When I was a spotty 21y/o I tried to explain to a 'senior' engineer that his bizarre problem was due to his use of dual port memories. His hardware had 4 separate inputs of a multiplexed analog signal called AMX192 which is much like a video signal. The hardware would sample each channel level and store it into one side of a dual port memory. There were four of these blocks. On the other side was some hardware that would read a common location from each of the dual port memories into latches. A micro would process the values and output the highest of the values. This would repeat. On first look this should all work - the killer was random noise on the output. The engineer was under the mistaken belief that you would either read the old data or the new data from the dual port ram if the was a conflict. Reality is that you read anything. Anyway, since I was the young smartarse who opened his mouth up too much, I was told to fix it. The solution was rather easy in that I had a common clock, when low the write side could write, when high the read side could read. Simple mutual exclusion. Lesson learnt - don't upstage the senior engineer, keep one's mouth shut!