1. Liebe Forumsgemeinde,

    aufgrund der Bestimmungen, die sich aus der DSGVO ergeben, müssten umfangreiche Anpassungen am Forum vorgenommen werden, die sich für uns nicht wirtschaftlich abbilden lassen. Daher haben wir uns entschlossen, das Forum in seiner aktuellen Form zu archivieren und online bereit zu stellen, jedoch keine Neuanmeldungen oder neuen Kommentare mehr zuzulassen. So ist sichergestellt, dass das gesammelte Wissen nicht verloren geht, und wir die Seite dennoch DSGVO-konform zur Verfügung stellen können.
    Dies wird in den nächsten Tagen umgesetzt.

    Wir danken allen, die sich in den letzten Jahren für Hilfesuchende und auch für das Forum selbst engagiert haben.

Pymol für G5

Dieses Thema im Forum "Software" wurde erstellt von MacELCH, 19. September 2003.

  1. MacELCH

    MacELCH New Member

    Ob da wohl ein Mac-Herz schlägt ?

    Dies ist die Kopie einer Email aus einem Wissenschaftlichen Forum, es paßt aber auch hier ganz gut rein:

    PyMOLer's & Friends,

    For those of you considering a new PowerMac G5, a special
    "native" PyMOL v0.91 for the G5 with some nice demonstrations has been
    released: http://pymol.sf.net. This version contains an assortment of
    G5 optimizations, and the next release will have even more. Current G4
    users will benefit from these as well.

    Although Apple's top-of-the-line 2.0 Ghz G5 systems aren't quite
    shipping yet, they did let us run some PyMOL benchmarks on one, and the
    numbers were very impressive. In my own hands, a dual 2.0 Ghz G5
    effectively competes with a dual 3.0 Ghz Pentium 4 Xeon for generating
    PyMOL ray-traced movies in head-to-head comparisons. OpenGL speed is
    also top notch. So in my opinion, Mac/PC performance parity has finally
    been achieved -- congratulations Apple!

    By the way, Steve Jobs' controversial "twice as fast as a PC"
    claim is only valid for code exploiting the Altivec vectorization unit.
    PyMOL doesn't use that, but it can still render images equally as fast
    on the dual-processor Mac as on the dual-processor PC. "Once as fast as
    a PC is fast enough", I say, without any magic code.

    Thus, the really good news for cross-platform developers (like
    me) is that it is not necessary to write G5-specific code in order to
    get good performance. In fact, most of the "G5 optimizations" we put
    into PyMOL have equally helped the PC version (much to Apple's chagrin).
    Code optimization is simply easier on a Mac using Apple's great free
    tools (especially Shark
    http://developer.apple.com/technotes/tn/tn2086.html#useshark ), and the
    resulting code runs measurably faster on every platform.

    So if you are looking to buy a fast dual-processor desktop
    system, you may want to take a look at the G5 once the dual 2 Ghz
    machines ship in a few weeks. Though you'll probably pay more for this
    Mac than you will for a no-name x86 Linux setup, you will get concrete
    value in return:

    - freedom from Linux hardware configuration and OpenGL driver
    hassles
    - accountable vendor support
    - a user-friendly operating system
    - desktop applications such as Office, Photoshop, and
    Illustrator.
    - awesome developer tools (at no additional cost!)
    - 64 bits
    - stereo 3D graphics capability
    - and of course, the unique Mac OS X "look and feel".

    Plus, since Mac OS X is Unix, most computational chemistry,
    bioinformatics, and structural biology research software already runs
    well on it.

    Whatever your platform disposition, it is absolutely in your
    interest to learn about the many new options which exist before your
    next big purchase. With ridiculously fast Pentium 4, G5, Athlon,
    Itanium2, and Opteron hardware now entering the market, and with at
    least four affordable or free operating systems to choose from, there
    are plenty of ways to build a great system for relatively little money.
    You should not spend the cost of a new car just to buy a desktop
    computer (sadly, some companies I know STILL do this...).

    These days, PyMOL can compile and run virtually anywhere, but
    our focus will continue to be on x86 Windows, x86 Linux, and G4/G5 Macs,
    with IRIX and Solaris support continuing as long as demand persists.
    Please do speak up if/when native Itanium2 and Opteron binaries are
    needed...

    Does this sound like an Apple commercial? I hope not -- I aim
    for long-term platform neutrality, but at the same time, I do want PyMOL
    users and potential PyMOL users to be fully informed as to their
    options. Plus, I am concerned about the distorting effect of dogmatic
    anti-Mac prejudice which pervades some corporations. The goal behind
    this message is to encourage everyone become aware of a unique new
    hardware offering so that you will take the time to evaluate it and can
    then make fully-informed decisions to buy whatever product best suits
    your scientific needs.

    Don't we owe it to ourselves and to the people who fund our
    research to apply our finite resources judiciously? In keeping with
    this, we should seek to reward those vendors who deliver the best values
    and to avoid those who do not. To do anything less is to waste precious
    resources or opportunities.

    So take a look at the dual 2 Ghz G5 even if it means overcoming
    historical anti-Mac bias within yourself or your organization. Sure,
    "Macintosh" may have become synonymous with "molasses" during the last
    couple of years, but that perception just doesn't apply to these
    forthcoming dual 2 Ghz G5 systems. If you do run into static within
    your IT department, remind them that these machines are long-sought
    replacements for those expensive old SGI workstations and
    "hard-to-support" Linux desktops, and that they pose no real threat to
    the Windows corporate enterprise. They are tools for doing science just
    like an X-ray diffractometer or an NMR, and who is IT to dictate what
    tools scientists should use?

    DISCLIAMER: Apple is NOT a financial sponsor of the PyMOL
    project, but they have taken steps to promote PyMOL on Macintosh by
    providing early access to hardware, exposure, and limited developer
    assistance.

    DISCLIAMER to the DISCLAIMER: So what? We're not afraid to
    criticize Apple. Apart from the iPod, the dual 2 Ghz G5 PowerMac is the
    ONLY system we recommend you purchase from the company at this time.
    Their laptops, cluster nodes, single-processor G5s, and mid-range G4
    systems are arguably still behind the competition, even after taking OS
    X's advantages into account. I'm no Mac-addict, but I do like to
    recognize and celebrate a good product when I see one. The dual 2 Ghz
    G5 running Mac OS X is a "killer" desktop computer for research --
    Apple's first in a long time, and hopefully more will come if we
    appropriately reward this one with our business.

    Cheers,
    Warren
     
  2. MacELCH

    MacELCH New Member

    Ob da wohl ein Mac-Herz schlägt ?

    Dies ist die Kopie einer Email aus einem Wissenschaftlichen Forum, es paßt aber auch hier ganz gut rein:

    PyMOLer's & Friends,

    For those of you considering a new PowerMac G5, a special
    "native" PyMOL v0.91 for the G5 with some nice demonstrations has been
    released: http://pymol.sf.net. This version contains an assortment of
    G5 optimizations, and the next release will have even more. Current G4
    users will benefit from these as well.

    Although Apple's top-of-the-line 2.0 Ghz G5 systems aren't quite
    shipping yet, they did let us run some PyMOL benchmarks on one, and the
    numbers were very impressive. In my own hands, a dual 2.0 Ghz G5
    effectively competes with a dual 3.0 Ghz Pentium 4 Xeon for generating
    PyMOL ray-traced movies in head-to-head comparisons. OpenGL speed is
    also top notch. So in my opinion, Mac/PC performance parity has finally
    been achieved -- congratulations Apple!

    By the way, Steve Jobs' controversial "twice as fast as a PC"
    claim is only valid for code exploiting the Altivec vectorization unit.
    PyMOL doesn't use that, but it can still render images equally as fast
    on the dual-processor Mac as on the dual-processor PC. "Once as fast as
    a PC is fast enough", I say, without any magic code.

    Thus, the really good news for cross-platform developers (like
    me) is that it is not necessary to write G5-specific code in order to
    get good performance. In fact, most of the "G5 optimizations" we put
    into PyMOL have equally helped the PC version (much to Apple's chagrin).
    Code optimization is simply easier on a Mac using Apple's great free
    tools (especially Shark
    http://developer.apple.com/technotes/tn/tn2086.html#useshark ), and the
    resulting code runs measurably faster on every platform.

    So if you are looking to buy a fast dual-processor desktop
    system, you may want to take a look at the G5 once the dual 2 Ghz
    machines ship in a few weeks. Though you'll probably pay more for this
    Mac than you will for a no-name x86 Linux setup, you will get concrete
    value in return:

    - freedom from Linux hardware configuration and OpenGL driver
    hassles
    - accountable vendor support
    - a user-friendly operating system
    - desktop applications such as Office, Photoshop, and
    Illustrator.
    - awesome developer tools (at no additional cost!)
    - 64 bits
    - stereo 3D graphics capability
    - and of course, the unique Mac OS X "look and feel".

    Plus, since Mac OS X is Unix, most computational chemistry,
    bioinformatics, and structural biology research software already runs
    well on it.

    Whatever your platform disposition, it is absolutely in your
    interest to learn about the many new options which exist before your
    next big purchase. With ridiculously fast Pentium 4, G5, Athlon,
    Itanium2, and Opteron hardware now entering the market, and with at
    least four affordable or free operating systems to choose from, there
    are plenty of ways to build a great system for relatively little money.
    You should not spend the cost of a new car just to buy a desktop
    computer (sadly, some companies I know STILL do this...).

    These days, PyMOL can compile and run virtually anywhere, but
    our focus will continue to be on x86 Windows, x86 Linux, and G4/G5 Macs,
    with IRIX and Solaris support continuing as long as demand persists.
    Please do speak up if/when native Itanium2 and Opteron binaries are
    needed...

    Does this sound like an Apple commercial? I hope not -- I aim
    for long-term platform neutrality, but at the same time, I do want PyMOL
    users and potential PyMOL users to be fully informed as to their
    options. Plus, I am concerned about the distorting effect of dogmatic
    anti-Mac prejudice which pervades some corporations. The goal behind
    this message is to encourage everyone become aware of a unique new
    hardware offering so that you will take the time to evaluate it and can
    then make fully-informed decisions to buy whatever product best suits
    your scientific needs.

    Don't we owe it to ourselves and to the people who fund our
    research to apply our finite resources judiciously? In keeping with
    this, we should seek to reward those vendors who deliver the best values
    and to avoid those who do not. To do anything less is to waste precious
    resources or opportunities.

    So take a look at the dual 2 Ghz G5 even if it means overcoming
    historical anti-Mac bias within yourself or your organization. Sure,
    "Macintosh" may have become synonymous with "molasses" during the last
    couple of years, but that perception just doesn't apply to these
    forthcoming dual 2 Ghz G5 systems. If you do run into static within
    your IT department, remind them that these machines are long-sought
    replacements for those expensive old SGI workstations and
    "hard-to-support" Linux desktops, and that they pose no real threat to
    the Windows corporate enterprise. They are tools for doing science just
    like an X-ray diffractometer or an NMR, and who is IT to dictate what
    tools scientists should use?

    DISCLIAMER: Apple is NOT a financial sponsor of the PyMOL
    project, but they have taken steps to promote PyMOL on Macintosh by
    providing early access to hardware, exposure, and limited developer
    assistance.

    DISCLIAMER to the DISCLAIMER: So what? We're not afraid to
    criticize Apple. Apart from the iPod, the dual 2 Ghz G5 PowerMac is the
    ONLY system we recommend you purchase from the company at this time.
    Their laptops, cluster nodes, single-processor G5s, and mid-range G4
    systems are arguably still behind the competition, even after taking OS
    X's advantages into account. I'm no Mac-addict, but I do like to
    recognize and celebrate a good product when I see one. The dual 2 Ghz
    G5 running Mac OS X is a "killer" desktop computer for research --
    Apple's first in a long time, and hopefully more will come if we
    appropriately reward this one with our business.

    Cheers,
    Warren
     
  3. MacELCH

    MacELCH New Member

    Demnächst kommen hier noch ein paar "Real World Test", die dank Ganimed an einem 1.8er G5 durchgeführt wurden mit Vergleichen zu einem Titanium@500 bzw. G4@867.

    Dieses Programm zeigt hauptsächlich die OpenGL Fähigkeiten des Rechners (aber nicht nur).
     
  4. MacELCH

    MacELCH New Member

    So und nun die erste Auswertung:

    Sieht ganz schön beeindruckend aus, was das für mich bedeutet ? Mit dem G5 kann man fließend Betonung liegt auf fließend mit OpenGL Programmen arbeiten, so daß es quasi in Echtzeit ist.

    Eben kam noch ein G4@733 mit nachgerüsteter ATI8500 64 MB in meinen Kasten rein:

    Hier die Werte:

    black screen - (vermutlich 100 fps)
    model screen - 50 fps
    rotation1 - 11 s
    hsc - 0.950 fps
    rotation2 - 6:30 min
     

Diese Seite empfehlen