jakubner/resolved-promise-performance

How fast is retrieval of a value from a resolved promise compared to retrieval of the value sans promise.

Benchmark

What is the overhead of requesting a value from a resolved promise?

Motivation

For a static value that is retrieved asynchronously subsequent calls return the same known value: what is the performance cost once it can be furnished as an already resolved promise?

For example a static data-point retrieved from some data-source is always retrieved as a promise. The first retrieval will incur asynchronous penalty as the value is unknown. What about subsequent retrievals? We're still using promises in an asynchronous call, but what's the cost?

Running

npm install
node index.js

Results

donePromise x 186,285 ops/sec ±39.04% (17 runs sampled)
value x 220,988 ops/sec ±19.03% (11 runs sampled)
fn x 80,614 ops/sec ±137.75% (10 runs sampled)
fastest :: value,donePromise,fn

donePromise x 171,476 ops/sec ±31.84% (13 runs sampled)
value x 183,915 ops/sec ±43.85% (11 runs sampled)
fn x 30,836 ops/sec ±192.23% (9 runs sampled)
fastest :: donePromise,value,fn

Retrieval from a resolved promise is almost as fast as retrieval of the value sans Promise wrapper.

Note that all tests are asynchronous calls--even when the value could be returned synchronously. This is to mimic the fact that resolved promises can only be used in asynchronous call handlers.