What defines latency in PC memory?
“Latency” in computer terms is how long it takes to get a request back. It’s the flip side of the coin from “bandwidth”, which is how fast the data flows. If you want to buy a movie, you can either download it or buy the DVD from Amazon. The download has good latency: you get a response to your request immediately, but the bandwidth sucks. (Or at least, it did before everybody got broadband.) Buying the DVD has great bandwidth (several gigabytes in just a few days, way faster than a modem) but the latency sucks. Good for downloading movies. Bad for surfing the web, despite the awesome bandwidth. The same thing applies up and down the computer scale. Making a request to RAM isn’t instantaneous. The electronic signal has to travel up to a few inches to get to the RAM chips, through the chips, and then back before you can use the memory. It may not seem like much, but compared to the onboard cache, it’s forever. Several nanoseconds during which your computer is just twiddling its thumbs.
