Dear ImageJ people,
I'm getting <Out of memory> errors at 653 MB in ImageJ64, regardless of the amount of memory assigned in Options/Memory & Threads. The patient is a MacBook Pro with 4 GB of memory running OS X 10.5.2. ImageJ version 1.4.1k, Java version 1.4.2_16. It does not seem to matter how I get to 653 MB: the same thing happens with a complicated plugin or just opening the mri-stack sample and scaling it up. The memory monitor shows only 35% used, or so. The problem is specific to ImageJ64. The 32-bit version of ImageJ works fine. I suspected a hardware problem because the memory expansion from 2 GB to 4 GB included breaking the little clips that retain the cards. (The proper way to disengage the clips is to spread them horizontally, not pry them up...) However, an extended hardware test revealed no problems. Any ideas? Another ImageJ64 puzzle is that Quicktime works on one MacBook pro, but not a seemingly identical other one when running ImageJ64. It works on both machines with regular ImageJ. Bob Robert Dougherty, Ph.D. President, OptiNav, Inc. 10900 NE 8th St, Suite 900 Bellevue, WA 98004 Tel. (425)990-5912 FAX (425)467-1119 www.optinav.com [hidden email] |
Bob,
Unless I am very confused, MacOSX supports 64-bit only for java 1.6.0, and not all 1.6.0 packages --you have to download it from the developer.apple.com. AFAIK the 1.4.2_16 won't work in 64-bit mode. Albert -- Albert Cardona http://www.mcdb.ucla.edu/Research/Hartenstein/acardona |
Albert,
That's it. Thanks. I'll try it soon. Bob On Sep 9, 2008, at 3:06 PM, Albert Cardona wrote: > Bob, > > Unless I am very confused, MacOSX supports 64-bit only for java > 1.6.0, and not all 1.6.0 packages --you have to download it from > the developer.apple.com. AFAIK the 1.4.2_16 won't work in 64-bit mode. > > > Albert > > -- > Albert Cardona > http://www.mcdb.ucla.edu/Research/Hartenstein/acardona > |
In reply to this post by Robert Dougherty
On Sep 9, 2008, at 5:46 PM, Robert Dougherty wrote:
> Dear ImageJ people, > > I'm getting <Out of memory> errors at 653 MB in ImageJ64, > regardless of the amount of memory assigned in Options/Memory & > Threads. The patient is a MacBook Pro with 4 GB of memory running > OS X 10.5.2. ImageJ version 1.4.1k, Java version 1.4.2_16. It > does not seem to matter how I get to 653 MB: the same thing happens > with a complicated plugin or just opening the mri-stack sample and > scaling it up. The memory monitor shows only 35% used, or so. The > problem is specific to ImageJ64. The 32-bit version of ImageJ > works fine. I suspected a hardware problem because the memory > expansion from 2 GB to 4 GB included breaking the little clips that > retain the cards. (The proper way to disengage the clips is to > spread them horizontally, not pry them up...) However, an extended > hardware test revealed no problems. Any ideas? ImageJ64 requires Java 1.5 or later, Mac OS X 10.5 and a 64-bit Intel processor. Java 1.6, which is 64-bit only, is available via Software Update, although it is is not made the default version of Java. To make it the default, switch to "Java SE 6 (64-bit)" in /Applications/ Utilities/Java/Java Preferences. > Another ImageJ64 puzzle is that Quicktime works on one MacBook > pro, but not a seemingly identical other one when running > ImageJ64. It works on both machines with regular ImageJ. QuickTime for Java is not 64-bit compatible. The machine it is working on is probably running 32-bit Java and the other one is probably running 64-bit Java. The title of the Edit>Options>Memory & Threads dialog box changes to "Memory (64-bit)" when ImageJ is running on a 64-bit version of Java. -wayne |
I get a similar <Out of Memory> error at around 640MB regardless of the Memory setting, like the original poster, except I'm running 64-bit Java--1.6.0_12. In "About ImageJ..." the last line says 465MB of 3000MB (15%). The title of the Edit>Options>Memory is "Memory (64-bit)". I tried rebooting and updating to the latest ImageJ 1.42m to no avail. I'm running 64-bit Vista w/ 4GB of RAM, no other programs open. |
OK, I solved my problem by running ImageJ as an Administrator, by right clicking on ImageJ.exe and then checking the box in the "Compatibility" panel. Mind you, running as Administrator was suggested in the Install notes, but there it is implied that this is only necessary to use the "Memory & Threads..." option. Indeed, I WAS initially prevented from using this option, but then I gave Users in the imagej directory (in "Program Files (x86)") full permissions and then I was able to generate the ImageJ.cfg file, and the max memory seemed to be updated. However, as mentioned in the previous post, I would still run out of memory at around 640MB, followed by other oddities. So I would recommend that the Run as Administrator box be checked as a matter of course if ImageJ is installed in the default folder. |
In reply to this post by sisan
I don't know if this is the same thing, but Windows (XP, ME) limits an
application to a total (physical and virtual) memory allocation around 600 Mbytes. This has bitten me before with Matlab, so I assume it is the same with imageJ. David On Fri, Apr 24, 2009 at 8:14 PM, sisan <[hidden email]> wrote: > Wayne Rasband wrote: > > > > On Sep 9, 2008, at 5:46 PM, Robert Dougherty wrote: > > > >> Dear ImageJ people, > >> > >> I'm getting <Out of memory> errors at 653 MB in ImageJ64, > >> regardless of the amount of memory assigned in Options/Memory & > >> Threads. The patient is a MacBook Pro with 4 GB of memory running > >> OS X 10.5.2. ImageJ version 1.4.1k, Java version 1.4.2_16. It > >> does not seem to matter how I get to 653 MB: the same thing happens > >> with a complicated plugin or just opening the mri-stack sample and > >> scaling it up. The memory monitor shows only 35% used, or so. The > >> problem is specific to ImageJ64. The 32-bit version of ImageJ > >> works fine. I suspected a hardware problem because the memory > >> expansion from 2 GB to 4 GB included breaking the little clips that > >> retain the cards. (The proper way to disengage the clips is to > >> spread them horizontally, not pry them up...) However, an extended > >> hardware test revealed no problems. Any ideas? > > > > ImageJ64 requires Java 1.5 or later, Mac OS X 10.5 and a 64-bit Intel > > processor. Java 1.6, which is 64-bit only, is available via Software > > Update, although it is is not made the default version of Java. To > > make it the default, switch to "Java SE 6 (64-bit)" in /Applications/ > > Utilities/Java/Java Preferences. > > > >> Another ImageJ64 puzzle is that Quicktime works on one MacBook > >> pro, but not a seemingly identical other one when running > >> ImageJ64. It works on both machines with regular ImageJ. > > > > QuickTime for Java is not 64-bit compatible. The machine it is > > working on is probably running 32-bit Java and the other one is > > probably running 64-bit Java. The title of the Edit>Options>Memory & > > Threads dialog box changes to "Memory (64-bit)" when ImageJ is > > running on a 64-bit version of Java. > > > > -wayne > > > > > > I get a similar <Out of Memory> error at around 640MB regardless of the > Memory setting, like the original poster, except I'm running 64-bit > Java--1.6.0_12. In "About ImageJ..." the last line says 465MB of 3000MB > (15%). The title of the Edit>Options>Memory is "Memory (64-bit)". I tried > rebooting and updating to the latest ImageJ 1.42m to no avail. I'm running > 64-bit Vista w/ 4GB of RAM, no other programs open. > -- > View this message in context: > http://n2.nabble.com/unexplainable-memory-limit-in-ImageJ64-tp1078657p2701372.html > Sent from the ImageJ mailing list archive at Nabble.com. > |
Hi David,
>I don't know if this is the same thing, but Windows (XP, ME) limits an >application to a total (physical and virtual) memory allocation around 600 >Mbytes. This has bitten me before with Matlab, so I assume it is the same >with imageJ. Where did you get this info from? Microsoft claims different (even on Windows 2000) http://msdn.microsoft.com/en-us/library/aa366778.aspx "User-mode virtual address space for each 32-bit process 2 GB: Up to 3 GB with IMAGE_FILE_LARGE_ADDRESS_AWARE and 4GT" Never heard about this before! Joachim ____________________________________________ Leica Microsystems CMS GmbH | GmbH mit Sitz in Wetzlar | Amtsgericht Wetzlar HRB 2432 Geschäftsführer: Dr. Stefan Traeger | Dr. Wolf-Otto Reuter | Dr. David Roy Martyr | Colin Davis ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________ |
Hi,
On Tue, 28 Apr 2009, Joachim Wesner wrote: > >I don't know if this is the same thing, but Windows (XP, ME) limits an > >application to a total (physical and virtual) memory allocation around > >600 Mbytes. This has bitten me before with Matlab, so I assume it is > >the same with imageJ. > > Where did you get this info from? Microsoft claims different (even on > Windows 2000) > > http://msdn.microsoft.com/en-us/library/aa366778.aspx > > "User-mode virtual address space for each 32-bit process > > 2 GB: Up to 3 GB with IMAGE_FILE_LARGE_ADDRESS_AWARE and 4GT" Umm. Did the subject not state that the issue is about 64-bit? Ciao, Dscho |
In reply to this post by Joachim Wesner
Joachim
Got it by direct experiment. For instance by using "zeros" to allocate memory in MatLab. This has been the same for Windows ME, NT, W2K, and XP. I suppose it coud be a MatLab problem as I've never tried it with say a C++ program, but MathWorks says not. David On Tue, Apr 28, 2009 at 12:25 PM, Joachim Wesner < [hidden email]> wrote: > Hi David, > > >I don't know if this is the same thing, but Windows (XP, ME) limits an > >application to a total (physical and virtual) memory allocation around 600 > >Mbytes. This has bitten me before with Matlab, so I assume it is the same > >with imageJ. > > Where did you get this info from? Microsoft claims different (even on > Windows 2000) > > http://msdn.microsoft.com/en-us/library/aa366778.aspx > > "User-mode virtual address space for each 32-bit process > > 2 GB: Up to 3 GB with IMAGE_FILE_LARGE_ADDRESS_AWARE and 4GT" > > Never heard about this before! > > Joachim > > ____________________________________________ > > Leica Microsystems CMS GmbH | GmbH mit Sitz in Wetzlar | Amtsgericht > Wetzlar HRB 2432 > Geschäftsführer: Dr. Stefan Traeger | Dr. Wolf-Otto Reuter | Dr. David Roy > Martyr | Colin Davis > > > ______________________________________________________________________ > This email has been scanned by the MessageLabs Email Security System. > For more information please visit http://www.messagelabs.com/email > ______________________________________________________________________ > |
In reply to this post by dscho
Hi Dscho,
originally, maybe, yes, but as I understood the later poster wanted to make the point that there is a limit also in Win32 , as there is no ME 64 bit and also the very same site I cited talks about 64 Bit windows and also indicates NO hint about a strange 640 MB limit especially in Win64! Joachim Johannes Schindelin <Johannes.Schinde An [hidden email]> Joachim Wesner <joachim.wesner@LEICA-MICROSYSTEMS 29.04.2009 00:14 .COM> Kopie [hidden email] Thema Re: Antwort: Re: unexplainable memory limit in ImageJ64 Hi, On Tue, 28 Apr 2009, Joachim Wesner wrote: > >I don't know if this is the same thing, but Windows (XP, ME) limits an > >application to a total (physical and virtual) memory allocation around > >600 Mbytes. This has bitten me before with Matlab, so I assume it is > >the same with imageJ. > > Where did you get this info from? Microsoft claims different (even on > Windows 2000) > > http://msdn.microsoft.com/en-us/library/aa366778.aspx > > "User-mode virtual address space for each 32-bit process > > 2 GB: Up to 3 GB with IMAGE_FILE_LARGE_ADDRESS_AWARE and 4GT" Umm. Did the subject not state that the issue is about 64-bit? Ciao, Dscho ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________ |
In reply to this post by David Webster
Hi Dan,
sorry, but this must be a problem of your setup (and that Matlab cannot use more than 2GB even if the system is set for 3GB user space) !? What does "memory" say on your system? On a somewhat loaded 32-bit system (some applications open) I get with Matlab 2008b Maximum possible array: 1279 MB (1.341e+009 bytes) * Memory available for all arrays: 1600 MB (1.677e+009 bytes) ** Memory used by MATLAB: 220 MB (2.308e+008 bytes) Physical Memory (RAM): 3326 MB (3.487e+009 bytes) and as you suggested I just allocated three arrays of about 1200MB total and there are now 63 MB left! On 64 Bit that limit should not occur any more http://www.mathworks.de/support/tech-notes/1100/1106.html Mit freundlichen Grüßen / Best regards Joachim ____________________________________________ Leica Microsystems CMS GmbH | GmbH mit Sitz in Wetzlar | Amtsgericht Wetzlar HRB 2432 Geschäftsführer: Dr. Stefan Traeger | Dr. Wolf-Otto Reuter | Dr. David Roy Martyr | Colin Davis David Webster <dwwebster00@GMAI L.COM> An Gesendet von: [hidden email] ImageJ Interest Kopie Group <[hidden email]. Thema GOV> Re: unexplainable memory limit in ImageJ64 29.04.2009 01:33 Bitte antworten an ImageJ Interest Group <[hidden email]. GOV> Joachim Got it by direct experiment. For instance by using "zeros" to allocate memory in MatLab. This has been the same for Windows ME, NT, W2K, and XP. I suppose it coud be a MatLab problem as I've never tried it with say a C++ program, but MathWorks says not. David On Tue, Apr 28, 2009 at 12:25 PM, Joachim Wesner < [hidden email]> wrote: > Hi David, > > >I don't know if this is the same thing, but Windows (XP, ME) limits an > >application to a total (physical and virtual) memory allocation around 600 > >Mbytes. This has bitten me before with Matlab, so I assume it is the same > >with imageJ. > > Where did you get this info from? Microsoft claims different (even on > Windows 2000) > > http://msdn.microsoft.com/en-us/library/aa366778.aspx > > "User-mode virtual address space for each 32-bit process > > 2 GB: Up to 3 GB with IMAGE_FILE_LARGE_ADDRESS_AWARE and 4GT" > > Never heard about this before! > > Joachim > > ____________________________________________ > > Leica Microsystems CMS GmbH | GmbH mit Sitz in Wetzlar | Amtsgericht > Wetzlar HRB 2432 > Geschäftsführer: Dr. Stefan Traeger | Dr. Wolf-Otto Reuter | Dr. David > Martyr | Colin Davis > > > ______________________________________________________________________ > This email has been scanned by the MessageLabs Email Security System. > For more information please visit http://www.messagelabs.com/email > ______________________________________________________________________ > ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________ |
Free forum by Nabble | Edit this page |