How Swift Compile Optimazation Level works

Optimization Level

Swift compiler has 3 optimization options (link)

-O
- for product code
-Onone
- for development and debugging mode.
-Osize
- for size optimization over the optimizing code

How to compile swift with optimization level?

Here is a simple swift code for checking execution time. (main.swift)

import Foundation
func sum(from: Int, to: Int) -> Int {
    var result = 0
    for i in from...to {
        result += i
    }
    return result
}

let startTime = Date()
let result = sum(from: 1, to: 10000000)
let endTime = Date()

print("Result: \(result)")
print("Calculation Time: \(endTime.timeIntervalSince(startTime)) seconds")

Compiled it with -Onone

swiftc -Onone main.swift && ./main

//Result
Result: 50000005000000
Time: 1.4896910190582275 seconds

Compiled it with -O

swiftc -O main.swift && ./main

//Result
Result: 50000005000000
Time: 0.006083011627197266 seconds

Results

-O option is 24,389% faster than -Onone option

Check SIL file

swiftc -O main.swift is generating executable file. It is not a human readable file. The optimization process transforms the source code into machine code, which is the binary code that can be executed directly by the computer.

Do you want to see the intermediate representation of the Swift code generated during the optimization process?

You can use the -emit-sil flag with the swiftc command. This will output the SIL (Swift Intermediate Language) code generated by the compiler, which is a human-readable representation of the Swift code that has been optimized

swiftc -O -emit-sil main.swift -o main.sil

swiftc -Onone -emit-sil main.swift -o main.sil

You can see the main.sil file. Let’s open it using TextEdit

Now we can see the details by checking sil file. If you want to learn more about it visit here

Conclusion

I wanted to test how Swift’s optimization levels work, so I used the -emit-sil option to generate an SIL (Swift Intermediate Language) file. This file serves as a bridge between the Swift source code and the machine code, and allows the Swift compiler to perform optimizations such as dead code elimination, constant folding, inlining, and loop unrolling. By inspecting and analyzing the SIL file, we can better understand how the Swift compiler is optimizing our code. Additionally, SIL can be used for debugging purposes since it provides a more detailed and low-level view of the generated code compared to the original Swift source

Notes
This blog post has been edited using ChatGPT to refine some of the sentences

Shawn