Performance win32 vs x64 on Windows XP x64 professional

Discussion in 'Windows 64bit' started by Eric, Jan 17, 2007.

  1. Eric

    Eric Guest

    Hi all,

    I did a performance test with the same application that I compile with
    VS2005 for win32 and x64. I run it on Windows XP x64 professional SP1.
    The computer is a Intel Pentium D 3 GHz with 2 GB RAM.

    The purpose of the test was to identify to gain in performance of an
    application 64-bit vs 32-bit. My result is the win32 version is far
    better than the win64 version. How can this be possible? All win32
    application are emulate in 64-bits? An 64-bit application use all the
    power (resource) available?

    What can explain this result? Will I get the same result between Vista
    32-bit VS Vista 64-bit?

    P.S.: I clean the cache between each test...

    Regards,
    Eric
    Eric, Jan 17, 2007
    #1
    1. Advertising

  2. The explanation could be Compiler Options?

    It is important, however, to establish that running 64bit is not one bit
    faster - if you need to push more data through the system 64bit will do it
    in less time, not because the system is faster but because it has a wider
    data-path. The short of all this is that if you need it you will think it is
    faster, if you don't need it - you won't see the advantage!

    But then, even if you don't need it - as I don't, I think it is a nice
    polished system that is more stable than anything else I've had.

    Tony. . .


    "Eric" <> wrote in message
    news:...
    > Hi all,
    >
    > I did a performance test with the same application that I compile with
    > VS2005 for win32 and x64. I run it on Windows XP x64 professional SP1.
    > The computer is a Intel Pentium D 3 GHz with 2 GB RAM.
    >
    > The purpose of the test was to identify to gain in performance of an
    > application 64-bit vs 32-bit. My result is the win32 version is far
    > better than the win64 version. How can this be possible? All win32
    > application are emulate in 64-bits? An 64-bit application use all the
    > power (resource) available?
    >
    > What can explain this result? Will I get the same result between Vista
    > 32-bit VS Vista 64-bit?
    >
    > P.S.: I clean the cache between each test...
    >
    > Regards,
    > Eric
    >
    Tony Sperling, Jan 17, 2007
    #2
    1. Advertising

  3. Eric

    Dshai Guest

    Agreed Tony, although I do occasionally see the difference as well.

    --

    Dshai

    Life is only limited by those living it...

    "Eric" <> wrote in message
    news:...
    > Hi all,
    >
    > I did a performance test with the same application that I compile with
    > VS2005 for win32 and x64. I run it on Windows XP x64 professional SP1.
    > The computer is a Intel Pentium D 3 GHz with 2 GB RAM.
    >
    > The purpose of the test was to identify to gain in performance of an
    > application 64-bit vs 32-bit. My result is the win32 version is far
    > better than the win64 version. How can this be possible? All win32
    > application are emulate in 64-bits? An 64-bit application use all the
    > power (resource) available?
    >
    > What can explain this result? Will I get the same result between Vista
    > 32-bit VS Vista 64-bit?
    >
    > P.S.: I clean the cache between each test...
    >
    > Regards,
    > Eric
    >
    Dshai, Jan 18, 2007
    #3
  4. Yes, me too - but keeping true to the analytical mindset of this society [
    snicker. . .] what is it that we are seeing? most of the time it is not the
    64bit effect but the effect of running a well greased system - I've said
    this before, and I am not about to revert this opinion. XP x64 is VASTLY
    better than any consumer OS when it comes to managing Virtual Resources -
    this is subjectively speeding up the OS considerably, you may not be able to
    verify this statement from any benchmarking, but we, the users we sit down
    and watch it every time we boot the machine - the 64bits, most of us never
    see!!!


    Tony. . .


    "Dshai" <> wrote in message
    news:...
    > Agreed Tony, although I do occasionally see the difference as well.
    >
    > --
    >
    > Dshai
    >
    > Life is only limited by those living it...
    >
    > "Eric" <> wrote in message
    > news:...
    > > Hi all,
    > >
    > > I did a performance test with the same application that I compile with
    > > VS2005 for win32 and x64. I run it on Windows XP x64 professional SP1.
    > > The computer is a Intel Pentium D 3 GHz with 2 GB RAM.
    > >
    > > The purpose of the test was to identify to gain in performance of an
    > > application 64-bit vs 32-bit. My result is the win32 version is far
    > > better than the win64 version. How can this be possible? All win32
    > > application are emulate in 64-bits? An 64-bit application use all the
    > > power (resource) available?
    > >
    > > What can explain this result? Will I get the same result between Vista
    > > 32-bit VS Vista 64-bit?
    > >
    > > P.S.: I clean the cache between each test...
    > >
    > > Regards,
    > > Eric
    > >

    >
    >
    Tony Sperling, Jan 18, 2007
    #4
  5. Eric

    Eric Guest

    I understand your point. Both projects (win32 and win64) have to same
    optimization options. But the result show clearly that the test (load
    CPU and memory access) was faster in 32-bit. Normaly like you said, in
    64-bit I should get better performance when we use a lot a memory
    access and with just load CPU the result should be similar.

    Maybe is a problem with some system resources (drivers motherboard,
    video card, etc) that are much optimize in win32 mode and not in win64.

    Eric

    Tony Sperling wrote:
    > Yes, me too - but keeping true to the analytical mindset of this society [
    > snicker. . .] what is it that we are seeing? most of the time it is not the
    > 64bit effect but the effect of running a well greased system - I've said
    > this before, and I am not about to revert this opinion. XP x64 is VASTLY
    > better than any consumer OS when it comes to managing Virtual Resources -
    > this is subjectively speeding up the OS considerably, you may not be able to
    > verify this statement from any benchmarking, but we, the users we sit down
    > and watch it every time we boot the machine - the 64bits, most of us never
    > see!!!
    >
    >
    > Tony. . .
    >
    >
    > "Dshai" <> wrote in message
    > news:...
    > > Agreed Tony, although I do occasionally see the difference as well.
    > >
    > > --
    > >
    > > Dshai
    > >
    > > Life is only limited by those living it...
    > >
    > > "Eric" <> wrote in message
    > > news:...
    > > > Hi all,
    > > >
    > > > I did a performance test with the same application that I compile with
    > > > VS2005 for win32 and x64. I run it on Windows XP x64 professional SP1.
    > > > The computer is a Intel Pentium D 3 GHz with 2 GB RAM.
    > > >
    > > > The purpose of the test was to identify to gain in performance of an
    > > > application 64-bit vs 32-bit. My result is the win32 version is far
    > > > better than the win64 version. How can this be possible? All win32
    > > > application are emulate in 64-bits? An 64-bit application use all the
    > > > power (resource) available?
    > > >
    > > > What can explain this result? Will I get the same result between Vista
    > > > 32-bit VS Vista 64-bit?
    > > >
    > > > P.S.: I clean the cache between each test...
    > > >
    > > > Regards,
    > > > Eric
    > > >

    > >
    > >
    Eric, Jan 18, 2007
    #5
  6. Eric

    Eric Guest

    Hi all,

    I did the same test with Vista 32-bit and Vista x64 (in win32 mode and
    win64 mode) always on the same computer. The results are very
    interesting.

    Here is the results in order : the better to the worst

    1- Windows XP 32-bit
    2- Windows XP x64 (win32 mode)
    3- Vista 32-bit, Vista x64 (win32 and win64 mode) => all same results
    4- Windows XP x64 (win64 mode)

    Like expected, the performance of Vista is lower than Windows XP
    caused by the new graphics Aero, but Windows XP x64 in win64 mode have
    the worst performance and it is far away from Vista. So I think the
    Windows XP x64 is a turnip. The only good think that Windows XP x64
    have is to permit the test application in win64 mode, but do not
    expect have better performance by switching your application in win64
    mode.

    Regards,
    Eric

    On Jan 18, 12:59 pm, "Eric" <> wrote:
    > I understand your point. Both projects (win32 and win64) have to same
    > optimization options. But the result show clearly that the test (load
    > CPU and memory access) was faster in 32-bit. Normaly like you said, in
    > 64-bit I should get better performance when we use a lot a memory
    > access and with just load CPU the result should be similar.
    >
    > Maybe is a problem with some system resources (drivers motherboard,
    > video card, etc) that are much optimize in win32 mode and not in win64.
    >
    > Eric
    >
    > Tony Sperling wrote:
    > > Yes, me too - but keeping true to the analytical mindset of this society [
    > > snicker. . .] what is it that we are seeing? most of the time it is not the
    > > 64bit effect but the effect of running a well greased system - I've said
    > > this before, and I am not about to revert this opinion. XP x64 is VASTLY
    > > better than any consumer OS when it comes to managing Virtual Resources -
    > > this is subjectively speeding up the OS considerably, you may not be able to
    > > verify this statement from any benchmarking, but we, the users we sit down
    > > and watch it every time we boot the machine - the 64bits, most of us never
    > > see!!!

    >
    > > Tony. . .

    >
    > > "Dshai" <> wrote in message
    > >news:...
    > > > Agreed Tony, although I do occasionally see the difference as well.

    >
    > > > --

    >
    > > > Dshai

    >
    > > > Life is only limited by those living it...

    >
    > > > "Eric" <> wrote in message
    > > >news:...
    > > > > Hi all,

    >
    > > > > I did a performance test with the same application that I compile with
    > > > > VS2005 for win32 and x64. I run it on Windows XP x64 professional SP1.
    > > > > The computer is a Intel Pentium D 3 GHz with 2 GB RAM.

    >
    > > > > The purpose of the test was to identify to gain in performance of an
    > > > > application 64-bit vs 32-bit. My result is the win32 version is far
    > > > > better than the win64 version. How can this be possible? All win32
    > > > > application are emulate in 64-bits? An 64-bit application use all the
    > > > > power (resource) available?

    >
    > > > > What can explain this result? Will I get the same result between Vista
    > > > > 32-bit VS Vista 64-bit?

    >
    > > > > P.S.: I clean the cache between each test...

    >
    > > > > Regards,
    > > > > Eric
    Eric, Jan 28, 2007
    #6
  7. I am sorry, but your 'results' -if we can call it that- contradicts two
    years of combined experience in this group, as far as I can make out. You
    have given no indication of what data you are working with, and no
    indication of how you wrote your timing routine(s), and no indication of
    what language you used. Furthermore, you are not telling us if any of the
    results actually seem to make any sense, when compared? Or if under
    analysis, they reveal some component of randomness? True benchmarking should
    be using proven scientific method, and should be written in machine-code, or
    it is of no more value than ordinary entertainment.

    I don't doubt that you have been digging into the issue with considerable
    iinterest and with much effort, and you may even have uncovered something of
    great general interest to the group - but what?

    I think you should publish your results to a developer's group, they will
    have people present, with the basic technology in shape, who might be more
    able to appreciate your work.

    But they probably would like to know how you have been treating
    Caching-Schemes and the Virtual Memory Subsystem, and stuff like that, which
    is vastly different across those OS's.


    Tony. . .


    "Eric" <> wrote in message
    news:...
    > Hi all,
    >
    > I did the same test with Vista 32-bit and Vista x64 (in win32 mode and
    > win64 mode) always on the same computer. The results are very
    > interesting.
    >
    > Here is the results in order : the better to the worst
    >
    > 1- Windows XP 32-bit
    > 2- Windows XP x64 (win32 mode)
    > 3- Vista 32-bit, Vista x64 (win32 and win64 mode) => all same results
    > 4- Windows XP x64 (win64 mode)
    >
    > Like expected, the performance of Vista is lower than Windows XP
    > caused by the new graphics Aero, but Windows XP x64 in win64 mode have
    > the worst performance and it is far away from Vista. So I think the
    > Windows XP x64 is a turnip. The only good think that Windows XP x64
    > have is to permit the test application in win64 mode, but do not
    > expect have better performance by switching your application in win64
    > mode.
    >
    > Regards,
    > Eric
    >
    > On Jan 18, 12:59 pm, "Eric" <> wrote:
    > > I understand your point. Both projects (win32 and win64) have to same
    > > optimization options. But the result show clearly that the test (load
    > > CPU and memory access) was faster in 32-bit. Normaly like you said, in
    > > 64-bit I should get better performance when we use a lot a memory
    > > access and with just load CPU the result should be similar.
    > >
    > > Maybe is a problem with some system resources (drivers motherboard,
    > > video card, etc) that are much optimize in win32 mode and not in win64.
    > >
    > > Eric
    > >
    > > Tony Sperling wrote:
    > > > Yes, me too - but keeping true to the analytical mindset of this

    society [
    > > > snicker. . .] what is it that we are seeing? most of the time it is

    not the
    > > > 64bit effect but the effect of running a well greased system - I've

    said
    > > > this before, and I am not about to revert this opinion. XP x64 is

    VASTLY
    > > > better than any consumer OS when it comes to managing Virtual

    Resources -
    > > > this is subjectively speeding up the OS considerably, you may not be

    able to
    > > > verify this statement from any benchmarking, but we, the users we sit

    down
    > > > and watch it every time we boot the machine - the 64bits, most of us

    never
    > > > see!!!

    > >
    > > > Tony. . .

    > >
    > > > "Dshai" <> wrote in message
    > > >news:...
    > > > > Agreed Tony, although I do occasionally see the difference as well.

    > >
    > > > > --

    > >
    > > > > Dshai

    > >
    > > > > Life is only limited by those living it...

    > >
    > > > > "Eric" <> wrote in message
    > > > >news:...
    > > > > > Hi all,

    > >
    > > > > > I did a performance test with the same application that I compile

    with
    > > > > > VS2005 for win32 and x64. I run it on Windows XP x64 professional

    SP1.
    > > > > > The computer is a Intel Pentium D 3 GHz with 2 GB RAM.

    > >
    > > > > > The purpose of the test was to identify to gain in performance of

    an
    > > > > > application 64-bit vs 32-bit. My result is the win32 version is

    far
    > > > > > better than the win64 version. How can this be possible? All

    win32
    > > > > > application are emulate in 64-bits? An 64-bit application use all

    the
    > > > > > power (resource) available?

    > >
    > > > > > What can explain this result? Will I get the same result between

    Vista
    > > > > > 32-bit VS Vista 64-bit?

    > >
    > > > > > P.S.: I clean the cache between each test...

    > >
    > > > > > Regards,
    > > > > > Eric

    >
    Tony Sperling, Jan 29, 2007
    #7
  8. Eric

    Jud Hendrix Guest

    On Mon, 29 Jan 2007 02:50:38 +0100, "Tony Sperling"
    <> wrote:

    >I am sorry, but your 'results' -if we can call it that- contradicts two
    >years of combined experience in this group,


    Tally ho! In my experience, XP64 in 64-bit mode, runs the best of them all,
    but I have to admit, I have very limited experience with Vista (and not
    used it under multi-media stress). But given the heavy system-requirements,
    i don't expect it to perform better, as a nice look will come with a loss.

    jud
    Jud Hendrix, Jan 29, 2007
    #8
  9. Eric

    Eric Guest

    My purpose wasn't to evaluate in a precise manner the performance. I
    just want to know if the gain in performance of an
    application 64-bit vs 32-bit on Windows XP x64 and Vista x64 is
    visible (so my test unit is seconds !!). Also, I want to evaluate the
    performance of a typical use of my software. I publish my result to
    start a debate on the possible causes of the performance problem with
    Windows XP x64.

    When you buy a car, you try it. You want to know if you like to drive
    it. So I want to know what can explain my lost of RPM with Windows XP
    x64 in 64-bit mode. I was expected result like I had with Vista x64
    i.e. similary performance between 32-bit and 64-bit mode.

    Regards,
    Eric

    On Jan 29, 3:25 am, Jud Hendrix <> wrote:
    > On Mon, 29 Jan 2007 02:50:38 +0100, "Tony Sperling"
    >
    > <> wrote:
    > >I am sorry, but your 'results' -if we can call it that- contradicts two
    > >years of combined experience in this group,Tally ho! In my experience, XP64 in 64-bit mode, runs the best of them all,

    > but I have to admit, I have very limited experience with Vista (and not
    > used it under multi-media stress). But given the heavy system-requirements,
    > i don't expect it to perform better, as a nice look will come with a loss.
    >
    > jud
    Eric, Jan 29, 2007
    #9
  10. Well, personally I cannot evaluate the results since I don't know what you
    have been doing, but if you have a dual-boot setup, consider this:

    Create a partition that is available to both Systems

    Create random size files up to 60% of the available space.

    Randomly delete half of those files.

    Repeat the file creation.

    Back up the partition.

    // This should have the partition being wildly Fragmented

    Get a Defragger that is released in both 32bit and 64bit versions.

    Run both versions while restoring the backup between runs.

    This is not scientific by any means, and may not be any more usefull than
    your own method, but it should give you an indication of how the user
    typically would experience the system at work, it would also be easier to
    time, perhaps. Only suggesting this to give you a chance of getting possibly
    different indications.

    Also, have you checked your present algorithm if the timing routine you use
    can resolve any usefull information, yours wouldn't be the first to return
    with contradicting results. Seconds are pretty long inside a CPU.


    Tony. . .


    "Eric" <> wrote in message
    news:...
    > My purpose wasn't to evaluate in a precise manner the performance. I
    > just want to know if the gain in performance of an
    > application 64-bit vs 32-bit on Windows XP x64 and Vista x64 is
    > visible (so my test unit is seconds !!). Also, I want to evaluate the
    > performance of a typical use of my software. I publish my result to
    > start a debate on the possible causes of the performance problem with
    > Windows XP x64.
    >
    > When you buy a car, you try it. You want to know if you like to drive
    > it. So I want to know what can explain my lost of RPM with Windows XP
    > x64 in 64-bit mode. I was expected result like I had with Vista x64
    > i.e. similary performance between 32-bit and 64-bit mode.
    >
    > Regards,
    > Eric
    >
    > On Jan 29, 3:25 am, Jud Hendrix <> wrote:
    > > On Mon, 29 Jan 2007 02:50:38 +0100, "Tony Sperling"
    > >
    > > <> wrote:
    > > >I am sorry, but your 'results' -if we can call it that- contradicts two
    > > >years of combined experience in this group,Tally ho! In my experience,

    XP64 in 64-bit mode, runs the best of them all,
    > > but I have to admit, I have very limited experience with Vista (and not
    > > used it under multi-media stress). But given the heavy

    system-requirements,
    > > i don't expect it to perform better, as a nice look will come with a

    loss.
    > >
    > > jud

    >
    Tony Sperling, Jan 30, 2007
    #10
    1. Advertising

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Harold

    XP professional vs. WINDOWS 2000 Professional

    Harold, Dec 12, 2003, in forum: Microsoft Certification
    Replies:
    4
    Views:
    1,009
    Zenner
    Dec 15, 2003
  2. D@Z
    Replies:
    5
    Views:
    729
    Liza Smorgaborgsson
    Jan 30, 2006
  3. jamesa01
    Replies:
    2
    Views:
    458
    Steve
    Feb 27, 2006
  4. =?Utf-8?B?UGFzY2Fs?=
    Replies:
    6
    Views:
    588
    Wayne Wastier
    Sep 25, 2005
  5. James Kosin

    Re: win32 replacement for edit.com in x64 windows

    James Kosin, Feb 18, 2010, in forum: Windows 64bit
    Replies:
    2
    Views:
    3,019
    Bobby Johnson
    Feb 19, 2010
Loading...

Share This Page