View Single Post
Posts: 1,808 | Thanked: 4,272 times | Joined on Feb 2011 @ Germany
#11
Originally Posted by pichlo View Post
Sorry in advance about the verbosity. I am typing this in a hurry. It might have found a more concise way of expressing myself if I had more time.
I found it just fine. Interesting observations!

Yes, that is the most common assumption. The whole American (and, since we monkey everything after the Americans, European too) production system is based on that. However, there was once a company that stopped assuming and started measuring. They found that, perhaps counter-intuitively, utilizing your production line close to full capacity is detrimental to its throughput. Any small variation at any link in the chain causes huge dropouts in productivity. Their findings were one of the bases for the Toyota Production System, or TPS as it has become commonly known.
When I was a child my father installed a game on our PC. It was made by some economist. You had a number of "production lines" (it was all a text-mode GUI) and you had to fiddle with some parameters, with the goal of maximizing production.

Without really understanding what the whole thing was about, I just played it like a game (most kids don't need instructions, they just find a way to, "evolutionarily", play better each time). After a few days I went to my father and said "that's weird, in order to get a better result, I have to stress a single line, where I would have thought that I need to distribute the load equally" (being a 10 year old, I used different wording, but the message was clear). His reply was that I was doing the right thing, and then showed me the book accompanying the software (no, we could not afford "cool" games) and there it was all explained.

Obviously *you* are right. Nothing is perfect, and the above only applies when no "slack" is needed, i.e. when everything happens deterministically and with no possibility of random delays or defects.

Still, in the particular case of the N900, or any computer for that matter, it seems counter-intuitive to "waste" resources (e.g. killing background programs to save memory, even if you will periodically need to start those programs again; moving stuff away from rootfs when you don't actually need that space for anything, etc.)