Re: Processor for photo editting

Discussion in 'Digital Photography' started by Rita Berkowitz, Mar 26, 2008.

  1. nospam wrote:

    >> Just a hint, it aint happening since Photoshop isn't optimized for
    >> 64-bit. You can have 10GB of memory and it aint making any
    >> difference. The only thing significant here is you are in total
    >> bliss with thinking the placebo of having extra memory is actually
    >> helping when it is not. Anything past
    >> 1.5GB of RAM is a waste for Photoshop.

    >
    > totally false.


    How so?




    Rita
     
    Rita Berkowitz, Mar 26, 2008
    #1
    1. Advertising

  2. Rita Berkowitz

    -hh Guest

    On Mar 26, 6:49 am, "Rita Berkowitz" <> wrote:
    > nospam wrote:
    > >> Just a hint, it aint happening since Photoshop isn't optimized for
    > >> 64-bit. You can have 10GB of memory and it aint making any
    > >> difference.  The only thing significant here is you are in total
    > >> bliss with thinking the placebo of having extra memory is actually
    > >> helping when it is not.  Anything past
    > >> 1.5GB of RAM is a waste for Photoshop.

    >
    > > totally false.

    >
    > How so?



    The issue here is not 32-vs-64 addressing, because 1.5GB is well
    within 32 bit addressing limits.


    Since contemporary OS's run virtual RAM and swaps out to the Hard
    Drive (HD) whenever the user has more stuff running than physical RAM,
    the system's performance takes a hit from the I/O access speed
    differences between RAM and HD.

    This is why there's a performance gain from adding more physical RAM,
    and that this is more significant than Disk I/O - - simply put, RAM I/
    O is faster than any hard drive configuration in existance today.


    Granted, these performance gains will be an area of diminishing
    returns, since the RAM is displacing Disk I/O. Diminishing returns
    means that the first +1GB of RAM has more benefit than the next +1GB
    of RAM added, etc.

    Thus, when someone like "Rita" claims that anything more than 1.5GB of
    RAM is a "waste", they're merely speaking for their level of personal
    need. Thus, I'm not surprised to hear "Rita" claims.

    For everyone else, the metric to use is to examine one's Page-In/Page-
    Out ratio after a period of typically serious use of Photoshop, etc,
    to then decide if the frequency with which the OS hits the HD due to
    lack of Physical RAM is acceptable to them or not.

    The exercise is left up to Rita to post a screenshot of their Page-In/
    Out values so as to back up that pesky little "Anything more than
    1.5GB is a waste" claim. Its trivial to do this on a Mac. :)


    -hh
     
    -hh, Mar 26, 2008
    #2
    1. Advertising

  3. -hh wrote:
    > On Mar 26, 6:49 am, "Rita Berkowitz" <> wrote:
    >> nospam wrote:
    >>>> Just a hint, it aint happening since Photoshop isn't optimized for
    >>>> 64-bit. You can have 10GB of memory and it aint making any
    >>>> difference. The only thing significant here is you are in total
    >>>> bliss with thinking the placebo of having extra memory is actually
    >>>> helping when it is not. Anything past
    >>>> 1.5GB of RAM is a waste for Photoshop.
    >>> totally false.

    >> How so?

    >
    >
    > The issue here is not 32-vs-64 addressing, because 1.5GB is well
    > within 32 bit addressing limits.
    >
    >
    > Since contemporary OS's run virtual RAM and swaps out to the Hard
    > Drive (HD) whenever the user has more stuff running than physical RAM,
    > the system's performance takes a hit from the I/O access speed
    > differences between RAM and HD.
    >
    > This is why there's a performance gain from adding more physical RAM,
    > and that this is more significant than Disk I/O - - simply put, RAM I/
    > O is faster than any hard drive configuration in existance today.
    >
    >
    > Granted, these performance gains will be an area of diminishing
    > returns, since the RAM is displacing Disk I/O. Diminishing returns
    > means that the first +1GB of RAM has more benefit than the next +1GB
    > of RAM added, etc.
    >
    > Thus, when someone like "Rita" claims that anything more than 1.5GB of
    > RAM is a "waste", they're merely speaking for their level of personal
    > need. Thus, I'm not surprised to hear "Rita" claims.
    >
    > For everyone else, the metric to use is to examine one's Page-In/Page-
    > Out ratio after a period of typically serious use of Photoshop, etc,
    > to then decide if the frequency with which the OS hits the HD due to
    > lack of Physical RAM is acceptable to them or not.
    >
    > The exercise is left up to Rita to post a screenshot of their Page-In/
    > Out values so as to back up that pesky little "Anything more than
    > 1.5GB is a waste" claim. Its trivial to do this on a Mac. :)


    No doubt you'll get a flippant answer that completely dodges the issue.

    --
    john mcwilliams
     
    John McWilliams, Mar 26, 2008
    #3
  4. -hh wrote:

    > Since contemporary OS's run virtual RAM and swaps out to the Hard
    > Drive (HD) whenever the user has more stuff running than physical RAM,
    > the system's performance takes a hit from the I/O access speed
    > differences between RAM and HD.


    That's your misunderstanding of how CS3 manages these files and utilizes
    memory. In a batch process of 150 D3 RAW files you will never break the
    1.5GB memory barrier if you have acceptable disk I/O.

    > This is why there's a performance gain from adding more physical RAM,
    > and that this is more significant than Disk I/O - - simply put, RAM I/
    > O is faster than any hard drive configuration in existance today.


    I agree with you when one upgrades from 1GB to 2GB or RAM. RAM I/O speed is
    irrelevant because you have to get your filed from and back to the disk.

    > Granted, these performance gains will be an area of diminishing
    > returns, since the RAM is displacing Disk I/O. Diminishing returns
    > means that the first +1GB of RAM has more benefit than the next +1GB
    > of RAM added, etc.


    Well, 1.5GB is the sweet spot for CS3.

    > Thus, when someone like "Rita" claims that anything more than 1.5GB of
    > RAM is a "waste", they're merely speaking for their level of personal
    > need. Thus, I'm not surprised to hear "Rita" claims.


    Could be since batching 150 D3 RAW files really is just a blip on the old
    SCSI I/O.

    > For everyone else, the metric to use is to examine one's Page-In/Page-
    > Out ratio after a period of typically serious use of Photoshop, etc,
    > to then decide if the frequency with which the OS hits the HD due to
    > lack of Physical RAM is acceptable to them or not.


    I haven't broke the 1.5GB barrier yet and don't know anyone else that has.
    If a person is breaking the 1.5GB barrier they are doing something seriously
    wrong.





    Rita
     
    Rita Berkowitz, Mar 26, 2008
    #4
  5. Rita Berkowitz

    nospam Guest

    In article <>, Rita Berkowitz
    <> wrote:

    > I haven't broke the 1.5GB barrier yet and don't know anyone else that has.
    > If a person is breaking the 1.5GB barrier they are doing something seriously
    > wrong.


    there's no such barrier and you are trolling.
     
    nospam, Mar 27, 2008
    #5
  6. Rita Berkowitz

    -hh Guest

    "Rita Berkowitz" <> wrote:
    > -hh wrote:
    > > Since contemporary OS's run virtual RAM and swaps out to the Hard
    > > Drive (HD) whenever the user has more stuff running than physical RAM,
    > > the system's performance takes a hit from the I/O access speed
    > > differences between RAM and HD.

    >
    > That's your misunderstanding of how CS3 manages these files and utilizes
    > memory.  In a batch process of 150 D3 RAW files you will never break the
    > 1.5GB memory barrier if you have acceptable disk I/O.


    But unfortunately, when your total system only has 1.5GB, the overhead
    from other activities will never let Photoshop reach this 1.5GB limit.

    YMMV on what else you choose to leave running, but between the OS,
    Mail, Browser & background tasks, its not unusual to have 800MB-1GB of
    real memory allocated. With the Rita 1.5GB system, this only leaves
    ~700MB of Real memory available for Photoshop...roughly half of the
    above claimed limit.



    > > This is why there's a performance gain from adding more physical RAM,
    > > and that this is more significant than Disk I/O - - simply put, RAM I/
    > > O is faster than any hard drive configuration in existance today.

    >
    > I agree with you when one upgrades from 1GB to 2GB or RAM.  


    Well, you had better go tell that to the "Other Rita", who on 3/25/08
    claimed that even 2GB of RAM is a waste:

    "You can have 10GB of memory and it aint making any
    difference....Anything past 1.5GB of RAM is a waste for Photoshop."


    > RAM I/O speed is irrelevant because you have to get your
    > filed from and back to the disk.


    Disk I/O is relevant only when opening & closing the file. For all of
    the other image manipulating actions, its CPU and RAM that determine
    how long it will take. It depends on what you're doing as to which
    ultimately takes "longer", but the fact remains that substituting Disk
    Swap for Physical RAM **always** increases the timeline.


    > > Granted, these performance gains will be an area of diminishing
    > > returns, since the RAM is displacing Disk I/O.  Diminishing returns
    > > means that the first +1GB of RAM has more benefit than the next +1GB
    > > of RAM added, etc.

    >
    > Well, 1.5GB is the sweet spot for CS3.


    2^30 = 1GB and 2^31 = 2GB; 1.5GB is approximately 2^30.5.

    Now how can a non-integer value of bits to have rationality as a
    memory addressing limit? Or was it simply an arbitrary value written
    into a Photoshop subroutine that's intended to avoid it from hogging
    some perceived value of "too much" from the OS's RAM?

    Unfortunately, without Rita providing a URL to a Geocities webpage
    that is a screenshot of Rita's "Activity Monitor" screen, to see RAM &
    VM allocations, Page In/Out values, etc...this is an empty claim.


    > > Thus, when someone like "Rita" claims that anything more than 1.5GB of
    > > RAM is a "waste", they're merely speaking for their level of personal
    > > need.  Thus, I'm not surprised to hear "Rita" claims.

    >
    > Could be since batching 150 D3 RAW files really is just a blip on the old
    > SCSI I/O.


    If you're running a batch on 150 files that's doing just one or two
    simple actions and then resaving the file, then disk I/O might be your
    predominant time factor. However, it is trivial to determine how much
    difference physical RAM contributes to your process: time how long it
    takes to do a batch of 100, then open the PC's lid, pull some RAM and
    time how long it takes to run another batch of 100 (ideally the same
    100). If your system isn't already choked, it will be obvious when
    you start to get into serious disk-swapping...and if you don't see any
    difference, your system needs more RAM. Report back your results,
    with nice PowerPoint Charts.


    > > For everyone else, the metric to use is to examine one's Page-In/Page-
    > > Out ratio after a period of typically serious use of Photoshop, etc,
    > > to then decide if the frequency with which the OS hits the HD due to
    > > lack of Physical RAM is acceptable to them or not.

    >
    > I haven't broke the 1.5GB barrier yet and don't know anyone else that has.


    It might be rare, but not impossible:

    <http://www.huntzinger.com/photo/2008/activity.jpg>


    > If a person is breaking the 1.5GB barrier they are doing something seriously
    > wrong.


    Or merely something different than 'Rita'.

    FYI, the above documented use of 1.6 real + 2 VM (3.7GB total) was
    from just two Photoshop files being open from the same project, namely
    the full scan of a Kodachrome @ 200 MP (17,433 x 11,551) and its
    first downsample to 8674 x 5776 (50 MP).


    -hh
     
    -hh, Mar 27, 2008
    #6
  7. -hh wrote:

    >> I agree with you when one upgrades from 1GB to 2GB or RAM.

    >
    > Well, you had better go tell that to the "Other Rita", who on 3/25/08
    > claimed that even 2GB of RAM is a waste:
    >
    > "You can have 10GB of memory and it aint making any
    > difference....Anything past 1.5GB of RAM is a waste for Photoshop."


    Funny! With 2GB one would only be wasting 512MB compared to 8.5GB with 10GB
    of RAM.

    >> RAM I/O speed is irrelevant because you have to get your
    >> filed from and back to the disk.

    >
    > Disk I/O is relevant only when opening & closing the file. For all of
    > the other image manipulating actions, its CPU and RAM that determine
    > how long it will take. It depends on what you're doing as to which
    > ultimately takes "longer", but the fact remains that substituting Disk
    > Swap for Physical RAM **always** increases the timeline.


    Again, you are assuming incorrectly that there is this magical need to keep
    everything up in RAM when it doesn't need to be there. Fix your disk I/O
    and memory won't be an issue.

    > FYI, the above documented use of 1.6 real + 2 VM (3.7GB total) was
    > from just two Photoshop files being open from the same project, namely
    > the full scan of a Kodachrome @ 200 MP (17,433 x 11,551) and its
    > first downsample to 8674 x 5776 (50 MP).


    It's just a simple case of poor disk I/O you're struggling with. Had you a
    decent disk array all of this overhead would have been offloaded back to the
    disk while your memory is waiting for more input.




    Rita
     
    Rita Berkowitz, Mar 27, 2008
    #7
  8. Rita Berkowitz

    -hh Guest

    John McWilliams <> wrote:
    > -hh wrote:
    > >
    > > The exercise is left up to Rita to post a screenshot of their Page-In/
    > > Out values so as to back up that pesky little "Anything more than
    > > 1.5GB is a waste" claim.  Its trivial to do this on a Mac. :)

    >
    > No doubt you'll get a flippant answer that completely dodges the issue.


    Yup, I did. Apparently, "Rita" has never realized that Photoshop's
    use of a scratch disk essentially means that its Virtual RAM will
    exceed its Physical. Hence, why "Rita" won't provide a screenshot of
    the Activity Monitor.

    In any event, my screenshot showed that Rita's flippant claim of a
    Photoshop hard limit ('barrier') at 1.5GB to be utterly false.


    -hh
     
    -hh, Mar 27, 2008
    #8
  9. Rita Berkowitz

    Ron Hunter Guest

    Rita Berkowitz wrote:
    > -hh wrote:
    >
    >>> I agree with you when one upgrades from 1GB to 2GB or RAM.

    >>
    >> Well, you had better go tell that to the "Other Rita", who on 3/25/08
    >> claimed that even 2GB of RAM is a waste:
    >>
    >> "You can have 10GB of memory and it aint making any
    >> difference....Anything past 1.5GB of RAM is a waste for Photoshop."

    >
    > Funny! With 2GB one would only be wasting 512MB compared to 8.5GB with
    > 10GB
    > of RAM.
    >
    >>> RAM I/O speed is irrelevant because you have to get your
    >>> filed from and back to the disk.

    >>
    >> Disk I/O is relevant only when opening & closing the file. For all of
    >> the other image manipulating actions, its CPU and RAM that determine
    >> how long it will take. It depends on what you're doing as to which
    >> ultimately takes "longer", but the fact remains that substituting Disk
    >> Swap for Physical RAM **always** increases the timeline.

    >
    > Again, you are assuming incorrectly that there is this magical need to keep
    > everything up in RAM when it doesn't need to be there. Fix your disk I/O
    > and memory won't be an issue.
    >
    >> FYI, the above documented use of 1.6 real + 2 VM (3.7GB total) was
    >> from just two Photoshop files being open from the same project, namely
    >> the full scan of a Kodachrome @ 200 MP (17,433 x 11,551) and its
    >> first downsample to 8674 x 5776 (50 MP).

    >
    > It's just a simple case of poor disk I/O you're struggling with. Had you a
    > decent disk array all of this overhead would have been offloaded back to
    > the
    > disk while your memory is waiting for more input.
    >
    >
    >
    >
    > Rita
    >


    Rita,
    Disk I/O is VASTLY slower than ram access. Worse, many programs seem
    to access 'scratchpad' files in a terribly inefficient, 'byte-wise'
    manner making things terribly slow. The best solution is to load the
    computer with all the ram it can handle, and get one with a quad-core
    (or two of them), and a fast HD. Frankly, Windows does a terrible job
    of swapping ram to and from disk.
     
    Ron Hunter, Mar 27, 2008
    #9
  10. Ron Hunter wrote:

    >
    > Rita,
    > Disk I/O is VASTLY slower than ram access. Worse, many programs seem
    > to access 'scratchpad' files in a terribly inefficient, 'byte-wise'
    > manner making things terribly slow. The best solution is to load the
    > computer with all the ram it can handle, and get one with a quad-core
    > (or two of them), and a fast HD. Frankly, Windows does a terrible job
    > of swapping ram to and from disk.


    Hey, Ron, it's been awhile. You'll be delighted to know the new Mac
    desktop is an eight core -Xeon-, can run OS X as well as Vista, and many
    flavors of other PC software if one were nostalgic.

    Arguing with "Rita" is like wrestling with a pig in mud. Even when
    you've got it pinned, it's just loving the mud.

    --
    john mcwilliams
     
    John McWilliams, Mar 27, 2008
    #10
  11. -hh wrote:
    > John McWilliams <> wrote:
    >> -hh wrote:
    >>> The exercise is left up to Rita to post a screenshot of their Page-In/
    >>> Out values so as to back up that pesky little "Anything more than
    >>> 1.5GB is a waste" claim. Its trivial to do this on a Mac. :)

    >> No doubt you'll get a flippant answer that completely dodges the issue.

    >
    > Yup, I did. Apparently, "Rita" has never realized that Photoshop's
    > use of a scratch disk essentially means that its Virtual RAM will
    > exceed its Physical. Hence, why "Rita" won't provide a screenshot of
    > the Activity Monitor.
    >
    > In any event, my screenshot showed that Rita's flippant claim of a
    > Photoshop hard limit ('barrier') at 1.5GB to be utterly false.


    Well, as the French say << Mais, oui!>>. All hat and no cattle was a
    phrase used by our ex-most-prolific poster, and "fits" the bill, too.

    --
    john mcwilliams
     
    John McWilliams, Mar 27, 2008
    #11
  12. Rita Berkowitz

    Ron Hunter Guest

    John McWilliams wrote:
    > Ron Hunter wrote:
    >
    >>
    >> Rita,
    >> Disk I/O is VASTLY slower than ram access. Worse, many programs
    >> seem to access 'scratchpad' files in a terribly inefficient,
    >> 'byte-wise' manner making things terribly slow. The best solution is
    >> to load the computer with all the ram it can handle, and get one with
    >> a quad-core (or two of them), and a fast HD. Frankly, Windows does a
    >> terrible job of swapping ram to and from disk.

    >
    > Hey, Ron, it's been awhile. You'll be delighted to know the new Mac
    > desktop is an eight core -Xeon-, can run OS X as well as Vista, and many
    > flavors of other PC software if one were nostalgic.
    >
    > Arguing with "Rita" is like wrestling with a pig in mud. Even when
    > you've got it pinned, it's just loving the mud.
    >


    With that many cores, it could probably run both at the same time.
    Never mind, I am schizo enough without thinking about that. Grin.
     
    Ron Hunter, Mar 27, 2008
    #12
  13. -hh wrote:

    > Yup, I did. Apparently, "Rita" has never realized that Photoshop's
    > use of a scratch disk essentially means that its Virtual RAM will
    > exceed its Physical. Hence, why "Rita" won't provide a screenshot of
    > the Activity Monitor.


    That's just it; I do know what's going on in my system. I have no problems
    with disk I/O or SMP.

    > In any event, my screenshot showed that Rita's flippant claim of a
    > Photoshop hard limit ('barrier') at 1.5GB to be utterly false.


    It only proves you have an improperly configured computer suffering from
    poor disk I/O and improper memory management.




    Rita
     
    Rita Berkowitz, Mar 27, 2008
    #13
  14. Ron Hunter wrote:

    > Disk I/O is VASTLY slower than ram access. Worse, many programs
    > seem to access 'scratchpad' files in a terribly inefficient,
    > 'byte-wise' manner making things terribly slow. The best solution is
    > to load the computer with all the ram it can handle, and get one with
    > a quad-core (or two of them), and a fast HD. Frankly, Windows does a
    > terrible job of swapping ram to and from disk.


    Ron, that's what I've been saying. Our boy -hh has some serious I/O
    problems. Had he been using a decent SCSI RAID array he would be offloading
    out of the old scratch disc so fast he'd barely break a GB of used memory.
    And yes, I also suggested SMP. It's only the people that are using
    substandard hardware and don't have a clue are the ones blindly and clueless
    recommending throwing more memory at the problem.





    Rita
     
    Rita Berkowitz, Mar 27, 2008
    #14
  15. Rita Berkowitz

    -hh Guest

    On Mar 27, 7:45 pm, "Rita Berkowitz" <> wrote:
    > Ron Hunter wrote:
    > >   Disk I/O is VASTLY slower than ram access.  Worse, many programs
    > > seem to access 'scratchpad' files in a terribly inefficient,
    > > 'byte-wise' manner making things terribly slow.  The best solution is
    > > to load the computer with all the ram it can handle, and get one with
    > > a quad-core (or two of them), and a fast HD.  Frankly, Windows does a
    > > terrible job of swapping ram to and from disk.

    >
    > Ron, that's what I've been saying.  Our boy -hh has some serious I/O
    > problems.  Had he been using a decent SCSI RAID array he would be offloading
    > out of the old scratch disc so fast he'd barely break a GB of used memory.


    Always so relevant when the system is idling.

    Next time, try looking at the %CPU being used by Photoshop (its at 2%)
    before jumping to conclusions about I/O:

    <http://www.huntzinger.com/photo/2008/activity.jpg>



    -hh
     
    -hh, Mar 28, 2008
    #15
  16. Rita Berkowitz

    Ron Hunter Guest

    Rita Berkowitz wrote:
    > Ron Hunter wrote:
    >
    >> Disk I/O is VASTLY slower than ram access. Worse, many programs
    >> seem to access 'scratchpad' files in a terribly inefficient,
    >> 'byte-wise' manner making things terribly slow. The best solution is
    >> to load the computer with all the ram it can handle, and get one with
    >> a quad-core (or two of them), and a fast HD. Frankly, Windows does a
    >> terrible job of swapping ram to and from disk.

    >
    > Ron, that's what I've been saying. Our boy -hh has some serious I/O
    > problems. Had he been using a decent SCSI RAID array he would be
    > offloading
    > out of the old scratch disc so fast he'd barely break a GB of used memory.
    > And yes, I also suggested SMP. It's only the people that are using
    > substandard hardware and don't have a clue are the ones blindly and
    > clueless
    > recommending throwing more memory at the problem.
    >
    >
    >
    >
    >
    > Rita
    >

    Adding RAM is not really very expensive these days, at least not
    compared to SCSI RAID arrays.
     
    Ron Hunter, Mar 28, 2008
    #16
  17. Ron Hunter wrote:

    >> Ron, that's what I've been saying. Our boy -hh has some serious I/O
    >> problems. Had he been using a decent SCSI RAID array he would be
    >> offloading
    >> out of the old scratch disc so fast he'd barely break a GB of used
    >> memory. And yes, I also suggested SMP. It's only the people that
    >> are using substandard hardware and don't have a clue are the ones
    >> blindly and clueless
    >> recommending throwing more memory at the problem.

    >
    > Adding RAM is not really very expensive these days, at least not
    > compared to SCSI RAID arrays.


    BINGO! We are going in circles chasing Adobe's propaganda tail. Thank you
    for admitting that it is cheaper and easier adding a pacifier than it is to
    do it right and suck up the added expense of correcting disk I/O and SMP
    issues. You're just saying the same things I've been saying, just in your
    own sweet little way.





    Rita
     
    Rita Berkowitz, Mar 28, 2008
    #17
  18. ["Followup-To:" header set to rec.photo.digital.slr-systems.]
    Ron Hunter <> wrote:

    > Adding RAM is not really very expensive these days, at least not
    > compared to SCSI RAID arrays.


    Rita's disks are faster than zir's RAM. That's not because
    zir disks are unusually fast, though. Zie's just a very, very
    slow thinker.

    -Wolfgang
     
    Wolfgang Weisselberg, Mar 28, 2008
    #18
  19. Rita Berkowitz

    Steve Guest

    On Fri, 28 Mar 2008 07:35:35 -0400, "Rita Berkowitz"
    <> wrote:

    >Ron Hunter wrote:
    >
    >>> Ron, that's what I've been saying. Our boy -hh has some serious I/O
    >>> problems. Had he been using a decent SCSI RAID array he would be
    >>> offloading
    >>> out of the old scratch disc so fast he'd barely break a GB of used
    >>> memory. And yes, I also suggested SMP. It's only the people that
    >>> are using substandard hardware and don't have a clue are the ones
    >>> blindly and clueless
    >>> recommending throwing more memory at the problem.

    >>
    >> Adding RAM is not really very expensive these days, at least not
    >> compared to SCSI RAID arrays.

    >
    >BINGO! We are going in circles chasing Adobe's propaganda tail. Thank you
    >for admitting that it is cheaper and easier adding a pacifier than it is to
    >do it right and suck up the added expense of correcting disk I/O and SMP
    >issues. You're just saying the same things I've been saying, just in your
    >own sweet little way.


    But you definitely get the best bang for the buck if you add ram to a
    system that doesn't have enough or the right type than adding a fast
    SCSI raid array to a system that just has a regular disk drive.

    It also depends a lot on what the program is doing. For instance, if
    you're using photoshop just as a photo viewer (open file, look at
    file, open the next) then boosting your disk access speed will make a
    noticable difference. But if you're using it for complex filters on
    high MP count images, then disk access speed won't make much
    difference if you have enough ram so it doesn't have to swap.

    Just to add some real life numbers to the mix, I did little
    experiments when I upgraded my system. Unfortunately I don't use
    photoshop. But I did take some measurements with Nikon
    PictureProject.

    First upgrade was the disk drive. My laptop came standard with an
    80GB 5400rpm drive. I upgraded it to a 100GB 7200rpm drive. Both are
    the same SATA interface but the 7200rpm has faster access, lower
    latency, etc. While overall the system seemed "snappier", the time to
    convert 10 NEF images from my D200 to jpegs while also doing
    brightness, contrast, sharpness, rotation, crop, etc., didn't change
    enough to be measurable. I already did an upgrade to 1.5GB of main
    memory so there was enough that it didn't have to swap. Maybe if it
    did have to swap, the faster hard drive might have made a difference.
    But it also would have been over much slower.

    Second upgrade was the main memory. As stated above, I had 1.5GB of
    memory. However, it was arranged poorly. I just added a 1GB memory
    stick to the 512MB stick that the laptop came with. The 512MB was
    PC4300 DDR533 memory and the 1GB stick was PC5400 DDR667 memory. But
    because the slower speed memory was in there, they 1GB was downgraded
    to 533MHz. Also, because they were not matched, the memory operated
    in single channel mode. The upgrade was to pull out the 512MB stick
    and replace it with a matched 1G PC5400 DDR667 stick. So now I have
    2GB of dual channel 667MHz memory instead of 1.5GB of single channel
    533MHz memory. With that upgrade, the time to convert the same 10 NEF
    images from my D200 to jpegs with brightness, contrast, sharpening,
    rotation, cropping, etc., dropped from 3:44 to 3:20.

    So here's a case where faster disk I/O = no difference but faster RAM
    = 12% increase in speed. But if there wasn't enough RAM in the first
    place, then faster disk I/O would have made a difference but overall
    would have been much slower than having sufficient RAM.

    So the real answer as to what upgrades make the biggest performance
    difference is "it depends." It depends on what the needs of your
    program are and also in what way your system is currently defficient.

    For photo editing in particular, if you have enough ram that your
    program isn't using swap space, then the only thing you might notice
    by adding a super fast disk raid array is that programs and files will
    load a little faster. Actually doing the editing (applying filters,
    corrections, etc.) isn't going to go any faster. But if you don't
    have enough ram and your photo editor is using swap space, the best
    upgrade you can make is add more ram. Adding a high I/O bandwidth
    disk array to such a system will make a difference. But a much higher
    cost and lower performance than adding ram. And add it in a smart way
    so that your system can take best advantage of it. I.e., highest
    speed your system will allow, and in matched pairs if your system
    supports dual channel.

    Steve
     
    Steve, Mar 28, 2008
    #19
  20. Rita Berkowitz

    -hh Guest

    Steve <> wrote:
    >
    > But you definitely get the best bang for the buck if you add ram to a
    > system that doesn't have enough or the right type than adding a fast
    > SCSI raid array to a system that just has a regular disk drive.


    It also depends on the flavor of SCSI, as SATA-II is comparable to SAS
    and Ultra-320. As such, "SCSI" is no longer a magic word per se, for
    a RAID 0 on a SATA bus buries a single spindle SCSI.


    > It also depends a lot on what the program is doing.  For instance,
    > if you're using photoshop just as a photo viewer (open file, look at
    > file, open the next) then boosting your disk access speed will make
    > a noticable difference.


    Precisely a workflow that 'Rita' alluded to. It would be nice to see
    some hardware specifics and performance numbers ... with tests
    repeated as RAM is pulled off ... but just like the System

    > But if you're using it for complex filters on high MP
    > count images, then disk access speed won't make much
    > difference if you have enough ram so it doesn't have to swap.


    Precisely a workflow that I referred to. Granted, one does take a HD
    I/O hit in that one has to read in the data file initially (and when
    done, write it out), but for all the time in between, CPU and RAM can
    quickly become far more significant because one is sitting at the
    keyboard waiting for a filter to complete, etc.


    > For photo editing in particular, if you have enough ram that your
    > program isn't using swap space, then the only thing you might notice
    > by adding a super fast disk raid array is that programs and files will
    > load a little faster.  Actually doing the editing (applying filters,
    > corrections, etc.) isn't going to go any faster.  But if you don't
    > have enough ram and your photo editor is using swap space, the best
    > upgrade you can make is add more ram.  Adding a high I/O bandwidth
    > disk array to such a system will make a difference.  But a much higher
    > cost and lower performance than adding ram.  And add it in a smart way
    > so that your system can take best advantage of it.  I.e., highest
    > speed your system will allow, and in matched pairs if your system
    > supports dual channel.


    In other words, "Objective Best Value"


    -hh
     
    -hh, Mar 29, 2008
    #20
    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. Lou

    editting forwarded emails

    Lou, Aug 2, 2005, in forum: Firefox
    Replies:
    3
    Views:
    481
    Moz Champion
    Aug 2, 2005
  2. Rita Berkowitz

    Re: Processor for photo editting

    Rita Berkowitz, Mar 24, 2008, in forum: Digital Photography
    Replies:
    11
    Views:
    470
  3. Rita Berkowitz

    Re: Processor for photo editting

    Rita Berkowitz, Mar 26, 2008, in forum: Digital Photography
    Replies:
    4
    Views:
    339
    Wolfgang Weisselberg
    Mar 27, 2008
  4. Rita Berkowitz

    Re: Processor for photo editting

    Rita Berkowitz, Mar 27, 2008, in forum: Digital Photography
    Replies:
    6
    Views:
    301
    Paul Allen
    Mar 28, 2008
  5. Giuen
    Replies:
    0
    Views:
    1,427
    Giuen
    Sep 12, 2008
Loading...

Share This Page