AM2432: What is the speed/clock-usage differences between using read R31 to get 12bit input data and using parallel 16bit input to get 12 bit input data?

Part Number: AM2432

Tool/software:

Hello Experts,

I am new to TI MCU (first time to design TI MCU), and I am not familiar with TI MCU PRU functions, and I only know the Sitara MCUs can operate I/O pins fast.

Because AM2432 PRU is fast, I want to use it to replace a FPGA to read a fast parallel output ADC.

I roughly know that the R31 of the PRU can be used to read the Voltage level state of up to 20 (EGIP) input pins. 

I also discovered that PRU has a 16bit parallel input mode (work with a PRU clock input pin, #17th EGIP pin).

(1). I was wondering the speed differences between using the above 2 kinds of read-12-input-pin case. 

(2). Also I roughly know that there is a very fast broadside connection command (32byte) that can be used to do fast I/O on MCU pins. 

Can the broadside command be used to do read-12-input-pin signal on EGIP pins (or GPIO pins)?

If yes, is the speed/clock-used by broadside command be faster than the above two read-12input-pin case?

If this is the fastest way, how to write code to extract those 12input-pin data from the 32byte data?

Thank you very much.

Best Regards,  JasonC.