Yes, they can be changed on the fly. Each element, such as frequency, gain or bandwidth can be reconfigured very rapidly by simply loading a couple of bytes of new data. That makes the devices great for software defined radio or cognitive radio systems. Add to that the ability to depower unused elements, e.g. ADCs (if you are using external convertors), or one of the dual channels for SISO operation.
By combining the programmable RF chips with an FPGA, such as a Zynq, provides a very capable mixed signal system with software control and hardware acceleration to process algorithms.
Very cool. The high end applications will be a great fit for these devices and the big SoC FPGAs. How about on the low end? Can an MCU with good low power operation modes be used? Can we do really crazy stuff like computing the setting we want to configure the part with and then change the setings as needed?
Low end applications can be served by MCUs. In fact, the LMS7002M incorporates an on-chip microcontroller (it runs 8051 code, I think) that offloads the baseband chip from several tasks like calibration. As I understand it, the on-chip microcontroller can also run code stored externally.
Both devices use SPI to transfer data into their chips, so if the MCU has sufficient intelligence it can look up the relevant code (say, what frequency to receive or transmit) and load that directly into the device. That's how devices in small cells, such as femtocells, can scan the spectrum to figure out what frequencies are being used locally as part of a self-organizing network. It's also why both devices have on-chip Received Signal Strength Indicators.