Case study - hardware performance
In this project, we will try to predict how much time it will take to multiply two matrices on different computers.
The dataset for this project originally comes from the paper Automatic selection of the fastest algorithm implementation by Sidnev and Gergel (2014), and it was made available at a machine learning competition organized by Mail.RU. You can check the details at http://mlbootcamp.ru/championship/7/.
Note
The content is in Russian, so if you do not speak it, it is better to use a browser with translation support.
You will find a copy of the dataset along with the code for this chapter.
This dataset has the following data:
m
,k
, andĂ‚Ân
Ă‚Â represent the dimensionality of the matrices, withm*k
being the dimensionality of matrixA
andk*n
Ă‚Â being the dimensionality of matrixB
- Hardware characteristics such as CPU speed, number of cores, whether hyper-threating is enabled or not, and the type of CPU
- The operation system
The solution for this problem can be...