Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon. Entire thread

Benchmarks

Name: Anonymous 2010-07-03 8:15

What sites offer multi-language benchmarks besides this
http://shootout.alioth.debian.org/u32/performance.php?test=regexdna
it shows fucking JavaScript as top result..obviously its rigged.

Name: Anonymous 2010-07-03 16:31

>>37
that was my first post in the thread, anonymity is hard ain't it. But I stick by my comment, I have only ever seen the shootout used as a pissing contest online, and never been put to any actual use. this thread could have proven me wrong, but as per usual, it didn't.

Name: Anonymous 2010-07-03 16:36

>>41
It's called a benchmark game for a reason. Get off your high horse ass-hole.

Name: Anonymous 2010-07-03 16:42

>>42
high horse ass-hole.
That's one exquisite sex toy.

Name: Anonymous 2010-07-03 16:43

>>42
Which would take me back to the "the only people who give credence" remark. Now, high ho silver away.

Name: Anonymous 2010-07-03 16:43

>>9
This may surprise you but V8 Irregexp was designed to improve sunspider benchmarks

Name: Anonymous 2010-07-03 16:44

>>42
Why should anyone think you know that reason?

Name: Anonymous 2010-07-03 16:45

>>45
And you know that because?

Name: Anonymous 2010-07-03 17:03

>>45
what does that have to do with >>9 ?

Name: Anonymous 2010-07-03 17:05

>>28
Haha quite edgy.

Name: Anonymous 2010-07-03 17:16

>>41 "I have only ever seen the shootout used as a pissing contest online, and never been put to any actual use."
Maybe that's because of the places you look.
"Several benchmarks from The Computer Language Benchmark Game were translated to Thorn and their runtime compared to the existing benchmarks running on Python 2.5."
"Thorn: robust, concurrent, extensible scripting on the JVM" OOPSLA '09
http://portal.acm.org/citation.cfm?id=1639949.1640098

Name: Anonymous 2010-07-03 17:36

>>1 "it shows fucking JavaScript as top result..obviously its rigged"

Not.
http://shootout.alioth.debian.org/u32/which-programming-languages-are-fastest.php

Name: Anonymous 2010-07-03 17:42

ooh im going to write an OS in javascript now because its obviously the sane choice

Name: Anonymous 2010-07-03 17:47

>>52
Why are you so mad that some javascript version of program may be faster than some other language's version of that program? What does it matter?

Name: Anonymous 2010-07-03 18:03

>>53
what im not mad

Name: Anonymous 2010-07-03 18:04

>>53
It's a good demonstration of the fact that these shoot-outs aren't just comparing language implementations, they're also comparing algorithm implementations, which renders them nearly worthless.

Name: Anonymous 2010-07-03 18:09

>>55
Do these programs implement different algorithms?
Java 6 -server    23.70
C++ GNU g++    23.82
http://shootout.alioth.debian.org/u32/performance.php?test=nbody

Name: Anonymous 2010-07-03 18:22

>>55
Watsamatta, did Racket perform better than your favorite language?

Name: Anonymous 2010-07-03 18:31

>>51
GHC is fourth after Sepples, C and Java? What the hell?

Name: Anonymous 2010-07-03 18:33

>>57
Many Schemes have fairly good performance. Ikarus, Gambit and probably Larceny should perform better than Racket and if you are willing to wait (and use a R4RS scheme) Stalin provides some really well optimised code.

Name: Anonymous 2010-07-03 18:33

>>58
Wait 'til you plot LuaJIT.

Name: Anonymous 2010-07-03 18:42

>>58
Where is GHC fourth after Sepples, C and Java?

Name: Anonymous 2010-07-03 19:38

>>56
The Sepples version is OMG OPTIMIZED based on assumptions that clearly don't apply to the machine the tests are run on, so yes.

(Or more exactly, they don't implement different algorithms, depending on your definitions, but they implement the same algorithm in different ways, and there lies the rub.)

Name: Anonymous 2010-07-03 19:43

It'd be interesting to write one of the examples in R4RS scheme then use stalin, and compare it to the C version.

But I don't care that much to do it myself.

Name: Anonymous 2010-07-03 20:06

>>62
Specifically what about the Sepples version doesn't apply to the machine the tests are run on?

If the programs didn't implement the same algorithm in different ways then we'd be left with Fortran written in Python, Fortran written in Java, Fortran written in Lisp, ...

Name: Anonymous 2010-07-03 20:07

>>64
I hope this obtusity is deliberate.

Name: Anonymous 2010-07-03 20:12

>>65
Be specific - it avoids unnecessary confusion.

Name: Anonymous 2010-07-03 20:13

>>66
SPECIFY MY ANUS

Name: Anonymous 2010-07-04 4:11

>>58
http://shootout.alioth.debian.org/fastest-programming-language.php
I can't find that particular result now but GHC is in the top 5 on all four architectures, just behind C, C++ and comparable to C# and Java. I did not expect an exotic language like Haskell to be so fast but, then again, I expect an average Haskell programmer to be much more competent than an average Java programmer.

Name: Anonymous 2010-07-04 4:51

>>68
What makes it exotic, you don't know it?

Name: Anonymous 2010-07-04 5:14

>>69
He should have said academic.

Name: FrozenVoid 2010-07-04 5:20

>>69
He meant "esoteric".



__________________
Orbis terrarum delenda est

Name: Anonymous 2010-07-04 12:29

>>68 "I did not expect an exotic language like Haskell to be so fast"
Which is another example of that website being put to actual use - data we can use to check our expectations.

What about "exotic" languages like ATS and Scala and F# ? (Is Ada "exotic"?)

Name: Anonymous 2010-07-05 4:33

Did google paid off the Lunix fags to exclude D(and include Golang and V8), since i was swearing i saw it a year ago at their site(and google cache is empty)?
D can only bee seen in the web archives, all benchmarks are gone.
http://web.archive.org/web/20060101191642/http://shootout.alioth.debian.org/

Name: FrozenVoid 2010-07-05 4:40

>>73 Yeah, there is something fishy about excluding D.
Their last archived page is from 2008(and Golang appeared in 2009) and it does include D. In general its nearly as fast as C++ or Ocaml, which should get into first 1-5 places on benchmarks.



__________________
Orbis terrarum delenda est

Name: Anonymous 2010-07-05 4:50

They should also include Stalin Scheme instead of the crappy interpreters.

Name: Anonymous 2010-07-05 5:02

>>73,74
It's because the shootout changed to 4 different computers instead of 1, and apparently DMD, or the D code written for the shootout, was not portable.

I don't really know why Walter Bright isn't doing what he needs to get D back in. D is so similar to C++ that people won't be willing to sacrifice performance for slightly better syntax. He should be trying to prove his performance claims.

>>75
Yes they should. The Stalin folks have made outrageous performance claims about its performance compared to C, despite the fact that Stalin compiles to C. This is because they compare it to polymorphic C code using function pointers rather than properly optimized code.

Name: Anonymous 2010-07-05 5:08

>>76
So you didn't read the whole mailing list thread then? It gave you all the information you fucking well wanted, but here we are
>>75
Stop baiting them

Name: Anonymous 2010-07-05 5:22

This thread has been closed and replaced with the following thread:

Subject: Compiling Haskell code with an Erlang compiler
Email:

It doesn't work.

Name: Anonymous 2010-07-05 5:27

>>78
Erlang isn't a compiler.

Name: Anonymous 2010-07-05 5:28

>>78
Just append "." to every line.
Newer Posts
Don't change these.
Name: Email:
Entire Thread Thread List
All trademarks and copyrights on this site are owned by their respective parties. All comments and posts are the responsibility of their own posters.
- Tinychan + 2ch Mode -