Benchmark: xoshiro128** various implementations

Script Preparation code:

class Xoshiro128StarStar {
	constructor(a, b, c, d) {
		this.a = a;
		this.b = b;
		this.c = c;
		this.d = d;
	}

	next() {
		const t = this.b << 9;
		let r = this.b * 5;
		r = ((r << 7) | (r >>> 25)) * 9;

		this.c ^= this.a;
		this.d ^= this.b;
		this.b ^= this.c;
		this.a ^= this.d;
		this.c ^= t;
		this.d = (this.d << 11) | (this.d >>> 21);

		return r >>> 0; // Ensure unsigned integer
	}

}
class Xoshiro128StarStar2 {
	constructor(a, b, c, d) {
		this.a = a;
		this.b = b;
		this.c = c;
		this.d = d;
	}

	next() {
		const s0 = this.a;
		const s1 = this.b;
		const s2 = this.c;
		const s3 = this.d;

		const x = s1 * 5;
		const r = (((x << 7) | (x >>> 25)) * 9) >>> 0;

		const t2 = s2 ^ s0;
		const t3 = s3 ^ s1;

		this.b = s1 ^ t2;
		this.a = s0 ^ t3;
		this.c = t2 ^ (s1 << 9);
		this.d = (t3 << 11) | (t3 >>> 21);

		return r;
	}
}

rngA = new Xoshiro128StarStar(3536, 4678, 2245, 778);
rngB = new Xoshiro128StarStar2(3536, 4678, 2245, 778);

Tests:

Xoshiro128StarStar
for(let i = 0; i < 100; i++) { const z = rngA.next(); }
Xoshiro128StarStar2
for(let i = 0; i < 100; i++) { const z = rngB.next(); }

Rendered benchmark preparation results:

Suite status: <idle, ready to run>

Previous results

Test case name	Result
Xoshiro128StarStar
Xoshiro128StarStar2

Fastest: N/A

Slowest: N/A

Latest run results:

Run details: (Test run date: one year ago)

User agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/137.0.0.0 Safari/537.36

Browser/OS: Chrome 137 on Windows

View result in a separate tab

Xoshiro128StarStar2

1.5M/s

Xoshiro128StarStar

771K/s

View exact numbers

Test name	Executions per second
✓ Xoshiro128StarStar2	1,467,687 Ops/sec
Xoshiro128StarStar	770,711 Ops/sec

Autogenerated LLM Summary (model llama3.2:3b, generated one year ago):

LLMs can make mistakes. Check important info.

Let's dive into the explanation.

**Benchmark Definition JSON**

The provided benchmark definition is for testing two different implementations of the Xoshiro 128* StarStar (XRNG) random number generator, which is a widely used and high-quality pseudorandom number generator.

**Script Preparation Code**

The script preparation code defines two classes: `Xoshiro128StarStar` and `Xoshiro128StarStar2`. These classes are the core of the XRNG algorithm. The main difference between them lies in how they generate the next random value (`next()` function). Here's a brief overview:

1.  **`Xoshiro128StarStar`**:

*   This implementation uses bitwise operations to combine and modify its internal state (`a`, `b`, `c`, and `d`) in each iteration.
    *   The formula used is similar to the one described in the paper by Makoto Matsumoto (2009) for Xoshiro 128* StarStar.

2.  **`Xoshiro128StarStar2`**:

*   This implementation also uses bitwise operations, but with some differences from `Xoshiro128StarStar`.
    *   The formula used is similar to the one described in the paper by Makoto Matsumoto (2009) for Xoshiro 128* StarStar, but with a different initial state (`s0`, `s1`, `s2`, and `s3`) and how they are combined.

**Options Compared**

The benchmark compares two options:

1.  **`Xoshiro128StarStar`**: This implementation uses the original formula for generating random numbers, which is considered to be one of the most efficient and reliable ways to generate high-quality pseudorandom numbers.
2.  **`Xoshiro128StarStar2`**: This implementation modifies the initial state (`s0`, `s1`, `s2`, and `s3`) and how they are combined, which may affect its performance and quality.

**Pros and Cons**

Here are some pros and cons of each approach:

*   **`Xoshiro128StarStar`**:
    *   Pros: It is widely used, known for its high-quality random numbers, and has a well-established reputation.
    *   Cons: The formula can be complex to understand and implement.

*   **`Xoshiro128StarStar2`**:
    *   Pros: This implementation may offer some performance improvements or slight variations in quality compared to the original `Xoshiro128StarStar`, depending on the specific use case.
    *   Cons: The modifications to the initial state and formula may affect its overall quality, making it less suitable for all applications.

**Other Considerations**

When choosing an implementation of XRNG, consider the following factors:

1.  **Performance**: If speed is a priority, `Xoshiro128StarStar2` might be faster due to simpler bitwise operations.
2.  **Quality**: For most applications requiring high-quality random numbers, `Xoshiro128StarStar` is still the preferred choice.

**Alternatives**

Other alternatives for generating pseudorandom numbers include:

1.  **Linear Congruential Generators (LCGs)**: These are simple and widely used but often have a lower quality compared to XRNGs.
2.  **Mersenne Twisters**: These offer better randomness properties than LCGs, but can be slower.
3.  **Fortuna PRNGs**: These are designed for high-quality random numbers and cryptographic applications.

In conclusion, the choice between `Xoshiro128StarStar` and `Xoshiro128StarStar2` depends on specific requirements, such as performance or quality of randomness needed.

xoshiro128** various implementations (version: 0)

Comparing performance of: `Xoshiro128StarStar vs Xoshiro128StarStar2`

Created: 2 years ago by: Registered User

Go to the latest result

Xoshiro128StarStar

Xoshiro128StarStar2

Suite status: <idle, ready to run>

Fastest: N/A

Slowest: N/A

Autogenerated LLM Summary (model llama3.2:3b, generated one year ago):

xoshiro128** various implementations (version: 0)

Comparing performance of: Xoshiro128StarStar vs Xoshiro128StarStar2

Created: 2 years ago by: Registered User

Go to the latest result

Xoshiro128StarStar

Xoshiro128StarStar2

Suite status: <idle, ready to run>

Fastest: N/A

Slowest: N/A

Autogenerated LLM Summary (model llama3.2:3b, generated one year ago):

Related benchmarks

Comparing performance of: `Xoshiro128StarStar vs Xoshiro128StarStar2`