Re: "Python is 71x Slower, Uses 75x More Energy, Than C" - YouTube

On 7/31/24 08:13, jseigh wrote:

On 7/30/2024 8:05 PM, Lynn McGuire wrote:

"Python is 71x Slower, Uses 75x More Energy, Than C" - YouTube
   https://www.youtube.com/watch?v=U4c6nFGt1iM

>
I don't know for sure but I suspect part of it is Python using
reference counting for it's memory managment. Maybe they should
compare it to jython.

>
I am not buying that C++ is slower than Rust.
>

Rustaceans make a lot of unsupported claims. They also claim without
any support that java is slower than rust. Apparently they are
completely ignorant of the fact that java uses JIT compilation and can
dynamically tune performance hot spots, and if they are using
reference counting to manage memory it's going to perform very badly
compared to Java GC.

Keep in mind that this is not an "unsupported claim" - it's the result
of measurements, with the method used to make the measurements
specified, though not completely. I hove no idea how, but you should be
able to ask the author of the paper to send you copies of the source
code used, and details of the compilation options chosen. In this kind
of academic study, providing such details on request is considered
mandatory.

In order to counter such evidence, you should ideally try to repeat the
tests. You might find that they are biased in some way. For instance,
the Rush versions of the programs might have been written by an expert,
while the C++ versions might have been written by someone who hates C++,
Alternatively, you might find that the particular algorithms chosen
create a bias towards Rust over C++. But you can't simply dismiss this
as due to the author of that study being a Rust fanboy - the study
covers many different languages, and if anything suggests that C vs.
Python is his main interest.