I think that benchmarks are made wrong way. They typically calculate how many times per second a browser can run some code. But this causes browser vendors to d

I think that benchmarks are made wrong way. They typically calculate how many times per second a browser can run some code. But this causes browser vendors to do trade-offs, for example, increase memory consumption which increases the probability of swapping which causes everything to stall. A more fair way would be to account for memory usage: score = speed ÷ (RAM usage)² Or they should run their tests at 2Gb system with lot of other applications running in background.