Skip to content
This repository was archived by the owner on Nov 6, 2022. It is now read-only.
This repository was archived by the owner on Nov 6, 2022. It is now read-only.

PHP performance profiling #485

Open
Open

Description

@kienstra

It'd be good to have an idea of this plugin's effect on front-end performance.

This might also point out areas that we could improve. For example, if a certain function is taking too long.

Activity

  1. kienstra commented on Dec 2, 2019

    @kienstra
    CollaboratorAuthor

    Block Lab Compared To Implementing Your Own Dynamic Block

    What's the PHP performance cost of using Block Lab for 1 block, compared to implementing your own dynamic block?

    In the test below, there was an average cost of about 87 milliseconds to use Block Lab. For reference, this is slightly less than the cost of Gutenberg or Yoast.

    Here's a Blackfire profile of Block Lab:
    https://blackfire.io/profiles/e4978c69-4bf2-462e-a195-69fa81d93372/graph

    This test was for the public-facing front-end, not for the block editor.

    Methodology

    Create 2 posts, one with a Block Lab block, and one with an equivalent dynamic block that doesn't use Block Lab

    With a Block Lab block

    1. Activate Twenty Nineteen. Deactivate all plugins except Block Lab, and a plugin that adds the snippets below.

    2. Add the Block Lab block:

    add_action(
    	'block_lab_add_blocks',
    	function() {
    		block_lab_add_block(
    			'test-hero',
    			array(
    				'title'    => 'Test Hero',
    				'category' => 'common',
    				'fields'   => array(
    					'headline' => array(
    						'label'   => 'Headline',
    						'control' => 'text',
    					),
    					'copy'     => array(
    						'label'   => 'Copy',
    						'control' => 'textarea',
    					),
    					'url'      => array(
    						'label'   => 'URL',
    						'control' => 'url',
    					),
    				),
    			)
    		);
    	}
    );
    1. Then, add the template for it in the theme or plugin, at blocks/block-test-hero.php:
    <?php
    /**
     * Template for test-hero block.
     */
    
    ?>
    <h1><?php esc_html_e( 'Testing Block Lab dynamic block', 'bl-testing' ) ?></h1>
    <p><?php esc_html_e( 'Here is the headline attribute:', 'bl-testing' ) ?></p>
    <p><?php echo esc_html( block_value( 'headline' ) ); ?></p>
    <p><?php esc_html_e( 'And the copy attribute:', 'bl-testing' ) ?></p>
    <p><?php echo esc_html( block_value( 'copy' ) ); ?></p>
    <p><?php esc_html_e( 'And the url attribute:', 'bl-testing' ) ?></p>
    <p><?php echo esc_html( block_value( 'url' ) ); ?></p>
    
    1. Populate the block via WP-CLI:
    wp post create --post_title="With Block Lab" --post_status='publish' --post_content='<!-- wp:block-lab/test-hero {"headline":"This is an example headline", "copy":"Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.","url":"https://example.com/baz"} /-->'
    1. Go to the front-end of the post, and run 10 Blackfire profiles of the back-end performance (results below)

    block-lab-dynamic-block

    Without Block Lab

    Using a custom dynamic block, instead of Block Lab

    1. Deactivate Block Lab
    2. Register a dynamic block, like the block above:
    add_action(
    	'init',
    	function() {
    		register_block_type(
    			'bl-testing/hero',
    			[
    				'attributes'      => [
    					'headline' => [
    						'type' => 'string',
    					],
    					'copy'     => [
    						'type' => 'string',
    					],
    					'url'      => [
    						'type' => 'string',
    					],
    				],
    				'render_callback' => 'bl_testing_render_dynamic_block',
    			]
    		);
    	}
    );
    
    /**
     * The render callback function for the dynamic block.
     *
     * @param array $attributes The attributes to render.
     */
    function bl_testing_render_dynamic_block( $attributes ) {
    	$attributes = wp_parse_args(
    		$attributes,
    		[
    			'headline' => '',
    			'copy'     => '',
    			'url'      => '',
    		]
    	)
    
    	?>
    		<h1><?php esc_html_e( 'Testing dynamic block', 'bl-testing' ) ?></h1>
    		<p><?php esc_html_e( 'Here is the headline attribute:', 'bl-testing' ) ?></p>
    		<p><?php echo esc_html( $attributes['headline'] ) ?></p>
    		<p><?php esc_html_e( 'And the copy attribute:', 'bl-testing' ) ?></p>
    		<p><?php echo esc_html( $attributes['copy'] ) ?></p>
    		<p><?php esc_html_e( 'And the url attribute:', 'bl-testing' ) ?></p>
    		<p><?php echo esc_html( $attributes['url'] ) ?></p>
    	<?php
    }
    1. Populate that block with WP-CLI:
    wp post create --post_title="Without Block Lab" --post_status='publish' --post_content='<!-- wp:bl-testing/hero {"headline":"This is an example headline", "copy":"Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.","url":"https://example.com/baz"} /-->'
    1. Go to the front-end of the post, and run 10 Blackfire profiles of the back-end performance

    Performance Results

    Milliseconds that PHP took to produce the HTML document (in 10 trials)

    With Block Lab Without Block Lab
    274 183
    269 173
    254 173
    267 172
    256 173
    267 184
    254 174
    264 174
    253 173
    265 172

    Average of Trials

    With Block Lab Without Block Lab
    262.3 175.1

    Difference: 87.2 milliseconds

    What does this profile test?

    Only the time PHP took to generate the HTML document, not the browser parsing the document. Block Lab doesn't use JavaScript on the front-end, only in the block editor.

  2. lukecarbis commented on Dec 2, 2019

    @lukecarbis
    Member

    It would be interesting to run the same comparison, but with 10 instances of the block in the content.

    It would also be interesting to run the same comparison, but without any blocks in the content (just the Block Lab plugin overhead).

  3. kienstra commented on Dec 2, 2019

    @kienstra
    CollaboratorAuthor

    Great idea! I'll work on a script for that.

  4. kienstra commented on Dec 2, 2019

    @kienstra
    CollaboratorAuthor

    Scalability

    I'll also test how well Block Lab scales.

    The main WP_Query usage is for the blocks added via the UI (stored in a custom post type). But this has a posts_per_page value of 100 to minimize scaling issues.

    blocks-via-ui

    Block Lab has a PHP API that should be used when there are more than 100 blocks needed:

    block_lab_add_block(
            'testing-block',
            ...

    I'll test whether this has memory or performance issues with extreme amounts of blocks, like in the hundreds.

  5. kienstra commented on Dec 2, 2019

    @kienstra
    CollaboratorAuthor

    Performance of Block Lab When There's No Block

    What's the PHP performance cost of Block Lab on the front-end of a post that doesn't have a Block Lab block?

    In the test below, the average cost of Block Lab was 83.4 milliseconds, very similar to the test above with 1 block.

    Here's a Blackfire profile of this:
    https://blackfire.io/profiles/f7414907-bbcb-4200-8c3c-6e66e615a1a5/graph

    Methodology

    1. Activate Twenty Nineteen
    2. Deactivate all plugins except Block Lab
    3. Create a new post with WP-CLI:
    wp post create --post_title="Lorem post" --post_status='publish' --post_content='Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.'
    1. Go to the front-end, and run 10 Blackfire profiles
    2. Deactivate Block Lab
    3. Repeat step 4

    front-end-post

    Performance

    Milliseconds that PHP took to produce the HTML document (in 10 trials)

    With Block Lab Without Block Lab
    275 200
    278 189
    274 190
    261 196
    284 188
    275 187
    282 200
    277 189
    278 188
    264 187

    Average of Trials

    With Block Lab Without Block Lab
    274.8 191.4

    Difference: 83.4 milliseconds

  6. kienstra commented on Dec 2, 2019

    @kienstra
    CollaboratorAuthor

    The test above might show some chances for improvement.

    For example maybe Block_Post::register_controls() could only run on the front-end when Loader::render_block_template() runs.

    If there's no Block Lab block on the front-end, there shouldn't be a need to register block controls.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions