Re:Amdahl's Law
by kache_ on Thursday March 25, 2021 @09:45PM
(#26587954)
This concept generalizes really well to many things outside of computing. For example, this is why adding more engineers to a team can slow things down.
[ Parent
| on HN ]
Amdahl's Law says that the marginal benefit of additional cores is always positive (or zero).
What you're describing is negative marginal benefit.
Communication overhead.
Mostly that, but I would add cultural and competence issues as well - more internal politics, and C-teamers sneaking in that drag the team down.
It's easy to make sure everyone in a small team is elite and has value alignment, and it's hard to maintain that as it gets bigger.
Re:Amdahl's Law
by loloquwowndueo on Thursday March 25, 2021 @11:00PM
(#26588351)
That’s more of a dispelled myth and not Amdahl's law.
https://en.m.wikipedia.org/wiki/The_Mythical_Man-Month Specifically :)
A surprising number of managers haven't got the memo about the "dispelled myth" part.
Re:Amdahl's Law
by anonytrary on Friday March 26, 2021 @04:11AM
(#26589801)
Amdahl's law is more simplistic than that. You're suggesting that the benefit of adding an additional worker is negative. Amdahl's law is not representative of most systems; your generalization seems better. The marginal benefit as a function of resources depends on the system in question. It isn't just a flat positive percentage.
Re:Amdahl's Law
by vegetablepotpie on Thursday March 25, 2021 @09:59PM
(#26588040)
I’m not so sure of that. Amdahls law tells you the speed up to expect when adding more cores to the parallelized portion of an algorithm. Whereas adding a new member to a team slows down one or more members for a matter of weeks/months because the members are spending time they would otherwise spend producing teaching the new member.