There's so many ways this benchmark can go wrong that there's pretty much no way I can trust this conclusion.
> All the loops call a dummy function DATA.doSomethingWithValue() for each array element to make sure V8 doesn't optimize out something too much.
This is probably the most worrying comment - what is "too much?" How are you sure it doesn't change between different implementations? Are you sure v8 doesn't do anything you don't expect? If you don't look into what's actually happening in the engine, you have no idea at this point. Either you do the real work and measure that, or do the fake work but verify that the engine does what you think it does.
There are a lot of "probably"s in the article. I was also suspicious that the author didn't say they did any pre measurement runs of the code to ensure that it was warmed up first. Nor did they e.g. use V8 arguments with Node (like --trace-opt) to check what was actually happening.
> All the loops call a dummy function DATA.doSomethingWithValue() for each array element to make sure V8 doesn't optimize out something too much.
This is probably the most worrying comment - what is "too much?" How are you sure it doesn't change between different implementations? Are you sure v8 doesn't do anything you don't expect? If you don't look into what's actually happening in the engine, you have no idea at this point. Either you do the real work and measure that, or do the fake work but verify that the engine does what you think it does.