Language Benchmarks

marco_polo

Martyr wrote: »

well marco most compilers would be smart enough to see this one, even i saw it before having to compile it

Must deserve some marks for effort. Good spot , Thought I might get ya.

Webmonkey

Martyr wrote: »

well marco most compilers would be smart enough to see this one, even i saw it before having to compile it

Really? which ones?

Martyr

what about a mersenne twister or Lagged Fibonacci generator psuedo random number generator, or even some crypto?

could you implement something like this?

marco_polo

Martyr wrote: »

what about a mersenne twister or Lagged Fibonacci generator psuedo random number generator, or even some crypto?

could you implement something like this?

I could, but I have neither the time or the inclination TBH.

Martyr

webmonkey wrote:

Really? which ones?

GNU C Compiler, MSVC, Pelles c compiler, watcom..most of them really.

its not exactly something you could miss.(unless you're java programmer, who thinks it would be worth running a loop that does nothing)

Martyr

I could, but I have neither the time or the inclination TBH.

yeah, i know and its because we already know that C/C++ is faster than Java.

marco_polo

Martyr wrote: »

GNU C Compiler, MSVC, Pelles c compiler, watcom..most of them really.

its not exactly something you could miss.(unless you're java programmer, who thinks it would be worth running a loop that does nothing)

Ooh, bitchy.

Martyr wrote: »

yeah, i know and its because we already know that C/C++ is faster than Java.

I am sure for the most part it still is marginally, what I don't understand is why you care so much.

Webmonkey

Martyr wrote: »

GNU C Compiler, MSVC, Pelles c compiler, watcom..most of them really.

its not exactly something you could miss.(unless you're java programmer, who thinks it would be worth running a loop that does nothing)

Where did you learn this?

Martyr

Ooh, bitchy.

i'm only kiddin, marco take it easy.

I am sure it still is for the most part, what I don't understand is why you care so much.

because it would be interesting to see where it goes.

this is a programming forum, isn't it?

its not against the rules to do a little programming now and then and show results of your work.

if you don't care, thats fine, you don't have to participate in the discussion.

Martyr

webmonkey wrote:

Where did you learn this?

learn what? are you talking about the loop? or my joke about java programmers?

the loop is obvious, webmonkey..what is there not to understand? it does nothing.
a c compiler won't generate code for that if used properly.

and the joke is just that..trying to be funny.

Webmonkey

Martyr wrote: »

learn what? are you talking about the loop? or my joke about java programmers?

the loop is obvious, webmonkey..what is there not to understand? it does nothing.
a c compiler won't generate code for that if used properly.

and the joke is just that..trying to be funny.

I don't see why a compiler would complain and prevent it from going on its way to assembly, it is syntactically correct, maybe not semantically but definitely syntactically. Have you ever tried it? - I'm still wondering where you learned it?

NeverSayDie

Webmonkey wrote: »

I don't see why a compiler would complain and prevent it from going on its way to assembly, it is syntactically correct, maybe not semantically but definitely syntactically. Have you ever tried it? - I'm still wondering where you learned it?

Modern compilers do all sorts of optimisation in the process of turning high level code into machine code. It won't complain, it'll just rearrange things to work faster. Ditching clearly redundant stuff like empty loops would be an example. See here for all sorts of crazy detail on the topic, if you're into that kind of thing;
http://en.wikipedia.org/wiki/Compiler_optimization

Re where he learnt it, possibly here;
http://www.javaperformancetuning.com/news/qotm028.shtml
http://www.javaperformancetuning.com/news/qotm029.shtml

Webmonkey

NeverSayDie wrote: »

Modern compilers do all sorts of optimisation in the process of turning high level code into machine code. It won't complain, it'll just rearrange things to work faster. Ditching clearly redundant stuff like empty loops would be an example. See here for all sorts of crazy detail on the topic, if you're into that kind of thing;
http://en.wikipedia.org/wiki/Compiler_optimization

Re where he learnt it, possibly here;
http://www.javaperformancetuning.com/news/qotm028.shtml
http://www.javaperformancetuning.com/news/qotm029.shtml

I'm not referring to Java, I'm referring to the statement of "most compilers", the GNU C compiler for example, especially since he making out C/C++ is faster than Java :pac: - Sorry I should have made it clearer.

Thanks for the links however, I am into that sort of thing :pac:

Martyr

NeverSayDie wrote:

Re where he learnt it, possibly here;

no, i didn't see that page before.

webmonkey wrote:

I'm not referring to Java, I'm referring to the statement of "most compilers", the GNU C compiler for example, especially since he making out C/C++ is faster than Java - Sorry I should have made it clearer.

yes, C/C++ is faster imho, would you like to prove me wrong? won't hold my breath

if java compensates for being a **** programmer, why can't c/c++ compilers do the same?

webmonkey wrote:

Thanks for the links however, I am into that sort of thing

hmm, i've a feeling theres alot of you expert programmers insecure about the end result of going against a noob like myself.

i'm not afraid of being proved wrong here, but i know i'm right

put your money where your mouth is, webmonkey..or any of ye for that matter :P

NeverSayDie

Martyr wrote: »

hmm, i've a feeling theres alot of you expert programmers insecure about the end result of going against a noob like myself.

i'm not afraid of being proved wrong here, but i know i'm right

put your money where your mouth is, webmonkey..or any of ye for that matter :P

Why on earth would anyone here be "insecure" about the end results, unless they'd written the compiler themselves or some such? Maybe you could discuss some of the technical reasons why you know you're right, and make things more useful for the OP?

Anima

Here is a good wikiarticle for java performance, where it is good and bad.

http://en.wikipedia.org/wiki/Java_performance#Comparison_to_other_languages

Seems to tie a lot of what I've been reading together.

Martyr

NeverSayDie wrote:

Why on earth would anyone here be "insecure" about the end results, unless they'd written the compiler themselves or some such? Maybe you could discuss some of the technical reasons why you know you're right, and make things more useful for the OP?

if you give me until sunday night or so, i should have enough time to produce some results on my own using both Java / C or C++

The reason i wanted an experienced Java programmer is so that there could be no arguement on the results, i'm not experienced writing Java but i'm sure i could cook up something..(i'm not experienced with C/C++ either, but i'm guessing it should be faster than Java anyway)

Sure, i can write some code in either and show results but chances are some expert on Java will claim to know a way of doing it better..

i'll do it, just don't have time right now, i'm busy in my call centre :pac: no doubt hobbes,webmonkey and others who disagree with me are programming some huge supercomputer using java or .net...."no time" to prove whether i'm right or wrong, just tell me i'm wrong..

Webmonkey

Martyr wrote: »

if you give me until sunday night or so, i should have enough time to produce some results on my own using both Java / C or C++

The reason i wanted an experienced Java programmer is so that there could be no arguement on the results, i'm not experienced writing Java but i'm sure i could cook up something..(i'm not experienced with C/C++ either, but i'm guessing it should be faster than Java anyway)

Sure, i can write some code in either and show results but chances are some expert on Java will claim to know a way of doing it better..

i'll do it, just don't have time right now, i'm busy in my call centre :pac: no doubt hobbes,webmonkey and others who disagree with me are programming some huge supercomputer using java or .net...."no time" to prove whether i'm right or wrong, just tell me i'm wrong..

Why do you assume I am a java expert. I am a C/C++ programmer and I hate working with Java, simply don't like it.

Where did I try to prove you wrong, I was simply saying that the code presented did not get optimized in my GNU compiler and I did try it.
However i did notice this:

for (j = 0; i < cycle1; i++)
    for(j = 0; j < cycle2; j++)

does get optimized while:

for (i = 0; i < cycle1; i++)
  for (j = 0; j < cycle2; j++)
  {
  }

does not.

Martyr

webmonkey wrote:

Why do you assume I am a java expert. I am a C/C++ programmer and I hate working with Java, simply don't like it.

ok, sorry for assuming, but i expected someone programming with C/C++ to know that most C compilers (if used correctly) would not generate machine code for a loop that does nothing.

Martyr

if compiler is used correctly, and i mean using its optimisation features..i don't think either of those loops would generate ANY machine code.

honestly, where would you see something like this used anyway?

its a bit of a bogus arguement.no offence, its not a practical example, its not a fair example to say java is better based on a loop like that ..

NeverSayDie

Martyr wrote: »

if you give me until sunday night or so, i should have enough time to produce some results on my own using both Java / C or C++

The reason i wanted an experienced Java programmer is so that there could be no arguement on the results, i'm not experienced writing Java but i'm sure i could cook up something..(i'm not experienced with C/C++ either, but i'm guessing it should be faster than Java anyway)

Sure, i can write some code in either and show results but chances are some expert on Java will claim to know a way of doing it better..

i'll do it, just don't have time right now, i'm busy in my call centre :pac: no doubt hobbes,webmonkey and others who disagree with me are programming some huge supercomputer using java or .net...."no time" to prove whether i'm right or wrong, just tell me i'm wrong..

Feel free, but that wasn't what I asked. I was interested in learning about the technical reasoning behind you knowing you're right, not millisecond counts from profiling code snippets.

Webmonkey

Martyr wrote: »

if compiler is used correctly, and i mean using its optimisation features..i don't think either of those loops would generate ANY machine code.

honestly, where would you see something like this used anyway?

its a bit of a bogus arguement.no offence, its not a practical, fair assumption to say java is better based on a loop like that ..

for (i = 0; i < cycle1; i++)
  for (j = 0; j < cycle2; j++)
  {
  }

This doesn't and it runs away.

There are times you use loops like this but that include a sleep to cause your program to pause and wait for an interrupt/signal or something.

marco_polo

Martyr wrote: »

if compiler is used correctly, and i mean using its optimisation features..i don't think either of those loops would generate ANY machine code.

honestly, where would you see something like this used anyway?

its a bit of a bogus arguement.no offence, its not a practical example, its not a fair example to say java is better based on a loop like that ..

It is just a joke example dude, using default options I believe it would take the gcc compiler about a month BTW .

Anima

Webmonkey are you using "-O3" when compiling? If there is nothing in the loops, then it should get optimised away.

Webmonkey

Done trick, cheers.

Martyr

NeverSayDie wrote:

Feel free, but that wasn't what I asked. I was interested in learning about the technical reasoning behind you knowing you're right, not millisecond counts from profiling code snippets.

Its a gamble, based on reading mainly, see the wikipedia entry by Anima.
its all based on research done so far, by other people of course but i'll try this myself later out of curiosity.

[PHP]for (i = 0; i < cycle1; i++)
for (j = 0; j < cycle2; j++)
{
}[/PHP]

webmonkey wrote:

There are times you use loops like this but that include a sleep to cause your program to pause and wait for an interrupt/signal or something.

you would use something like that to signal/interrupt? :eek:
not very sophisticated..but what do i know, i work in a call centre :pac:

what about this:

[PHP]
int cycle1,cycle2;

cycle1 = 1000;
cycle2 = 2000;

for (i = 0; i < cycle1; i++)
for (j = 0; j < cycle2; j++)
{
}[/PHP]

what compiler are you using? what options?

marco_polo wrote:

It is just a joke example dude, using default options I believe it would take the gcc compiler about a month BTW

LOL..maybe if you can't use it properly

Martyr

nevermind, Anima answered.

these options i would normally use for GNU c compiler:

-fomit-frame-pointer -O3 -fexpensive-optimizations

also, writing good code helps.

Agner Fog updates a very good manual on optimizing C++

located here: http://www.agner.org/optimize/

marco_polo

Martyr wrote: »

...
but what do i know, i work in a call centre :pac:

Judging from your posts, everything.

Martyr

i do work in a call centre, ask anyone that knows me.

i'm not a programmer by profession, never have been which is why anyone here should feel comfortable proving that Java is faster than C/C++ at generating machine code.

Webmonkey

Marty wrote:

you would use something like that to signal/interrupt?
not very sophisticated..but what do i know, i work in a call centre

How would you do it so? We talking about using os primitives for handling signal handlers etc. You clearly know what your talking about.