How the hell is json encode faster , compared to every thing else on benchmark

[KAKAROT379]

> protobufjs@7.5.4 bench
> node bench

benchmarking encoding performance ...

protobuf.js (reflect) x 472,670 ops/sec ±3.46% (84 runs sampled)
protobuf.js (static) x 503,024 ops/sec ±2.78% (93 runs sampled)
JSON (string) x 620,443 ops/sec ±2.48% (92 runs sampled)
JSON (buffer) x 511,103 ops/sec ±4.41% (90 runs sampled)
google-protobuf x 278,468 ops/sec ±3.29% (87 runs sampled)

          JSON (string) was fastest
          JSON (buffer) was 19.1% ops/sec slower (factor 1.2)
   protobuf.js (static) was 19.2% ops/sec slower (factor 1.2)
  protobuf.js (reflect) was 24.5% ops/sec slower (factor 1.3)
        google-protobuf was 55.5% ops/sec slower (factor 2.2)

benchmarking decoding performance ...

protobuf.js (reflect) x 1,407,764 ops/sec ±1.39% (92 runs sampled)
protobuf.js (static) x 1,518,683 ops/sec ±2.12% (93 runs sampled)
JSON (string) x 444,182 ops/sec ±0.81% (89 runs sampled)
JSON (buffer) x 386,086 ops/sec ±2.25% (91 runs sampled)
google-protobuf x 308,558 ops/sec ±4.06% (89 runs sampled)

   protobuf.js (static) was fastest
  protobuf.js (reflect) was 6.6% ops/sec slower (factor 1.1)
          JSON (string) was 70.4% ops/sec slower (factor 3.4)
          JSON (buffer) was 74.6% ops/sec slower (factor 3.9)
        google-protobuf was 80.1% ops/sec slower (factor 5.0)

benchmarking combined performance ...

protobuf.js (reflect) x 356,577 ops/sec ±2.40% (93 runs sampled)
protobuf.js (static) x 377,525 ops/sec ±0.50% (91 runs sampled)
JSON (string) x 248,215 ops/sec ±2.41% (94 runs sampled)
JSON (buffer) x 206,619 ops/sec ±2.62% (89 runs sampled)
google-protobuf x 140,726 ops/sec ±0.81% (90 runs sampled)

   protobuf.js (static) was fastest
  protobuf.js (reflect) was 7.3% ops/sec slower (factor 1.1)
          JSON (string) was 35.5% ops/sec slower (factor 1.5)
          JSON (buffer) was 46.4% ops/sec slower (factor 1.9)
        google-protobuf was 62.8% ops/sec slower (factor 2.7)

benchmarking fromJSON performance ...

isolated x 471 ops/sec ±4.83% (83 runs sampled)
isolated (resolveAll) x 491 ops/sec ±2.10% (87 runs sampled)
shared (unique) x 4.28 ops/sec ±5.74% (15 runs sampled)

  isolated (resolveAll) was fastest
               isolated was 6.6% ops/sec slower (factor 1.1)
        shared (unique) was 99.2% ops/sec slower (factor 118.8)

benchmarking resolveAll performance ...

isolated x 61,481,810 ops/sec ±4.94% (77 runs sampled)

benchmarking load performance ...

sync x 3,442 ops/sec ±6.53% (74 runs sampled)

OS: Ubuntu 24.04.3 LTS x86_64
Host: *VPS*
Kernel: 6.6.111+
Uptime: 1 hour, 21 mins
Packages: 843 (dpkg)
Shell: bash 5.2.21
Terminal: /dev/pts/0
CPU: Intel Xeon (2) @ 2.199GHz
Memory: 377MiB / 7947MiB

Npm: 11.6.2
Node: v24.11.1

I tried it multiple times and i am getting similar results