Added benchmarks.

[n1k0]

r=? @gijsk @leibovic

[gijsk]

Can we get an average of a fixed set of those ops/s so it's easier to compare the output ?

[gijsk]

(the total time for all benchmarks is less useful because new commits tend to add a new testcase)

[n1k0]

Can we get an average of a fixed set of those ops/s so it's easier to compare the output ?

I couldn't find an option for this, except exporting the result as CSV (brrr). We could create a separate suite with a fixed subset of these test pages, and mark it as a constant reference point, though.

[gijsk]

Can we get an average of a fixed set of those ops/s so it's easier to compare the output ?

I couldn't find an option for this, except exporting the result as CSV (brrr). We could create a separate suite with a fixed subset of these test pages, and mark it as a constant reference point, though.

That sounds like a good idea, yeah.

[n1k0]

[n1k0]

@gijsk now we have these two distinct commands, npm run perf and npm run perf-reference, but we need to select the right subset of reference test pages. Could you please take over this, then land the patch when you're happy? It's a little late here :)

[n1k0]

Ah just spotted: all this should be removed from this file, as we're not in a test context.

[gijsk]

Done.