Benchmark: shaders strings comparisons with Set has()

Script Preparation code:

var a = `
struct Matrices {
	model: mat4x4f,
	world: mat4x4f,
	modelView: mat4x4f,
	modelViewProjection: mat4x4f
};

fn getVertex2DToUVCoords(vertex: vec2f) -> vec2f {
  return vec2(
    vertex.x * 0.5 + 0.5,
    0.5 - vertex.y * 0.5
  );
}

fn getVertex3DToUVCoords(vertex: vec3f) -> vec2f {
  return vec2(
    vertex.x * 0.5 + 0.5,
    0.5 - vertex.y * 0.5
  );
}


fn getUVCover(uv: vec2f, textureMatrix: mat4x4f) -> vec2f {
  return (textureMatrix * vec4f(uv, 0.0, 1.0)).xy;
}

@group(1) @binding(0) var<uniform> matrices: Matrices;

struct VSOutput {
  @builtin(position) position: vec4f,
  @location(0) uv: vec2f,
  @location(1) normal: vec3f,
};

@fragment fn main(fsInput: VSOutput) -> @location(0) vec4f {
  // normals
  return vec4(normalize(fsInput.normal) * 0.5 + 0.5, 1.0);
}`

var b = `
struct Matrices {
	model: mat4x4f,
	world: mat4x4f,
	modelView: mat4x4f,
	modelViewProjection: mat4x4f
};

fn getVertex2DToUVCoords(vertex: vec2f) -> vec2f {
  return vec2(
    vertex.x * 0.5 + 0.5,
    0.5 - vertex.y * 0.5
  );
}

fn getVertex3DToUVCoords(vertex: vec3f) -> vec2f {
  return vec2(
    vertex.x * 0.5 + 0.5,
    0.5 - vertex.y * 0.5
  );
}


fn getUVCover(uv: vec2f, textureMatrix: mat4x4f) -> vec2f {
  return (textureMatrix * vec4f(uv, 0.0, 1.0)).xy;
}

@group(1) @binding(0) var<uniform> matrices: Matrices;

struct VSOutput {
  @builtin(position) position: vec4f,
  @location(0) uv: vec2f,
  @location(1) normal: vec3f,
};

@fragment fn main(fsInput: VSOutput) -> @location(0) vec4f {
  // normals
  return vec4(normalize(fsInput.normal) * 0.5 + 0.5, 1.0);
}`

var c = `struct Attributes {
	@builtin(vertex_index) vertexIndex : u32,
	@builtin(instance_index) instanceIndex : u32,
	@location(0) position: vec3f,
	@location(1) uv: vec2f,
	@location(2) normal: vec3f
};

struct Matrices {
	model: mat4x4f,
	world: mat4x4f,
	modelView: mat4x4f,
	modelViewProjection: mat4x4f
};

struct Camera {
	model: mat4x4f,
	view: mat4x4f,
	projection: mat4x4f
};

fn getOutputPosition(position: vec3f) -> vec4f {
  return matrices.modelViewProjection * vec4f(position, 1.0);
}

fn getUVCover(uv: vec2f, textureMatrix: mat4x4f) -> vec2f {
  return (textureMatrix * vec4f(uv, 0.0, 1.0)).xy;
}

@group(0) @binding(0) var<uniform> camera: Camera;

@group(1) @binding(0) var<uniform> matrices: Matrices;

struct VertexOutput {
  @builtin(position) position: vec4f,
  @location(0) uv: vec2f,
  @location(1) normal: vec3f,
};

@vertex fn main(
  attributes: Attributes,
) -> VertexOutput {
  var vsOutput: VertexOutput;

  vsOutput.position = getOutputPosition(attributes.position);
  vsOutput.uv = attributes.uv;
  vsOutput.normal = attributes.normal;
  
  return vsOutput;
}`

var d = new Set([a, b, c])

Tests:

a, b equality
a === b
a, b locale compare
a.localeCompare(b) === 0
a, c equality
a === c
a, c locale compare
a.localeCompare(c) === 0
d has a
d.has(a)
d has b
d.has(b)
d has c
d.has(c)
d has something not in d
d.has('test')

Rendered benchmark preparation results:

Suite status: <idle, ready to run>

Previous results

Test case name	Result
a, b equality
a, b locale compare
a, c equality
a, c locale compare
d has a
d has b
d has c
d has something not in d

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/126.0.0.0 Safari/537.36

Browser/OS: Chrome 126 on Windows

View result in a separate tab

Test name	Executions per second
a, b equality	15968287.0 Ops/sec
a, b locale compare	15832560.0 Ops/sec
a, c equality	15952454.0 Ops/sec
a, c locale compare	15292805.0 Ops/sec
d has a	15293982.0 Ops/sec
d has b	15454039.0 Ops/sec
d has c	15493578.0 Ops/sec
d has something not in d	30871632.0 Ops/sec

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

LLMs can make mistakes. Check important info.

The test results show the performance of various benchmarking tests on different browsers and devices. The tests appear to be checking for the presence or equality of strings in an object `d`.

Here's a summary of the top-performing results:

1. **TestName: "d has something not in d"** (Chrome 126, Desktop, Windows): 30871632 executions/second
This test checks if the object `d` contains a string that is not present in it, which suggests that the browser is handling the lookup efficiently.
2. **TestName: "a, b equality"** (Chrome 126, Desktop, Windows): 15968287 executions/second
This test checks for equality between two strings `a` and `b`. The performance suggests that the browser's string comparison is optimized.
3. **TestName: "d has a"** (Chrome 126, Desktop, Windows): 15454039 executions/second
This test simply checks if the object `d` contains the string `a`. The high execution count indicates fast lookup performance.

The tests with lower performance seem to involve more complex comparisons or lookups. For example:

* **TestName: "d has c"** (Chrome 126, Desktop, Windows): 15493578 executions/second
This test also checks for the presence of a string `c` in object `d`, but with slightly slower execution count.
* **TestName: "a, b locale compare"** (Chrome 126, Desktop, Windows): 15293982 executions/second
This test compares two strings `a` and `b` using locale-aware comparison, which may involve additional complexity.

Keep in mind that these results are based on a single data point from the latest benchmark run. Other runs or different configurations might yield different performance characteristics.

Related benchmarks:

shaders strings comparisons with Set has() (version: 0)

Comparing performance of: a, b equality vs a, b locale compare vs a, c equality vs a, c locale compare vs d has a vs d has b vs d has c vs d has something not in d

Created: 2 years ago by: Guest

Jump to the latest result

a, b equality

a, b locale compare

a, c equality

a, c locale compare

d has a

d has b

d has c

d has something not in d

Suite status: <idle, ready to run>

Fastest: N/A

Slowest: N/A

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