They have recently published a blog series which started with 35,000x and ended with 68,000x [1]. The thing is anyone familiar with performance sensitive code is not looking at Python vs the other thing. They are looking for whether or not it is in the ballpark of C, Julia, Cython, Numba, etc. Those are the contenders, not Python.

[1] https://www.modular.com/blog/mojo-a-journey-to-68-000x-speed...

I think the reason that they focus on pure Python is because a big use case would be pure Python code bases that can be incrementally ported over to Mojo for better performance, without needing to totally abandon the Python language and ecosystem.

If you read the description, they started with 35,000x, then tripled the number of cores to get 68,000x. If your triple the cores and wind up with less than twice the performance, your scaling isn't very good.

Depends on the task not all tasks scales linearly

It is very good, especially if you don't have to rewrite your code.

Try to do it yourself (say take a merge sort implementation and try to make it scale), it'll be roughly this much after a few hours of work.

Now imagine the compiler is doing this for you.

Keyword is imagine, because with Mojo you do have to write your code in a new language. You cannot use the Python parts if you want speed. That's writing code in a new language.

That's the stated goal, it's not mature technology.

You can have your python parts source compatible with the new language.

Of course they are far off from that.

But not that far off if your python code is typical ML framework consumer code.

I have never taken them seriously precisely because of this statement. They are a machine learning specific language and everyone who works in machines learning knows the speedup is cherry picked bullshit. If they're basic their markering around an obviously false claim, why would I take a closer look?

The claim is likely not false just like a horse pulling a buggy is several orders of magnitude faster than a snail.