I've been reading this thread agian and I keep coming back to the explaination given for why this oil pump becomes ineffective after the engine has been worked hard.
I accept that it happens, but the "why" offered has never rung true. Cavitation occurs under a couple circumstances and eventually that may part of the failure chain, but the primary cause I think lies elsewhere.
When oil, water, air or any other fluid is pushed at some pressure through a bypass, orifice, or leak, the fluid heats up due to friction with itself. If left unchecked in hydraulic systems, this can create temperatures high enough to burn up lots of o-rings and other seals. I once had a faulty power washer at work that when the trigger on the cleaning wand was not depressed, the water circulated in the pump through a bypass and heated up. When the trigger was squeezed the water pressure in the hose dropped and the water coming out of the pump turned to steam. When the steam got to the nozzle it came out at a much higher velocity vs the water. This greatly increased the recoil of the wand and made it unsafe to use until it was repaired.
I'm guessing he same thing is happening here. Since unlike most oil pumps, this one bypasses into the inlet rather than dumping the bypassed oil back to the pan, in situations where there is likely to be a lot of high speed running, (and as one would expect a lot of bypassed oil due to high pressures,) the oil which is heated a little more every time it gets bypassed, is fed right back into the pump thus overheating the pump.
Overheating can loosen up the clearances in the pump making less effective while it's hot. Heat can also make cavitation more likely since the oil is now very hot and could easily boil if exposed to a low pressure as is found on the suction side of the pump.
I've also puzzled over why the Mazda engineers would build a pump this way, and I don't think it was by mistake.
First and foremost these cars and these engines are meant for the general public, not racers, and not enthusiasts who will beat the daylights out of them like we do. Yes, I know, the Mazda marketing department pushes all that ZOOM-ZOOM stuff, and all that, "On any given weekend there are more Mazdas on race tracks than any other brand." The engineering department may not have gotten the memo that said, "We need to make our oiling systems capable of extended service on the race track." They probably WERE given the memo that said, "Find ways to give better mileage especially on cold starts."
I think they intentionally made the pump work this way to effectively heat the oil on cold starts. When an engine is cold, the viscosity of the oil is pretty high. In addition, the clearances are slightly tighter. All this makes the engine work harder just to stay running at idle. It also makes the oil pressure pretty high as well. There is likely some bypassing going on that heats the oil and helps shorten the engine warmup. Once the engine heats up to normal temperatures, pressure drops and the bypassing is not a factor and never will be for the average driver.
But sustained high speed work will overheat the pump and cause problems.
My $0.02.