Julia High Performance by Avik Sengupta

Author:Avik Sengupta
Language: eng
Format: epub, pdf
Tags: COM051000 - COMPUTERS / Programming / General, COM051010 - COMPUTERS / Programming Languages / General, COM042000 - COMPUTERS / Natural Language Processing
Publisher: Packt Publishing
Published: 2019-06-07T06:34:27+00:00

We can then benchmark these functions with and without forcing subnormal numbers to zero. Take a look at the following:

julia> set_zero_subnormals(false)

true

julia> t=rand(1000);

julia> @btime heatflow($t, 1000)

1.559 ms (1 allocation: 7.94 KiB)

julia> set_zero_subnormals(true)

true

julia> @btime heatflow($t, 1000)

1.062 ms (1 allocation: 7.94 KiB)

We can see a significant increase in speed by forcing subnormal numbers to zero.

Summary

In this chapter, we discussed how Julia uses a machine representation of numbers to achieve a C-like performance for its arithmetic computations. We noted how to work within these design constraints, and considered the edge cases that are introduced.

Working with single numbers, however, is the easy part. Most numerical computations, as we noted throughout this chapter, consist of working on large sets of numbers. In the next chapter, we will take a look at how to work with arrays in a performant manner.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.