I'm benchmarking the HP FPD slave interface of a TE0802 board featuring a xczu2cg-sbva484-1-e SOC.
I'm usign a custom AXI master to flood the interface with read or write operations.
I noticed that, when the burst size is not a power of two, the slave will eventually get stuck and stops replying to read requests.
I'm attaching two waveforms, showing a successful (burst 4) and and an unsuccessful (burst 3) case, with a reading benchmark.
Is there anything that I'm missing?