deno/website/benchmarks.html

193 lines
6.3 KiB
HTML

<!-- Copyright 2018-2019 the Deno authors. All rights reserved. MIT license. -->
<!DOCTYPE html>
<html>
<head>
<title>Deno Benchmarks</title>
<link rel="shortcut icon" href="favicon.ico">
<link rel="stylesheet" href="https://unpkg.com/c3@0.6.7/c3.min.css" />
<link rel="stylesheet" href="style.css" />
<meta content="width=device-width, initial-scale=1.0" name="viewport" />
</head>
<body>
<div id="spinner-overlay">
<div class="spinner"></div>`
</div>
<main>
<a href="/"><img src="images/deno_logo_4.gif" width=200></a>
<h1>Deno Continuous Benchmarks</h1>
<p>
These plots are updated on every commit to
<a href="https://github.com/denoland/deno">master branch</a>.
</p>
<p>Make sure your adblocker is disabled as some can block the chart rendering.</p>
<p><a href="#recent">recent data</a></p>
<p><a href="#all">all data</a> (takes a moment to load)</p>
<h3 id="req-per-sec">Req/Sec <a href="#req-per-sec">#</a></h3>
<p>
Tests HTTP server performance. 10 keep-alive connections do as many
hello-world requests as possible. Bigger is better.
</p>
<ul>
<li>
<a
href="https://github.com/denoland/deno/blob/master/tools/deno_tcp.ts"
>deno_tcp</a>
is a fake http server that doesn't parse HTTP. It is comparable to
<a
href="https://github.com/denoland/deno/blob/master/tools/node_tcp.js"
>node_tcp</a>
.
</li>
<li>
<a
href="https://github.com/denoland/deno_std/blob/master/http/http_bench.ts"
>deno_http</a>
is a web server written in TypeScript. It is comparable to
<a
href="https://github.com/denoland/deno/blob/master/tools/node_http.js"
>node_http</a>
.
</li>
<li>deno_core_single and deno_core_multi are two versions of
a minimal fake HTTP server. It blindly reads and writes fixed HTTP
packets. It is comparable to deno_tcp and node_tcp.
This is a standalone executable that uses <a
href="https://crates.io/crates/deno">the deno rust crate</a>. The
code is in
<a
href="https://github.com/denoland/deno/blob/master/core/examples/http_bench.rs"
>http_bench.rs</a>
and
<a
href="https://github.com/denoland/deno/blob/master/core/examples/http_bench.js"
>http_bench.js</a>. single uses <a
href="https://docs.rs/tokio/0.1.19/tokio/runtime/current_thread/index.html">tokio::runtime::current_thread</a>
and multi uses <a
href="https://docs.rs/tokio/0.1.19/tokio/runtime/">tokio::runtime::threadpool</a>.
</li>
<li>
<a
href="https://github.com/denoland/deno/blob/master/tools/hyper_hello.rs"
>
hyper
</a>
is a Rust HTTP server and represents an upper bound.
</li>
</ul>
<div id="req-per-sec-chart"></div>
<h3 id="max-latency">Max Latency <a href="#max-latency">#</a></h3>
<p>
Max latency during the same test used above for requests/second. Smaller is better.
</p>
<div id="max-latency-chart"></div>
<h3 id="exec-time">Execution time <a href="#exec-time">#</a></h3>
<p>
This shows how much time total it takes to run a few simple deno
programs:
<a
href="https://github.com/denoland/deno/blob/master/tests/002_hello.ts"
>
tests/002_hello.ts
</a>,
<a
href="https://github.com/denoland/deno/blob/master/tests/003_relative_import.ts"
>tests/003_relative_import.ts</a
>,
<a
href="https://github.com/denoland/deno/blob/master/tests/workers_round_robin_bench.ts"
>tests/workers_round_robin_bench.ts</a
>, and
<a
href="https://github.com/denoland/deno/blob/master/tests/workers_startup_bench.ts"
>tests/workers_startup_bench.ts</a
>. For deno to execute typescript, it must first compile it to JS. A
warm startup is when deno has a cached JS output already, so it should
be fast because it bypasses the TS compiler. A cold startup is when deno
must compile from scratch.
</p>
<div id="exec-time-chart"></div>
<h3 id="throughput">Throughput <a href="#throughput">#</a></h3>
<p>
Time it takes to pipe a certain amount of data through Deno.
<a
href="https://github.com/denoland/deno/blob/master/tests/echo_server.ts"
>
echo_server.ts
</a>
and
<a href="https://github.com/denoland/deno/blob/master/tests/cat.ts">
cat.ts </a
>. Smaller is better.
</p>
<div id="throughput-chart"></div>
<h3 id="max-memory">Max Memory Usage <a href="#max-memory">#</a></h3>
<p>
Max memory usage during execution. Smaller is better.
</p>
<div id="max-memory-chart"></div>
<h3 id="size">Executable size <a href="#size">#</a></h3>
<p>deno ships only a single binary. We track its size here.</p>
<div id="binary-size-chart"></div>
<h3 id="threads">Thread count <a href="#threads">#</a></h3>
<p>How many threads various programs use.</p>
<div id="thread-count-chart"></div>
<h3 id="syscalls">Syscall count <a href="#syscalls">#</a></h3>
<p>
How many total syscalls are performed when executing a given script.
</p>
<div id="syscall-count-chart"></div>
</main>
<script src="https://unpkg.com/d3@5.7.0/dist/d3.min.js"></script>
<script src="https://unpkg.com/c3@0.6.7/c3.min.js"></script>
<script type="module">
import { drawCharts } from "./app.js";
window.chartWidth = 800;
const overlay = document.getElementById("spinner-overlay")
function showSpinner () {
overlay.style.display = "block";
}
function hideSpinner () {
overlay.style.display = "none";
}
function updateCharts () {
const u = window.location.hash.match("all") ? "./data.json" : "recent.json";
showSpinner()
drawCharts(u).finally(hideSpinner)
}
updateCharts()
window.onhashchange = updateCharts
</script>
</body>
</html>