Python consumes 38x more energy than Java

https://stratoflow.com/efficient-and-environment-friendly-programming-languages/

by misiax • 3 years ago

52 140 3 years ago

stratoflow.com

_han• 3 years ago

I would like to point out that the data in the table is from 2017, and CPython recently focused a lot on performance.

boesboes2• 3 years ago

The actual benchmark: https://benchmarksgame-team.pages.debian.net/benchmarksgame/...

EDIT: fwiw, classic Fortran is twice as fast as java :P As is Rust, but that's less funny to me

zelphirkalt5• 3 years ago

Interesting, what is "Chapel"? Will need to look that up.

EDIT: seems to be https://chapel-lang.org/

yread• 3 years ago

It is very fast and very memory efficient apparently

thumbuddy5• 3 years ago

Fortran is a beast. People crap on it all the time yet use it often millions of times a day without realizing it.

chii• 3 years ago

People crap on it all the time

only clueless people looking for clout online!

skissane2• 3 years ago

Fortran is a beast. People crap on it all the time

I think most of the people who "crap on" Fortran, when they say "Fortran", mean FORTRAN 77, not Fortran 2018

smcl• 3 years ago

I don't think Fortran gets beaten up that much, it's just sort of ... ignored

ilyt4• 3 years ago

use in what ?

I severely doubt anything I do touches fortran millions times a day unless it's google search that runs on it.

xioxox• 3 years ago

As a Fortran user, it's pretty awful for string handling. It's fine for numerics. However, there's a lack of libraries for non-numeric stuff (e.g. data structures). The free compilers are buggy, too.

hmaarrfk• 3 years ago

Thanks for this link.

They also seemed to have avoided libraries like numpy.

In my mind, python made it possible to hardware optimize with libraries like numpy quite easily. Avoiding it is a mistake. I'll try to see if I have time to play the game myself and throw my attempt in there.

davidgrenier• 3 years ago

But isn't what makes numpy efficient written in C?

KingOfCoders• 3 years ago

If your web service runs with numpy, excellent.

igouy17• 3 years ago

No, this is "the actual benchmark" —

"Source: Energy Efficiency across Programming Languages, SLE’17"

and google finds —

https://greenlab.di.uminho.pt/wp-content/uploads/2017/09/pap...

Uninen• 3 years ago

Even more clickbaity title would have been "Python consumes 70x more energy than Rust"

ndlan• 3 years ago

clickbaity and correct

igouy17• 3 years ago

The link is marketing for some consulting firm.

ChrisRR• 3 years ago

Everything consumes more energy than C

ttoinou• 3 years ago

Except ASM, AVX etc.

SomeRndName11• 3 years ago

no, VHDL and Verilog consume less

thumbuddy5• 3 years ago

Sometimes I wonder if that's true long term. The amount of bugs, security issues, and code churn surrounding C code bases is pretty wild. If everybody and all their deployments has to recompile a C code base 100 times over the course of three years to get patches, fixes, etc it makes me wonder what the real cost of C is?

My intuition states that this isn't the case, but intuition can be wrong.

seba_dos1• 3 years ago

C is not low-level enough for it to be even remotely true.

thumbuddy5• 3 years ago

Yea but it's HN you can't mention Rust just being rust without a bunch of people losing their minds over the fact that someone said "Rust".

inglor_cz• 3 years ago

"We are the knights who say Rust!"

igouy17• 3 years ago

The link is marketing for some consulting firm.

zelphirkalt5• 3 years ago

While Java's compiler or JIT is probably one of the most optimized softwares in existence, how do we measure, what overhead the language nudges the programmers to build? I am thinking of AbstractProviderFactoryProxy and similar. Or that for a very long time it forced you (or still does to a degree) to put everything into classes, breeding 1 or 2 generations of programmers, who see a noun and jump to making a class and then subsequently initializing the class to be able to call a method, which actually would only need to be a standalone function.

It probably does not balance the 38x, and probably the JIT optimizes parts of it away, but surely the cost is far from zero.

Also what about the computational resources it takes to just run a JVM, compared to running a Python program?

DarkNova6• 3 years ago

Please explain how "AbstractProviderFactoryProxy" is related to the language.

I have taken over as lead for a Python project and what I see all the Java problems people used to make fun of (none have worked in Java before) and worse. On top, everything has an interface starting with "I", C# style. Methods in entities, service classes are instantiated with state etc. And side-effects are all over the place.

If you are writing modern Java syntax, there is little that is more stable and readable than good modern Java syntax.

zelphirkalt5• 3 years ago

Please explain how "AbstractProviderFactoryProxy" is related to the language.

The language Java did, until some time ago, not allow you to pass functions as arguments. You had to make an anonymous inner class for example, satisfying some interface. Many design patterns are very much oriented towards a language like Java, which does/did not allow for higher order functions.

Take the visitor pattern for example. It will get you a "Visitor" in your class name implementing a visitor. How would it work in other languages, which provide higher order functions or always did so? Well, you would simply write a function and depending on whether your language has static types, annotate its types for arguments and return value. You would then pass this function in as an argument, whose name can be "visitor".

Take a factory pattern as an example. How would it work? Well, a factory can be expressed using a function, that returns another function. It takes the arguments, that specify/narrow down how the returned function works. In Java it would become a Factory class instead. Instead of using a simple function, one needs (needed?) to build a whole class around that, because one could not return a function. So one would build that class and then make it return an object instead, which of course again must be derived from some class ...

Just 2 examples that quickly come to mind.

I have taken over as lead for a Python project and what I see all the Java problems people used to make fun of (none have worked in Java before) and worse.

It is certainly true, that people also write shit code in Python. I have seen very popular libraries, which wrap REST APIs in objects. It is silly, because objects are meant to have some lifetime in which they communicate with each other and possibly change their state. There is no changing state though. Everything is a response from the REST API, which by the definition of REST should transfer the representative state in its responses. I don't want any in between stored state. I want the state that the API gives me. It is very much a more functional view on things. But Python programmers will go "make classes!" nevertheless. Because noun. Because not thinking about whether they really need a class. Because thinking that one approach fits it all and the only approaches they learned were procedural at the beginning of their learning and then OOP, which is complex enough to take years to learn properly, so that is where they stopped.

However, Python always (for a very long time?) has allowed you to pass procedures as arguments and as such does not encourage "AbstractProviderFactoryProxy" as much as Java does. Still, sometimes former Java developers try their hand at some Python code ...

If you are writing modern Java syntax, there is little that is more stable and readable than good modern Java syntax.

Modern Java certainly has improved. But there is a lot of relearning to be done for those generations of Java programmers, to get rid of the "every noun a class" mentality.

ptx• 3 years ago

see a noun and jump to making a class

Even when what you Utils.see Utils.is a verb, you Utils.must still Utils.jump to Utils.make a class.

Java 5's static imports at least allow cleaning up the call site, although the documentation advises (in bold) using this feature "very sparingly".

kaba010• 3 years ago

How is a class that holds a static method/function any different to simply namespacing? Never understood people’s problem with that. For example, what’s the problem with Java’s Math “class”? It only has stateless math functions like any other language, nothing is forced on you.

zelphirkalt5• 3 years ago

It does not have to be a problem in terms of it working or not.

The problem is rather, that a class comes with loads of conceptual baggage. It is conceptually not fit for the purpose of merely namespacing.

A good language will have a concept fit for the purpose of namespacing. Either something directly called "namespace" or something that is meant to namespace things, like modules. A class is rather for grouping methods that interact with the state of the objects.

Of course, if you don't have anything else available, you gotta take what you have. It will not serve making the code more readable though, as a potentially new reader of the code will expect a class to be instanced somewhere, as is typical for classes. It can lead to confusion. If there was however a concept "namespace", they will immediately know that it is for grouping a category of things.

awelxtr• 3 years ago

Does this table take into account energy used while developing/maintaining?

thumbuddy5• 3 years ago

Sincerely doubt it does because that's a difficult thing to measure. If you are writing a lambda function at a FAANG company you do way more than break even with a language that has a heavy compilation time. If you're fixing a comment for a hobby project that doesn't ship binaries that you and your nephew use, definitely not.

The deployment and consumption of said code matters a lot for that.

speedgoose• 3 years ago

No.

Here is the source of the table: https://greenlab.di.uminho.pt/wp-content/uploads/2017/10/sle...

marginalia_nu5• 3 years ago

One should always be skeptical of these sorts of benchmarks, especially against Java, and I say this as primarily a Java developer.

Well optimized Java is very lean and can be incredibly performant, but the median modern Java app is an obese cronenberg made out of gigabytes of SpringBoot dependencies. That's not particularly fast; and well optimized python will run circles around it.

boesboes2• 3 years ago

This is an important point, real life performance looks nothing like benchmarks in most cases.

From my personal experience: benchmarks showed Jruby was MUCH MUCH faster the 'normal' Cruby. But try running a basic rails application on jruby and it's 10-100x slower, even after minutes of repeating the same request.

(disclaimer, this was a few years back and purely anecdotal)

renegade-otter• 3 years ago

"Artificial tests produce artificial results".

ramblerman2• 3 years ago

You seem to be conflating speed and memory usage.

A spring boot app will generally be very performant. It will indeed use more memory though.

estebank3• 3 years ago

I did an excercise some time ago and managed to get two file transfer apps, one in Rust and one in Java, where the performance was effectively the same (in some cases with a slight Java edge!), but the memory consumption was orders of magnitude in favor of Rust. The JVM is a wonder of engineering, and Java is indeed quite fast.

RamblingCTO• 3 years ago

You should get your hands on say, go and go-fiber, and compare that with spring boot regarding speed and memory. And I say that with our main backend being kotlin/java and spring boot. It's a nightmare and absolute and utter garbage.

marginalia_nu5• 3 years ago

No man spring boot is hideously slow as well compared to vanilla java. It's pretty fast compared to many other things, but that's just because vanilla java is extremely fast.

You often see like 15-30 second start times for springboot apps, which is hilarious and 13-28 seconds longer than it has any business taking.

BoardsOfCanada2• 3 years ago

As a data point I ported a spring boot application to rust and max memory lowered from 200 mb to 9 mb.

kaba010• 3 years ago

As opposed to the lean Python Django webservices? What a bad take..

marginalia_nu5• 3 years ago

As opposed to well written code that doesn't use bloated frameworks.

radicalbyte2• 3 years ago

This is why I'm learning Rust: C# gives fast development speeds than Python with decent performance. Python and Go for when their native libraries smoke C# and now Rust for performance or low level.

kaba010• 3 years ago

Go is not generally faster than C#/Java. There are scenarios where each can come up on top.

radicalbyte2• 3 years ago

Smoke as in have better support for specific (niche) use cases. Python has strong numeric libraries and Go has better support for modern/interesting/crazy cryptography systems than C#/Java.

So in reality there are very few places where you would want to use Go over modern C#.

boredumb5• 3 years ago

Just write in Rust, it's easier to build good software than either of these languages (yea, yea, do you data analyst stuff in python until you start using POLARS). At this point it's a far superior ecosystem from a developer experience point of view and the fact it's going to get you as close to efficient as possible without you thinking too far into it is a welcome side effect.

Using VScode with the Sqlx package I get compile time errors on my SQL - as a result I haven't ran a binary with a typo from my SQL in the last 12 months.

TachyonicBytes• 3 years ago

Do you use an extension other than just `rust-analyzer` in order to get compile-time errors for SQL with sqlx?

boredumb5• 3 years ago

I believe it's just the rust-analyzer - I only installed the rust extension via vscode so it may have included an extra bit.

boredumb5• 3 years ago

To follow up - no, I reproduced an error - disabled my rust-analyzer extension in vscode and it went away.

movpasd• 3 years ago

Are there any good interactive programming options in Rust? That's the main draw of a dynamic language like Python for ad hoc data science tasks, the semantics just lines up well with the highly iterative nature of the work.

boredumb5• 3 years ago

I know I saw earlier in the year a Jupyter notebook for rust - but I haven't used it personally so I'm not sure how nicely it will play. I use jupyter+python to view and play with the data usually and then write my actual job in rust with Polars to use as the ETL or whatever i'm doing with it at that moment.

slowmovintarget3• 3 years ago

Easy? No. Rust is a far more difficult language to work in than Python.

It will require you to write more correct code, so the end product is more likely to come out better. But it is not easy, even for building "good" software. (Assumption: "good" := correct, maintainable, efficient, and robust -- Rust nails the efficient part.)

fancyfredbot• 3 years ago

C++ uses 34% more energy than C?

I thought C code was generally also valid C++ code and so you really should be able to get the same performance and energy consumption. Very odd.

tasubotadas• 3 years ago

You can write C in C++ but it won't be idiomatic C++. I've seen plenty of experienced "C++" developers that have no idea what is "shared_ptr".

dahart6• 3 years ago

True partly because shared_ptr is a relatively new language feature; I’ve been locked to a ~decade-old C++ compiler for most of my professional career, and so has everyone I know, just because support for a current compiler across platforms at any given time has been historically so spotty. It seems to be improving at the moment.

it won’t be idiomatic

I just looked at the first example:

C: https://benchmarksgame-team.pages.debian.net/benchmarksgame/...

C++: https://benchmarksgame-team.pages.debian.net/benchmarksgame/...

The C++ doesn’t use shared_ptr or STL or anything, and isn’t that far from the C code, but it isn’t that close either. I wonder if it’s OMP configuration that’s causing it to be slower, or maybe just instruction cache or something since the C++ is larger.

The results say this one has C++ going 50% slower. I’m a little skeptical it’s the compiler’s or the language’s fault. I’d speculate it might have more to do with how the program is written.

igouy17• 3 years ago

Last time I tried, only 3 of the 9 selected C programs seemed to compile as C++.

mg3• 3 years ago

Is Python's "values don't change, they get replaced by new ones stored elsewhere" approach what makes it slow?

I would like to see some micro-benchmarks which pin down the bottlenecks.

For example, this minimal benchmark with one variable and a few control structures:

    x=0
    for i in range(int(10e6)):
        if i<50: x=x+1
    print (x)

Runs 5 times slower here than the same in PHP:

    $x=0;
    for ($i=0; $i<10e6; $i++)
        if ($i<50) $x++;
    echo "$x\n";

Tested with:

    time python3 loop.py

Which gives me 0.31s

And:

    time php loop.php

Which gives me 0.06s

A factor of five seems to be roughly the average when comparing Python to PHP for a bunch of different code constructs I tried.

snicker7• 3 years ago

You can’t use time command in this sort of benchmark. You are including start up t times here.

mg3• 3 years ago

Startup times seem to be the same and also negligible:

    time python3 nothing.py
    0.011s

    time php nothing.php
    0.011s

igouy17• 3 years ago

PHP JIT ?

misja111• 3 years ago

Amazing. I have seen this list popping up in my LinkedIn feed for more than a year already and it keeps coming back.

While any benchmark is of course interesting when considered on its own, the conclusions that people draw from this list tend to be complete nonsense: "Python is bad for the environment", "we should switch to language xxx to combat global warming" etc.

First of all, especially for the slower languages in this list, it is extremely rare that application code written in that language is the bottleneck of a performance critical application. Typically the bottlenecks of every day applications tend to be databases and network. If you need heavy number crunching, there tend to be excellent libraries available written in lower level languages, such as e.g. for Python NumPy and Pandas. If you are really concerned about the energy footprint of your application, you're probably better off with optimizing these components. Or optimizing the higher level architecture of your own application, which tends to be a lot easier in a higher level langguage.

Second, coding in a faster language is often not economically possible. Programming the same application in C will normally take much longer than coding it in Python. Developing time costs money and can also mean loss of opportunity.

And finally, the 'greenest' language of them all was not even included in this list: assembly. I guess the author must have realized in the back of his head that such a difficult language wasn't a realistic option for switching to a more energy efficient solution. He/she just failed to realize that this argument applies to many other languages as well.

mempko3• 3 years ago

I think you bring up a good point. Our economic system isn't designed to reward low energy consumption. In fact, GDP and energy usage are very tightly linked. And because we are seeking growth in GDP at a system level, we are doomed to use more energy.

screamingninja2• 3 years ago

GDP and energy usage are very tightly linked

Correlation or causation?

boredumb5• 3 years ago

Considering energy costs money, our economic system is literally designed to reward lower energy consumption.

screamingninja2• 3 years ago

I would add that proof of work algorithms (Bitcoin type) where the goal is to endlessly compute hashes with no real computational productivity gain are the real offenders, regardless of which language they're written in.

slowmovintarget3• 3 years ago

LLM has entered the chat...

Most of the implementations for LLMs runtimes I know of are Python using PyTorch. I believe most of the heavy lifting is written in C++ (llama.cpp), though.

Qwertious2• 3 years ago

regardless of which language they're written in.

And given that a major limitation of bitcoin-hashing is the cost of electricity, I would guess that they're written in a very energy-efficient language already.

kaba010• 3 years ago

"we should switch to language xxx to combat global warming" etc.

These takes fail to take into account the huge discrepancy between the market value produces by computer systems per unit of energy. It is such a huge value that it barely matters compared to most other industries, plus the net amounts are not particularly high either. Nonetheless, it can matter in certain cases, and Java is a great choice for server setups due to its GC being very efficient in doing only the necessary amount of work.

Also, I’m not convinced that a full on complex assembly app would be leaner than the same program written in a lower level language. That’s an area where compilers are very good, and humans only have so much working memory/hair on their head to hand-optimize whole programs.

igouy17• 3 years ago

The link is marketing for some consulting firm.

number6• 3 years ago

So Python is 38x more powerful?!

pella• 3 years ago

The research: "Energy Efficiency across Programming Languages" SLE’17, October 23–24, 2017

PDF: https://repositorio.inesctec.pt/bitstream/123456789/5492/1/P...

The human mental energy required for programming, unfortunately, has not been measured. ;-)

bigfryo• 3 years ago

So python programmers are destroying the planet.. I just knew it all along

Rochus5• 3 years ago

There is a more recent paper by Pereira et al. from 2021 (see https://www.sciencedirect.com/science/article/abs/pii/S01676... ); if geomean is used (which is the correct method) instead of what they apparently do in the paper, there is yet a different order (see http://software.rochus-keller.ch/Ranking_programming_languag...).

igouy17• 3 years ago

ods ?

"How do you want to open this file?"

Rochus5• 3 years ago

It's OpenOffice format; you should be able to open it with MS Excel too in case you prefer that.

jjaken• 3 years ago

Now compare mental energy

tmaly2• 3 years ago

I can see C having 1.00 in table 4 in the article.

But how does Python get 71.90 and JavaScript 4.45 ?

Something seems off there.

igouy17• 3 years ago

71.90 arithmetic average more than the 10 selected C programs.

6.52 arithmetic average more than the 10 selected C programs.

Seems off because?

kaba010• 3 years ago

JS has a JIT compiler, so for long running programs it can very well execute native machine code in the exact same way as C (plus a GC occasionally meddling with things). Also, it’s not even a bad JIT compiler, huge amount of development went into making it good.

Standard Python is interpreted all the time, going from instruction to instruction giving a layer of abstraction that never disappears.

tmaly2• 3 years ago

I agree a JIT works well for long running programs, but is that the situation we have here with these tests?

Python consumes 38x more energy than Java

Related Stories