BENCHMARKS GAME POZZED

No, because pthreads aren't C, but C11 threads would be allowed. If concurrency is part of the language and it happens to use pthreads, it would be allowed. GMP would not be allowed, but if big numbers are part of the language and they happen to be implemented with GMP, it would be allowed.

These do it wrong, according to what I wish the rules were, but this is what most of the benchmarks do.
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=gcc&id=1
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=gpp&id=4
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=racket&id=2
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=sbcl&id=4
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=gnat&id=2
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=go&id=3
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=java&id=2
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=fpascal&id=3

These do it right. OCaml big_int is part of the standard OCaml library.
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=vw&id=4
benchmarksgame.alioth.debian.org/u64q/program.php?test=pidigits&lang=ocaml&id=3

There are so many programs that use libraries that are not part of the language, which isn't a fair comparison of language speed.

That site had been gamed incredibly hard and the guy probably got tired of people taking it as gospel. There are all sorts of ways to cheat at microbenchmarks. For example, you can unroll fucking everything and as long as the tiny benchmark fits in the icache it looks fast, but real-world performance will be shit as you destroyed the icache.

Microbenchmarks stopped being relevant ages ago. Every implementation is designed with twisting the compiler to unrealistically output instructions as close to the C implementation as possible. If compiler authors started taking benchmarks seriously, they'd end up implementing retarded javascript shit like Date optimizations or a fucking Math.sin() cache.

Thanks for highlighting specific tests, having a look at them I wouldn't of assumed that such code was allowed.

I'd really like to see a set of tests that use code a normal dev would output; this highly optimised junk really does seem like cheating.

Having read through some of the racket and chicken scheme documentation both seem to concede that if you want performance, you're going to have to do some weird shit.

...

The old site used to have charts and comparison of arbitrary languages.
The current site has nothing of that it's dumbed down as fuck and amazingly useless.

They still autobuild charts comparing each language.

benchmarksgame.alioth.debian.org/u64q/which-programs-are-fastest.html

Benchmarks game website has lots of retarded code. I remember there was one in c that was one of the fastest. And then the c++ version one of the slowest thanks to some retard using a slow boost library. Sometimes I suspect boost abd modern c++ cultists are the ptiduct if some social engineering to make c++ less relevant.

THE SJWS CANT KEEP GETTING AWAY WITH THIS

WHY IS EVERYONE HERE DOING DAMAGE CONTROL?

I guess It's time to go there and make sure to write code in real languages that is faster than meme languages.

reminder that benchmarks are the ONLY sacred thing.

everything else is a lie. scripting vs compiled. static vs dynamic typing. object oriented or functionaly. java vs rust vs python vs go. ALL OF THIS IS FAKE THEORETICAL BULLSHIT.

THE ONLY THING THAT IS REAL IS TAKING TWO PROGRAMS AND STARTING THEM OFF AT THE SAME TIME AND RACING THEM.

We should start a site where programmers write code and people bet on which one will be faster, like horse races

Are you talking about programmers competing who will solve problem faster or whose code would take less time to get answer?

It's ridiculous. The only way to do this is contest-style in real time where there is no disassembly allowed. See how the code people actually write with these languages performs.

What in the fuck..

Why not both?

Yeah, I "like" c++, but I have to admit they have a community problem, with all their boost prozelitizing and modern c++ extremism, even when it obviously makes them lose the edge they have over other languages.

What's the point of writing c++ code, and bear with all the bullshit when you end up including bloated libraries and having retarded discussions when obvious performance problems are treated as negligible because they have a hard on for wathever is new and shiny.

The only way to fix c++ right now is to cut stuff. not to add more.

Worst of all, there are lots of things that c++ could fix over plain c and they keep doing these horrid mistakes, they keep, even today, duplicating plain c features and offering "modern ones" that in some way end up being WORSE than the originals.

Not a good claim. I develop in C for low latency projects. We spend days tackling the most efficient way to code a particular piece of code. dod, benchmarks, asm, simd, etc.

I care very little if X language can achieve a really good speed faster than Y language... if Y language can reach absolutely the best speed if you invest 10 times more hours and Y can't.

I'm not going to say that the info you want might not be useful. But frankly we have too many people acting as if performance isn't important. (performance matters very little to you if you can say "performance is important but," or if you subtly redefine the speed of a language as "implementation speed"). We need benchmarks to disregard implementation time, because we will stuck with one trick pony meme languages that can't achieve max speed if not.

These people are snake oil sellers. They can't even bother to write benchmarks properly.

Doing the work in classes are safer. That's why the C++ standard wraps all the C functions in classes for C++.

Have you even looked at the Fortran benchmarks, faggot? There's at least one benchmark missing and another one or two are single threaded when the competition is multithreaded. Obviously not an apples-to-apples comparison. The Benchmarks Game is much more a measure of how much effort has gone into optimising benchmarks rather than accurate language to language comparisons. The Ada lines of code counts back in the day weren't representative either (tons of FFI import lines).

this /thread
too bad the conclusion your typical faggot will derive from this is
while missing the entire point that there's no practical reason to care which PL is "faster" than the other, unless we're talking about something like your typical imperative PL vs Prolog

...and then we could garbage collect it to reduce errors.
And now you've got Go.

C++ is fine. Those of us that do it for a living already know what to avoid like template metaprogramming, streams, 'using', etc.. It needs more, not less. We got move support which was a big deal as most of C++ was sitting in the "never use this" category until that, and we still need template constraints, a total redesign/rewrite of streams, and a new debugging format for ELF that handles templates properly.

If there are complete features that you need to avoid then it does need things removed. Doesn't mean other things shouldn't be added.

Not really.

I am one of those people who have only basic knowledge about C++.
What is wrong with streams?

There are positives and negatives. C++ streams are typically a bit slower, generates larger code, and integrates much worse with localization (Using `printf("You have eaten %d dogs, and only %d are left in your %s", eatendogs, dogsleft, location)` is a hell of a lot easier to localize than c++ streams doing `cout

In some cases they're actually calling a virtual method per character written due to the insane design of streambuf (especially with iterator templates). The whole thing is a dumpster fire.
I like how you list the ways that they are unusable then suggest people saying they are unusable are wrong. The inability to internationalize them or use them sanely with unicode would on their own get them banned at every large company without all the other ways they're cancer.

If you were a cuteboi I'd objectively jiggle you, user.

This needs another (You)
It's getting ridiculous the amount of plain cheating nowadays.

ladies and gentlemen this here is a CIA NSA FBI PSYOP SHILL POST.

Even worse, faggots working outside their fields of knowledge already took it away long ago. But that's basically any tech talk.