From the course: Complete Guide to Parallel and Concurrent Programming with C++

Unlock this course with a free trial

Join today to access over 24,900 courses taught by industry experts.

Speedup, latency, and throughput

Speedup, latency, and throughput - C++ Tutorial

From the course: Complete Guide to Parallel and Concurrent Programming with C++

Start my 1-month free trial Buy for my team

Speedup, latency, and throughput

“

- There are several reasons for using multiple processors to execute a program in parallel. One reason might be to increase the size of the problem you can tackle in a certain amount of time. For example, we're going to a party and I promise to bring 10 cupcakes. Working by myself, I can decorate 10 cupcakes in one hour. They're very fancy cupcakes. But if Baron joins me as a second processor doing the same type of work in parallel, together we can decorate 20 cupcakes in one hour. This type of parallelization is called weak scaling. We're keeping the size of the problem for each processor constant, but we're bringing in more processors to accomplish more work in the same amount of time. - Another reason for parallelization and bringing in more processors is to accomplish a given task faster. If Olivia promised to bring 10 cupcakes to the party, then working alone, it would take her one hour to decorate all of them. But if we split the workload, so she'll do half and I'll do half…

Contents