In class Kayvon said that Amdahl's Law is more like Newton's laws of motion and some people don't think it's any law at all. I was wondering why that was. I can see that it obviously doesn't take synchronization or communication overhead between the threads into account, so is that why? I'm sure there are other reasons as well.
This comment was marked helpful 0 times.
kayvonf
@jmnash: Sorry to disappoint, but it was a joke. I commented that "Amdahl's Law" despite its name, is more a simple fact than a deep fundamental truth about the world or computation in general. Physicists and theorists would certainly say that's a prerequisite for something to be called a "law". Systems folks have a considerably lower bar :-) ... leading to observations like Amdahl's Law and Moore's Law being handed names that ironically echo the great principles of the world.
In class Kayvon said that Amdahl's Law is more like Newton's laws of motion and some people don't think it's any law at all. I was wondering why that was. I can see that it obviously doesn't take synchronization or communication overhead between the threads into account, so is that why? I'm sure there are other reasons as well.
This comment was marked helpful 0 times.
@jmnash: Sorry to disappoint, but it was a joke. I commented that "Amdahl's Law" despite its name, is more a simple fact than a deep fundamental truth about the world or computation in general. Physicists and theorists would certainly say that's a prerequisite for something to be called a "law". Systems folks have a considerably lower bar :-) ... leading to observations like Amdahl's Law and Moore's Law being handed names that ironically echo the great principles of the world.
This comment was marked helpful 1 times.