Java vs C++ speed (IO & Sorting) - Java

This is a discussion on Java vs C++ speed (IO & Sorting) - Java ; * Razii: > On Wed, 09 Apr 2008 18:49:28 +0200, "Alf P. Steinbach" > <alfps@start.no> wrote: > >> By the way, this has nothing to do with either C++ or Java so is off-topic; a >> language war is off-topic ...

+ Reply to Thread
Page 53 of 54 FirstFirst ... 3 43 51 52 53 54 LastLast
Results 521 to 530 of 536

Java vs C++ speed (IO & Sorting)

  1. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    * Razii:
    > On Wed, 09 Apr 2008 18:49:28 +0200, "Alf P. Steinbach"
    > <alfps@start.no> wrote:
    >
    >> By the way, this has nothing to do with either C++ or Java so is off-topic; a
    >> language war is off-topic in both groups posted to; and using person names in
    >> subject line is also contrary to all Usenet etiquette.

    >
    > There is no such thing as being off topic. You don't make the rules on
    > what is on and off-topic in an unmoderated newsgroup. Do you
    > ununderstand that, asshole?


    Hm, finally. :-)

    Perhaps people now will stop responding to you.

    Hopefully.


    Cheers, & dhth.,

    - Alf

  2. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    On Wed, 09 Apr 2008 20:47:18 +0200, "Alf P. Steinbach"
    <alfps@start.no> wrote:

    >Hm, finally. :-)
    >
    >Perhaps people now will stop responding to you.


    Just as you were able to resist responding to the last post? One thing
    is clear: You aren't the sharpest knife in the drawer.




  3. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    Razii wrote:
    > On Wed, 09 Apr 2008 18:49:28 +0200, "Alf P. Steinbach"
    > <alfps@start.no> wrote:
    >
    >> By the way, this has nothing to do with either C++ or Java so is off-topic; a
    >> language war is off-topic in both groups posted to; and using person names in
    >> subject line is also contrary to all Usenet etiquette.

    >
    > There is no such thing as being off topic. You don't make the rules on
    > what is on and off-topic in an unmoderated newsgroup. Do you
    > ununderstand that, asshole?
    >
    >


    Being unmoderated doesn't mean there are no rules, what it means is that
    there is no enforcement other than posters who repeatedly break the
    rules will find themselves in kill files. They can then continue posting
    messages, but the number of readers will decline.

    Mark Thornton

  4. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    Bo Persson wrote:
    > Stefan Ram wrote:
    >> "Chris Thomasson" <cristom@comcast.net> writes:
    >>> The JVM your using right now was "probably" created with C/C++,
    >>> and of course, assembly language.

    >> There actually are several JVMs written in Java.
    >> Some run on another JVM, some are being compiled
    >> to some other language.
    >>
    >>> So, C/C++ can be used to create a language that performs faster
    >>> than itself?

    >> Late compiling can take advantage of its knowledge of the
    >> execution environment and processor. C++ usually is compiled
    >> early (before distribution).
    >>

    >
    > If you need absolutely maximum performance, you will have to check out
    > the executing environment anyway - C++ or Java. If you end up on a
    > 386, no JIT compiler in the world can compensate for that.


    I've seen code on a '386 running a custom interpreted port of HP Business
    BASIC run faster then HP's own compiled version did on their HP 9000
    mainframe, back in the early 90s. Algorithms trump optimizers every time.

    I don't know what language HP's own port was written in, but ours was written
    in C and ran on the QNX operating system. QNX was a major factor in the
    performance being so good. Intel chips have had some pretty good support for
    multitasking since the '286, and the OS took full advantage of it. (You
    shoulda seen it fly on a '486 or Pentium - whoo-wah!)

    This has been an amusing thread, albeit more than a little silly. I am
    convinced, especially since it was kicked off with essentially a challenge to
    exceed 700 posts to the thread, that the primary purpose of this thread is to
    create Usenet noise.

    To which, yes, I just contributed. Please go back now to your pointless
    discussion already in progress.

    --
    Lew

  5. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    Razii wrote:
    > There is no such thing as being off topic. You don't make the rules on
    > what is on and off-topic in an unmoderated newsgroup. Do you
    > ununderstand that, asshole?


    PLONK, you troll!

    Good work, Razii, you managed in one post to completely invalidate this entire
    thread with one rude, stupid post. You can reply, but I won't see it.

    Put Loser On Newsgroup Killfile. Buh-bye!

    --
    Lew

  6. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    On Wed, 09 Apr 2008 20:12:24 -0400, Lew <lew@lewscanon.com> wrote:

    >PLONK, you troll!


    That's the sprit. Just use filters instead of whine about the threads,
    like I did with that idiot 'stan'.

    My last response was directed at an idiot who said this to me:

    "Silly persons, I pee on you.

    he is indeed not only an idiot but has no clue how Usenet works.



  7. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    In article <5uWdnaXgff8EbmHanZ2dnUVZ_oaonZ2d@comnet>, alfps@start.no
    says...

    [ ... ]

    > Huh? A 386 computer is a Rolls Royce. At least compared to the ordinary 286 PC
    > AT. If you can't make a program fast enough on what is essentially a
    > super-computer (the 386 exceeds the specs for some 1970's supercomputers), then
    > you're just a bad bad & very incompetent programmer. But I expect most are...


    Not really. Supercomputers are normally rated based on floating point
    performance. IIRC, the fastest 386 ran a 40 MHz. With a 387, it still
    needed a minimum of 23 clocks to execute a floating point addition.

    For comparison, a CDC 7600 (first available in 1969) ran at 36 MHz, but
    needed only 4 clocks to execute a floating point addition, so it was
    roughly 5 times as fast. Of course, that's just the very beginning of
    the 1970's -- by the end of the 1970's they'd gotten a LOT faster still.

    --
    Later,
    Jerry.

    The universe is a figment of its own imagination.

  8. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    On Apr 9, 7:05 am, asterisc <Rares....@ni.com> wrote:
    > For small benchmarking problems (which are some structured algorithms)
    > if you compileJavainto ASM (and an exe file) then it may be as fast
    > as C/C++. But then you will loose the portability. For being fast, it
    > need to be native, which isn't portable.


    I do not get your point here. A portable C++ program loses portability
    when you compile it for a particular platform. A Java program loses
    portability if it uses a native library specific to a particular
    platform (e.g. to access Windows registry).

    However you define "portability", programs are not written for the
    sake of being portable, they are written to solve problems. The only
    exception I can think of are programs written specifically to teach
    the concept of portability.

    Finally, Excelsior JET is not the only spec-compliant Java
    implementation with an Ahead-Of-Time compiler. IBM has it in WebSphere
    Realtime:

    http://www.ibm.com/developerworks/ja...tj2/index.html

    (scroll down to "AOT Java compilation")

    LDV

  9. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    On Apr 9, 10:38 am, "Chris Thomasson" <cris...@comcast.net> wrote:
    > >>And the conclusion is also there, once again:
    > >>"thus raising theJavaperformance level up to that of C/C++."

    >
    > > but the common claim in this newsgroup is that c++ is way faster. How
    > > many threads we had? In most cases, the difference was really small.

    >
    > [...]
    >
    > The common claim is not true. Comparing C/C++ toJavais not very
    > productive. When you want to understand the power of low-level languages,
    > vs. sayJava, you can ask yourself a simple question:
    >
    > What language(s) would you use to create a lean, mean high-performanceJava
    > Virtual Machine?
    >
    > The JVM your using right now was "probably" created with C/C++, and of
    > course, assembly language.


    This is true for Sun HotSpot, but not for Excelsior JET, which is
    written mostly in Oberon-2 (AOT compiler) and Modula-2 with some
    assembly (runtime).

    Also, Java with its strong typing is quite a low-level language
    compared to all those fancy scripting languages. If only it it had
    true multidimensional arrays... which C and C++ do not have either, by
    the way.

    For most applications, the only disadvantage of using Java vs. C/C++
    is the size of the JRE. The standard Java SE library is way too
    bloated. Excelsior JET implements a nice trick to reduce that overhead
    (http://www.excelsior-usa.com/java-download-size.html), but not to the
    extent to bring it down to unmanaged C++ level.

    > So, C/C++ can be used to create a language that performs faster than itself?
    > Cool!


    It is always the case with bootstrapped optimizing compilers - you use
    version N to create version N+1 that produces faster code than version
    N.

    LDV

  10. Default Re: Fastest! Counting words (Mirek Fidler.. continues)

    * Jerry Coffin:
    > In article <5uWdnaXgff8EbmHanZ2dnUVZ_oaonZ2d@comnet>, alfps@start.no
    > says...
    >
    > [ ... ]
    >
    >> Huh? A 386 computer is a Rolls Royce. At least compared to the ordinary 286 PC
    >> AT. If you can't make a program fast enough on what is essentially a
    >> super-computer (the 386 exceeds the specs for some 1970's supercomputers), then
    >> you're just a bad bad & very incompetent programmer. But I expect most are...

    >
    > Not really. Supercomputers are normally rated based on floating point
    > performance. IIRC, the fastest 386 ran a 40 MHz. With a 387, it still
    > needed a minimum of 23 clocks to execute a floating point addition.
    >
    > For comparison, a CDC 7600 (first available in 1969) ran at 36 MHz, but
    > needed only 4 clocks to execute a floating point addition, so it was
    > roughly 5 times as fast. Of course, that's just the very beginning of
    > the 1970's -- by the end of the 1970's they'd gotten a LOT faster still.


    Argh, destroying my point by an infusion of facts.

    Would you please stop that.

    This is not that kind of thread!


    Cheers,

    - Alf :-)

+ Reply to Thread
Page 53 of 54 FirstFirst ... 3 43 51 52 53 54 LastLast