Measure performance impact of recent async changes #444
Assignees
Milestone
Comments
|
I've looked into it over the week, and I have a problem on how to approach it.
I did prepare a basic example that rerenders a lot of times on click. Templates<head>
<title>blaze-benchmarks</title>
<meta name="viewport" content="width=device-width, initial-scale=1.0">
</head>
<body>
{{> benchmark template='let_alias' iterations=1000}}
{{> benchmark template='let_object_shallow' iterations=1000}}
{{> benchmark template='let_object_deep' iterations=1000}}
</body>
<template name="benchmark">
<button>Run</button>
<code style="display: inline-block; width: 20ch"><b>{{template}}</b></code>
{{#if summary}}
<code>{{summary}}</code>
{{/if}}
<br>
<div style="display: none">
{{> Template.dynamic data=data template=template}}
</div>
</template>
<template name="let_alias">
{{#let x=a}}{{x}}{{/let}}
</template>
<template name="let_object_shallow">
{{#let x=a.b}}{{x}}{{/let}}
</template>
<template name="let_object_deep">
{{#let x=a.b.c.d.e.f}}{{x}}{{/let}}
</template>import { Template } from 'meteor/templating';
import { ReactiveVar } from 'meteor/reactive-var';
import './main.html';
const values = new ReactiveVar({
a: { b: { c: { d: { e: { f: '' } } } } }
});
Template.benchmark.onCreated(function () {
this.summary = new ReactiveVar(null);
});
Template.benchmark.events({
'click button'(event, { data: { iterations }, summary }) {
const timings = [];
// Warmup.
for (let iteration = 0; iteration < iterations; ++iteration) {
values.dep.changed();
Tracker.flush();
}
// Measurement.
for (let round = 0; round < 100; ++round) {
const now = performance.now();
for (let iteration = 0; iteration < iterations; ++iteration) {
values.dep.changed();
Tracker.flush();
}
timings.push(performance.now() - now);
}
// Summary.
const ps = [25, 50, 90, 95];
timings.sort();
summary.set(ps.map(p => `p${p}=${Math.floor(timings[p - 1])}`).join(' '));
},
});
Template.benchmark.helpers({
data: () => values.get(),
summary: () => Template.instance().summary.get(),
});But, as expected, these all take basically the same amount of time. What do others think about it? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
While getting ready for Meteor 3.0, we've done a lot of internal changes (#412, #413, #424, #428; soon also one for #443). Most of them are simply adding on top of the existing logic to support
Promises virtually everywhere; some are actual redesigns (e.g., shape of_scopeBindings).Of course, adding more logic always comes with a cost, and we should check, whether it's acceptable. So far we didn't hear about any visible regressions, but it should be checked nevertheless.
It's strongly connected to #383, since the majority of the benchmarks should remain synchronous-only, so we could compare the results with previous versions.
The text was updated successfully, but these errors were encountered: