NodeJs how to create a non-blocking computation

This is a classic misunderstanding of how the event loop is working.

This isn't something that is unique to node - if you have a long running computation in a browser, it will also block. The way to do this is to break the computation up into small chunks that yield execution to the event loop, allowing the JS environment to interleave with other competing calls, but there is only ever one thing happening at one time.

The setImmediate demo may be instructive, which you can find here.

If you computation can be split into chunks, you could schedule executor to poll for data every N seconds then after M seconds run again. Or spawn dedicated child for that task alone, so that the main thread wouldn't block.

You can't do that directly, without using some of the IO modules in node (such as fs or net). If you need to do a long-running computation, I suggest you do that in a child process (e.g. child_process.fork) or with a queue.

We (Microsoft) just released napajs that can work with Node.js to enable multithreading JavaScript scenarios in the same process.

your code will then look like:

var napa = require('napajs');

// One-time setup. 
// You can change number of workers per your requirement. 
var zone = napa.zone.create('request-worker-pool', { workers: 4 });

http.createServer(function(req, res) {
    console.log(req.url);

    zone.execute((request) => {
        var result = null;
        // Do heavy computation to get result from request
        // ...
        return result;
    }, [req]).then((result) => {
        res.end(result.value);
    }
}).listen(8080, function() { console.log("ready"); });

You can read this post for more details.