Very likely the slowest speed. I have a couple 60ns SPD rated 72-pin sticks with all 50ns chips next to 60ns parity.

Seems like a waste of more expensive chips. Has anyone tested this?

I've done some simple tests in which I know a particular motherboard at 40 MHz FSB needs 60 ns chips because when I use 70 ns with 0WS, it will get general protection faults in Win 3.11. When I put 60 ns chips back in, even though 2 of them have 70 ns parity, the system works.

However, I did not ensure that those two chips (out of 8 total) with the 70 ns parity were in the front of the line. They may have been in the back where Windows hadn't used those modules yet.

I ran the test again, with less RAM (16 MB) and had the two sticks with 70 ns parity in the first slots. I ran IE3 and Netscape 2 for a bit, and no crashes. When using 70 ns RAM, the system crashes just after Windows 3.11 boot. For the 70 ns RAM, I need to add 1 ws in the BIOS, for 60 ns, I do not.

Does this mean that 60 ns RAM with 70 ns parity is qualified for 60 ns? I still do not know.

Does this mean that 60 ns RAM with 70 ns parity is qualified for 60 ns? I still do not know.

Yes, it does. The parity chip can be slower due to the way it is accessed.

I had a 286 / 12.5 MHz running for years with 80ns RAM and 100ns Parity with no waitstate.

Certainly it's going to be a lot more stable at 60ns than an all-70ns stick, but the question as I see it is whether a slower parity chip can officially rate as faster memory. I think most motherboards are very loose on memory for stability's sake, like in Phil's recent test with a couple speed grades that found no difference.