# Pierre Hires an Assistant  URL:: https://brilliant.org/courses/computer-science-essentials/computational-problem-solving-2/parallelism/3 Author:: brilliant.org ## Highlights > One way to utilize parallelism is by splitting up a single task into several similar tasks that can be dealt with independently. ([View Highlight](https://read.readwise.io/read/01g74fjhzn24axsbkra6wzv8nz)) > When multiple workers can split up a problem without adding in coordination overhead, computer scientists sometimes call the problem **embarrassingly parallel**. ([View Highlight](https://read.readwise.io/read/01g74fjz9sxqqtcx23cnan3fxa)) > A physics simulation can be parallelized, but it wouldn't be correct to call it embarrassingly parallel. ([View Highlight](https://read.readwise.io/read/01g74fk07dztq92bvwhx41rh8b)) > Everyone from computer scientists to bakers and architects can fall into the trap of thinking that they can speed something up by throwing more workers into the mix. But that isn't true. When you think about parallel and non-parallel tasks, it should be obvious that adding more workers will only make you faster at the parts of a task that can be done in parallel. ([View Highlight](https://read.readwise.io/read/01g74fkg35rnk82rz97ydh1wnv)) > This chart summarizes that observation — with many workers, most of the time in a task is spent on the non-parallelizable parts of it: ([View Highlight](https://read.readwise.io/read/01g74fkt58gvqjmm3avj7wqy34)) > This common-sense observation is important enough in computer science that it has a name, [Amdahl's law](https://en.wikipedia.org/wiki/Amdahl%27s_law), after Gene Amdahl who wrote a paper about it in 1967. ([View Highlight](https://read.readwise.io/read/01g74fkvcm5aqnxe12fqjp6h8b)) --- Title: Pierre Hires an Assistant Author: brilliant.org Tags: readwise, articles date: 2024-01-30 --- # Pierre Hires an Assistant  URL:: https://brilliant.org/courses/computer-science-essentials/computational-problem-solving-2/parallelism/3 Author:: brilliant.org ## AI-Generated Summary What about a physics simulation that uses the laws of physics to understand how gravity affects the motion of all the stars in a galaxy? ## Highlights > One way to utilize parallelism is by splitting up a single task into several similar tasks that can be dealt with independently. ([View Highlight](https://read.readwise.io/read/01g74fjhzn24axsbkra6wzv8nz)) > When multiple workers can split up a problem without adding in coordination overhead, computer scientists sometimes call the problem **embarrassingly parallel**. ([View Highlight](https://read.readwise.io/read/01g74fjz9sxqqtcx23cnan3fxa)) > A physics simulation can be parallelized, but it wouldn't be correct to call it embarrassingly parallel. ([View Highlight](https://read.readwise.io/read/01g74fk07dztq92bvwhx41rh8b)) > Everyone from computer scientists to bakers and architects can fall into the trap of thinking that they can speed something up by throwing more workers into the mix. But that isn't true. When you think about parallel and non-parallel tasks, it should be obvious that adding more workers will only make you faster at the parts of a task that can be done in parallel. ([View Highlight](https://read.readwise.io/read/01g74fkg35rnk82rz97ydh1wnv)) > This chart summarizes that observation — with many workers, most of the time in a task is spent on the non-parallelizable parts of it: ([View Highlight](https://read.readwise.io/read/01g74fkt58gvqjmm3avj7wqy34)) > This common-sense observation is important enough in computer science that it has a name, [Amdahl's law](https://en.wikipedia.org/wiki/Amdahl%27s_law), after Gene Amdahl who wrote a paper about it in 1967. ([View Highlight](https://read.readwise.io/read/01g74fkvcm5aqnxe12fqjp6h8b))