts, with processing in between, instead of real-time signal/protocol shifting.
I'm one of the founders of Per Vices. To confirm;
1. We are currently selling our device as test and measurement equipment. The intended use case is academic or research.
2. You're mostly correct about power - although it turns out that, depending on regulatory jurisdiction and the frequency or band of interest, power level can matter.
3. Actually, we're aiming for real time programming using gnuradio. You're absolutely right about the bursting, though - our initial programs were almost exactly as you described.
Long term, one of our driving goals is to drastically reduce the barriers (specifically cost and expertise) standing between the average software programmer, and a 'hello world' program that can usefully transmit information.
But I also think it's much easier to address problems in software than in hardware...
I am not familiar with the specifics of wifi protocols. Is the wifi protocol fit into the rx/tx burst+processing or RT signal/protocol?
We're aiming to have the user manual up by the end of the week, which should provide much more information.
There's a fair bit of work going on behind the scenes there, and wifi is a pretty heavy protocol to work with like this. I would be very surprised if Per Vices doesn't move to a seperate wifi chip in a future version, if they stick to the 'universal bridge' product. For such a popular protocol, it would make much much more sense.
The pessimist in me says no, but OTOH it is really nice to see something hanging off the PCIe bus, instead of trying to move all the data to the computer over USB or a network card.
More specifically it's an EP4CXG22. Minimum speed grade is C7, but I think we ended up getting some C6's in there (doesn't really change much).