Benchmark: (ColorfulCakeChen) CNN FusedConv (285226079309c705d1eaef149ddee0b61a5f6f94) 19

(ColorfulCakeChen) CNN FusedConv (285226079309c705d1eaef149ddee0b61a5f6f94) 19 (version: 19)

Comparing performance of: Async wait for init (WASM) vs (WASM) depthwise_bias_pointwise_bias vs (WASM) depthwise_pointwise_bias vs (WASM) fusedConv_bias vs (WASM) fusedConvWithBias vs Async wait for init (WebGL) vs (WebGL) depthwise_bias_pointwise_bias vs (WebGL) depthwise_pointwise_bias vs (WebGL) fusedConv_bias vs (WebGL) fusedConvWithBias vs Async wait for init (CPU) vs (CPU) depthwise_bias_pointwise_bias vs (CPU) depthwise_pointwise_bias vs (CPU) fusedConv_bias vs (CPU) fusedConvWithBias

Created: 2 years ago by: Registered User

Jump to the latest result

HTML Preparation code:

<script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@4.8.0/dist/tf.min.js"></script>
<script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs-backend-wasm/dist/tf-backend-wasm.js"></script>

Script Preparation code:

globalThis.backendName_testing = undefined;
globalThis.jsPerf_Module = undefined;
globalThis.testCorrectnessIterator = undefined;

async function init( backendName ) {
  if ( !globalThis.jsPerf_Module ) {
    console.log( "import jsPerf_FusedConv..." );
    globalThis.jsPerf_Module = await import(
      "https://cdn.jsdelivr.net/gh/colorfulcakechen/query-submit-canvas@285226079309c705d1eaef149ddee0b61a5f6f94/CNN/jsPerf/jsPerf_FusedConv.js" );
  }

  if ( backendName_testing != backendName ) {
    console.log( `Change backend to "${backendName}"...` );
    globalThis.backendName_testing = backendName;

    globalThis.jsPerf_Module.disposeResources();

    await tf.ready();
    await tf.setBackend( backendName );

    globalThis.jsPerf_Module.init();
  	globalThis.testCorrectnessIterator = jsPerf_Module.testCorrectness();
  }

  if ( !globalThis.testCorrectnessIterator )
    return true;

  let testCorrectnessYield = globalThis.testCorrectnessIterator.next();
  if ( testCorrectnessYield.done ) {
    console.log( `${backendName}, initialization done.` );
    globalThis.testCorrectnessIterator = null;
    return true;
  } else {
	return false;
  }
}

Tests:

Async wait for init (WASM)
if ( ( globalThis.initPromiseWASM ) && ( globalThis.initPromiseWebGL ) && ( globalThis.initPromiseCPU ) ) { // For benchmark re-run. globalThis.initPromiseWASM = globalThis.initPromiseWebGL = globalThis.initPromiseCPU = null; } const backendName = "wasm"; //const backendName = "webgl"; //const backendName = "cpu"; if ( !globalThis.initPromiseWASM ) { globalThis.initPromiseWASM = init( backendName ); } else { globalThis.initPromiseWASM.then( ( done ) => { if ( !done ) { globalThis.initPromiseWASM = init( backendName ); } else { } }); }
(WASM) depthwise_bias_pointwise_bias
globalThis.testSet.test_depthwise_bias_pointwise_bias();
(WASM) depthwise_pointwise_bias
globalThis.testSet.test_depthwise_pointwise_bias();
(WASM) fusedConv_bias
globalThis.testSet.test_fusedConv_bias();
(WASM) fusedConvWithBias
globalThis.testSet.test_fusedConvWithBias();
Async wait for init (WebGL)
const backendName = "webgl"; if ( !globalThis.initPromiseWebGL ) { globalThis.initPromiseWebGL = init( backendName ); } else { globalThis.initPromiseWebGL.then( ( done ) => { if ( !done ) { globalThis.initPromiseWebGL = init( backendName ); } else { } }); }
(WebGL) depthwise_bias_pointwise_bias
globalThis.testSet.test_depthwise_bias_pointwise_bias();
(WebGL) depthwise_pointwise_bias
globalThis.testSet.test_depthwise_pointwise_bias();
(WebGL) fusedConv_bias
globalThis.testSet.test_fusedConv_bias();
(WebGL) fusedConvWithBias
globalThis.testSet.test_fusedConvWithBias();
Async wait for init (CPU)
const backendName = "cpu"; if ( !globalThis.initPromiseCPU ) { globalThis.initPromiseCPU = init( backendName ); } else { globalThis.initPromiseCPU.then( ( done ) => { if ( !done ) { globalThis.initPromiseCPU = init( backendName ); } else { } }); }
(CPU) depthwise_bias_pointwise_bias
globalThis.testSet.test_depthwise_bias_pointwise_bias();
(CPU) depthwise_pointwise_bias
globalThis.testSet.test_depthwise_pointwise_bias();
(CPU) fusedConv_bias
globalThis.testSet.test_fusedConv_bias();
(CPU) fusedConvWithBias
globalThis.testSet.test_fusedConvWithBias();

Rendered benchmark preparation results:

Suite status: <idle, ready to run>

Previous results

Test case name	Result
Async wait for init (WASM)
(WASM) depthwise_bias_pointwise_bias
(WASM) depthwise_pointwise_bias
(WASM) fusedConv_bias
(WASM) fusedConvWithBias
Async wait for init (WebGL)
(WebGL) depthwise_bias_pointwise_bias
(WebGL) depthwise_pointwise_bias
(WebGL) fusedConv_bias
(WebGL) fusedConvWithBias
Async wait for init (CPU)
(CPU) depthwise_bias_pointwise_bias
(CPU) depthwise_pointwise_bias
(CPU) fusedConv_bias
(CPU) fusedConvWithBias

Fastest: N/A

Slowest: N/A

Latest run results:

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

User agent: Mozilla/5.0 (X11; CrOS x86_64 14541.0.0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/114.0.0.0 Safari/537.36

Browser/OS: Chrome 114 on Chrome OS 14541.0.0

View result in a separate tab

Test name	Executions per second
Async wait for init (WASM)	228412.9 Ops/sec
(WASM) depthwise_bias_pointwise_bias	721.6 Ops/sec
(WASM) depthwise_pointwise_bias	967.8 Ops/sec
(WASM) fusedConv_bias	112.1 Ops/sec
(WASM) fusedConvWithBias	138.8 Ops/sec
Async wait for init (WebGL)	915179.4 Ops/sec
(WebGL) depthwise_bias_pointwise_bias	1505.8 Ops/sec
(WebGL) depthwise_pointwise_bias	1776.0 Ops/sec
(WebGL) fusedConv_bias	527.7 Ops/sec
(WebGL) fusedConvWithBias	427.3 Ops/sec
Async wait for init (CPU)	967705.7 Ops/sec
(CPU) depthwise_bias_pointwise_bias	68.6 Ops/sec
(CPU) depthwise_pointwise_bias	65.9 Ops/sec
(CPU) fusedConv_bias	6.6 Ops/sec
(CPU) fusedConvWithBias	8.4 Ops/sec

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

LLMs can make mistakes. Check important info.

Let's dive into the data!

After analyzing the data, I noticed that the test names suggest that we're dealing with WebAssembly (WASM) and WebGL tests. The test names also indicate that we have multiple types of bias tests being executed: depthwise, pointwise, fusedConv_withBias, and fusedConv_bias.

Here are some observations:

1. **DevicePlatform**: Most tests are running on Desktop Chrome OS devices.
2. **Browser**: All tests are running with Chrome 114.
3. **OperatingSystem**: Chrome OS 14541.0.0 is the most common operating system used for these tests.
4. **ExecutionsPerSecond**: The highest number of executions per second is around 527-540, indicating that some tests run relatively fast.

Some questions come to mind:

* What is the purpose of these bias tests?
* Are these tests related to machine learning or computer graphics?
* Do these test results have any implications for WebGL or WASM performance?

If you'd like to provide more context or information about these tests, I'd be happy to help further!

Related benchmarks:

(ColorfulCakeChen) CNN FusedConv (285226079309c705d1eaef149ddee0b61a5f6f94) 19 (version: 19)

Created: 2 years ago by: Registered User

Jump to the latest result

Async wait for init (WASM)

(WASM) depthwise_bias_pointwise_bias

(WASM) depthwise_pointwise_bias

(WASM) fusedConv_bias

(WASM) fusedConvWithBias

Async wait for init (WebGL)

(WebGL) depthwise_bias_pointwise_bias

(WebGL) depthwise_pointwise_bias

(WebGL) fusedConv_bias

(WebGL) fusedConvWithBias

Async wait for init (CPU)

(CPU) depthwise_bias_pointwise_bias

(CPU) depthwise_pointwise_bias

(CPU) fusedConv_bias

(CPU) fusedConvWithBias

Suite status: <idle, ready to run>

Experimental features:

Fastest: N/A

Slowest: N/A

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