Is There A Chance That MicroPython will get as good as C++?

[ovexro]

Hello.

I hope this question doesn't offend anyone, but Claude compares C++ and MicroPython and says that MicroPython must catch up to C++.

Is there a chance that MicroPython will get as good as C++?

Thanks.

[robert-hh]

Is there a chance that MicroPython will get as good as C++?

The questions seems nonsense to me. In my ears it sounds like: Will apples get the same as oranges?

Besides that: what do you call good? Python and C++ are different, and both are good at their playgrounds. MicroPython is a restricted version of Python, it is not a replacement for code written in C++.

[sosi-deadeye]

The question is about: Development Speed vs. Execution Speed

Micropython: A dynamic language with a bytecode compiler and interpreter. If you know python very well, you can develop very fast. Execution speed could not compare with C/C++. Optimization could be done with decorators (micropython.native, micropython.viper) or with assembler. I think you can also use compiled mpy-Files, which includes different languages, but this afik. in the making.

C/C++: Static complied language, runs on everything, runs fast and let you shoot yourself in the foot.

[peterhinch]

As one who's coded in both languages, I know which I prefer. Life offers more enjoyable experiences than hours spent hunting down that last memory leak in an a big application.

[sk8board]

In general, which is faster, viper or Native machine code in .mpy files or MicroPython external C modules?

[jonnor]

Native code is faster than any bytecode, as long as FFI overheads are kept small (meaning a sizable chunk of work is done per function invocation). Dynamic module or C module can be quite similar. Though .mpy files execute from RAM - whereas custom C modules execute from FLASH. Which can have different performance characteristics - hardware dependent.
But the best is to do some benchmarks for the workloads you are interested in. And share the results :)

[peterhinch]

The phrase "Native machine code" is perhaps ambiguous. Carefully crafted code using the inline assembler is extremely fast. By contrast the native decorator often doesn't make much difference (in my experience). With care and the "right" kind of code the viper decorator can be fast - comparable to C or Assembler. The "right" kind of code being bit manipulations, addressing of arrays through pointers, integer arithmetic etc.

This is all hand-waving, albeit based on experience and measurement. I don't know of any generally accepted quantitative results.