Just for grins I looked up the datasheet for the 27C512 to make sure and yeah, it doesn't have any "data good" signal or anything else that would allow a computer to "clock faster" with, say, a 100ns part vs. a 150ns one. It has a "Chip Enable" signal which tells the chip to "wake up" and look at its address lines for the location of the byte you're requesting, and an "Output Enable" which tells it to actually shove the byte out its data lines. (It has this so a computer designer can dispense with the need to have a protective buffer in front of the device to keep it from fouling the bus while it's "thinking".) If the computer it's plugged into brings up OE too quickly and tries to use a value read from it before the output has stabilized then the value is likely to be garbage, and the part has no way to tell the host it's doing it. So, yeah. No matter how fast a part you stick in there it's going to run at the speed of the original unless you make changes to the circuitry of the computer to reduce the delay/increase the clock speed/whatever.

Anyway. If it was broken before and works now it runs infinitely faster. So you're still right!