Benchmark yielding similar results against encoding/json

[nuric]

Hello,

I'm probably missing something simple here. When I was trying to compare against standard library using the same large fixture:

func BenchmarkDecodeJSON(b *testing.B) {
	payload := ... // gojson.benchmark.LargeFixture
	for _, decodeLibrary := range []string{"encoding/json", "goccy/go-json"} {
		b.Run(decodeLibrary, func(b *testing.B) {
			for i := 0; i < b.N; i++ {
				// Unmarshal the payload
				var v map[string]any
				var err error
				switch decodeLibrary {
				case "encoding/json":
					err = json.Unmarshal(payload, &v)
				case "goccy/go-json":
					err = gojson.Unmarshal(payload, &v)
				}
				if err != nil {
					b.Fatal(err)
				}
			}
		})
	}
}

I get:

cpu: Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz
BenchmarkDecodeJSON/encoding/json-2         	    3286	    352568 ns/op	  209375 B/op	    2785 allocs/op
BenchmarkDecodeJSON/goccy/go-json-2         	    3772	    320198 ns/op	  249825 B/op	    3490 allocs/op
PASS

Is this not the same data used for the benchmarking in the README? I was expecting to a significant difference, especially for the large fixture payload in decoding.

Running them in seperate benchmarks (instead of using b.Run) gives the same results.

Thank for your help.