There's variance in the output, sometimes showing an 8% difference, but it's always faster. I noticed this when doing profiling of an application we are building, where protobuf parsing and serialization is where most of the work is. Our app needs to be very highly performant so this makes a big difference.
Using named arguments on the protobuf.js wrapper instead of the
arguments
expression improves performance by a small but noticeable amount.before: $ node test/perf/perfTest.js serializing and parsing a message 1,000,000 times... serialize: 8876ms parse: 8054ms
after: $ node test/perf/perfTest.js serializing and parsing a message 1,000,000 times... serialize: 8378ms # 5% faster parse: 7713ms # 4% faster
There's variance in the output, sometimes showing an 8% difference, but it's always faster. I noticed this when doing profiling of an application we are building, where protobuf parsing and serialization is where most of the work is. Our app needs to be very highly performant so this makes a big difference.