From: Warren D. <wa...@de...> - 2006-03-23 19:49:56
|
Pete, We have been suffering this same exact symptom with our nVidia-base 64-bit RHEL workstation for over a year. I have spent enough time troubleshooting the issue to have convinced myself that the problem isn't PyMOL -- it is an inability of certain combinations of hardware, Linux kernels, and proprietary OpenGL drivers to handle the vast quantity of immediate mode rendering geometry that PyMOL generates. It seems to be either a sporadic buffer overflow or a race condition within the kernel layer. =20 The proprietary nature of the OpenGL drivers prevents anyone other than the hardware vendor from being able to actually understand and solve the problem. Plus, the problem is sporadic -- most of the time everything works fine, but sometimes PyMOL hangs, sometimes all of X11 freezes, and sometimes the whole system comes crashing down. The intractable nature of these Linux stability problems is one of the primary reasons why I recommend Mac OS X over Linux for unix-based PyMOL visualization. =20 Note that the settings "nvidia_bugs" and "use_display_lists" may reduce the frequency of the problem on your system. Other than that, my suggestion would be to try different Linux kernels and OpenGL driver versions. =20 The combinatorial matrix of possibilities is huge, but maybe we should create a table on the Wiki for people to post working and non-working combinations of Linux kernels, cpu and graphics hardware, and opengl drivers? That's how we used to do things in the old days. I suspect only 5-10% of configurations are truly unworkable, and once enough data is collected from enough people, meaningful patterns would likely emerge. =20 Cheers Warren -- Warren L. DeLano, Ph.D. =20 Principal Scientist . DeLano Scientific LLC =20 . 400 Oyster Point Blvd., Suite 213 =20 . South San Francisco, CA 94080 USA =20 . Biz:(650)-872-0942 Tech:(650)-872-0834 =20 . Fax:(650)-872-0273 Cell:(650)-346-1154 . mailto:wa...@de... =20 =20 > -----Original Message----- > From: pym...@li...=20 > [mailto:pym...@li...] On Behalf Of=20 > Peter Adrian Meyer > Sent: Thursday, March 23, 2006 9:32 AM > To: pym...@li... > Subject: [PyMOL] sporadic pymol/X-server freezes? >=20 > Hi everyone, >=20 > Has anyone else run into cases where pymol will freeze the=20 > X-server (pymol doesn't repond to mouse or keyboard input;=20 > keyboard non-responsive to caps-lock on/off)? The system is=20 > still stable, I'm able to login remotely and see that pymol=20 > is running at 100% CPU. Killing pymol results in xorg=20 > running at 100% CPU (killing the xorg process as root doesn't=20 > do anything). >=20 > I've only seen this while looking at several maps/masks (but=20 > that's most of what I use pymol for, so I'm not sure if=20 > that's the cause). I've seen this error on both suse 10.0=20 > and kubuntu (3.10? whatever the lastest apt-get dist upgrade=20 > got me; two different machines) using source builds of=20 > pymol-0.98, and 0.99rc6 with ext-0.98,ext-0.99rc6, and ext-0.95. >=20 > Any suggestions on how to fix this (or get debugging=20 > information out of pymol to narrow down the problem)? >=20 > Thanks, > Pete >=20 >=20 > Pete Meyer > Fu Lab > BMCB grad student > Cornell University >=20 >=20 >=20 > ------------------------------------------------------- > This SF.Net email is sponsored by xPML, a groundbreaking=20 > scripting language that extends applications into web and=20 > mobile media. Attend the live webcast and join the prime=20 > developer group breaking into this new coding territory! > http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D110944&bid=3D241720& > dat=3D121642 > _______________________________________________ > PyMOL-users mailing list > PyM...@li... > https://lists.sourceforge.net/lists/listinfo/pymol-users >=20 >=20 >=20 >=20 |
From: Soisson, S. M. <ste...@me...> - 2006-03-24 12:39:36
|
My new laptop has an ATI FireGLV3200 card, and I am using the fglrx driver. I was having problems with seg faults, but using "set use_display_lists" fixed the problem. I did try using the "set nvidia" option that Warren suggested and it crashed my X-server. Had to do a hard reboot. So far so good though with the display_lists option. There clearly are some issues with the ATI GL implementation and pymol, at some level. Cheers, Steve -----Original Message----- From: pym...@li... [mailto:pym...@li...] On Behalf Of Tim Fenn Sent: Friday, March 24, 2006 3:30 AM To: Peter Adrian Meyer Cc: Warren DeLano; pym...@li... Subject: Re: [PyMOL] sporadic pymol/X-server freezes? On Thu, Mar 23, 2006 at 08:56:27PM -0500, Peter Adrian Meyer wrote: > > I guess it shouldn't supprise me that there are issues with proprietary > OpenGL on linux (FYI the two machines I've seen this on are using ATI and > intel graphics cards). I'll try out the options you mentioned. > So the ATI card is using the fglrx driver, I'm assuming (right?), and is the intel card using the Intel graphics drivers from intel.com? I've been told i810 that comes with XOrg is actually better (besides having full 3d support)... but what intel chipset are you using? Have you tried the NON-proprietary drivers? > Your idea about known-working (and non-working) combinations of hardware > seems like a very good thing to do. > If its a proprietary driver, my money is on a buggy driver that doesn't cooperate with the kernel. > > > > > >> Has anyone else run into cases where pymol will freeze the > >> X-server (pymol doesn't repond to mouse or keyboard input; > >> keyboard non-responsive to caps-lock on/off)? The system is > >> still stable, I'm able to login remotely and see that pymol > >> is running at 100% CPU. Killing pymol results in xorg > >> running at 100% CPU (killing the xorg process as root doesn't > >> do anything). So xorg quits, but the monitor stays frozen? Or the xorg process doesn't die? If its the latter, can you just send it a KILL (try an INT signal first, of course)? Anything in dmesg or the XOrg logs? Can you grab the XOrg pid with gdb and get a stack trace? Can you switch over to a console (ctrl-alt-f1, typically)? -Tim -- --------------------------------------------------------- Tim Fenn fe...@st... Stanford University, School of Medicine James H. Clark Center 318 Campus Drive, Room E300 Stanford, CA 94305-5432 Phone: (650) 736-1714 FAX: (650) 736-1961 --------------------------------------------------------- ------------------------------------------------------- This SF.Net email is sponsored by xPML, a groundbreaking scripting language that extends applications into web and mobile media. Attend the live webcast and join the prime developer group breaking into this new coding territory! http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642 _______________________________________________ PyMOL-users mailing list PyM...@li... https://lists.sourceforge.net/lists/listinfo/pymol-users ------------------------------------------------------------------------------ Notice: This e-mail message, together with any attachments, contains information of Merck & Co., Inc. (One Merck Drive, Whitehouse Station, New Jersey, USA 08889), and/or its affiliates (which may be known outside the United States as Merck Frosst, Merck Sharp & Dohme or MSD and in Japan, as Banyu) that may be confidential, proprietary copyrighted and/or legally privileged. It is intended solely for the use of the individual or entity named on this message. If you are not the intended recipient, and have received this message in error, please notify us immediately by reply e-mail and then delete it from your system. ------------------------------------------------------------------------------ |
From: Sabuj P. <sab...@va...> - 2006-03-23 20:14:49
|
Hi, >>Has anyone else run into cases where pymol will freeze the >>X-server (pymol doesn't repond to mouse or keyboard input; >>keyboard non-responsive to caps-lock on/off)? no. >>I've only seen this while looking at several maps/masks (but >>that's most of what I use pymol for, so I'm not sure if >>that's the cause). same here, how big are your map files in MB, how much ram and swap space do you have? I've seen this error on both suse 10.0 >>and kubuntu (3.10? whatever the lastest apt-get dist upgrade >>got me; two different machines) using source builds of >>pymol-0.98, and 0.99rc6 with ext-0.98,ext-0.99rc6, and ext-0.95. >> >>Any suggestions on how to fix this (or get debugging >>information out of pymol to narrow down the problem)? I haven't the slightest clue. Does this happen immediately after loading the map or while rotating or what? But I can tell you that I'm running gentoo, w/vanilla sources 2.6.15 running in x86_64 mode (straight from kernel.org), 1.0.8178 nvidia drivers on a quadro4 980xgl video card. You might try as was suggested to upgrade your kernel. You can get 2.6.16 from kernel.org, copy your current .config into the root of the source directory, run make oldconfig, hit enter a bunch of times, then compile the kernel with make && make modules_install. You will have to recompile the nvidia drivers for the new kernel too. Hope that helps, Sabuj Pattanayek |
From: Peter A. M. <pa...@co...> - 2006-03-24 02:14:22
|
> same here, how big are your map files in MB, how much ram and swap space > do you have? ~5MB/map, 512MB RAM, ~2GB swap. I'd expect a segfault (or an exception at least) if I was overloading the memory/swap, and CPU would drop if it was running out of physical RAM and swapping too much. > I haven't the slightest clue. Does this happen immediately after loading > the map or while rotating or what? But I can tell you that I'm running > gentoo, w/vanilla sources 2.6.15 running in x86_64 mode (straight from > kernel.org), 1.0.8178 nvidia drivers on a quadro4 980xgl video card. No consistent pattern. Both machines are x86, non-nvidia graphics cards (which probably isn't helping). But, score at least one working setup (yours) as a fully functional combination. > You might try as was suggested to upgrade your kernel. You can get > 2.6.16 from kernel.org, copy your current .config into the root of the > source directory, run make oldconfig, hit enter a bunch of times, then > compile the kernel with make && make modules_install. You will have to > recompile the nvidia drivers for the new kernel too. It might come to that, but it'll be a last resort. I'm reluctant to make major modifications to a system that I'm using to do research in the middle of a project (I've had 2 independent backups fail me at the same time). Particularly for a problem that isn't predictably reproducable, so I wouldn't know if an upgraded kernel/module had actually fixed the problem or not (it could be done w\ testing livecd distributions or within a vm, or trying to get a debug build of pymol talking happily to a debugger, but at a certain point it becomes more trouble to track down and fix the bug than it does to deal with it). Thanks for the suggestions, Pete Pete Meyer Fu Lab BMCB grad student Cornell University |
From: Sabuj P. <sab...@va...> - 2006-03-24 16:27:26
|
> No consistent pattern. Both machines are x86, non-nvidia graphics cards > (which probably isn't helping). But, score at least one working setup > (yours) as a fully functional combination. Yes always go with nvidia on linux x86/x86_64 if you can. Do you have an ati/sis/3dlabs/intel product? |
From: Peter A. M. <pa...@co...> - 2006-03-24 01:56:31
|
Warren, Thanks for getting back to me so quickly. It's good to hear that this problem has been seen before (it's been one of those weeks for strange computer issues). I guess it shouldn't supprise me that there are issues with proprietary OpenGL on linux (FYI the two machines I've seen this on are using ATI and intel graphics cards). I'll try out the options you mentioned. Your idea about known-working (and non-working) combinations of hardware seems like a very good thing to do. Pete > Pete, > > We have been suffering this same exact symptom with our nVidia-base 64-bit RHEL workstation for over a year. I have spent enough time troubleshooting the issue to have convinced myself that the problem isn't PyMOL -- it is an inability of certain combinations of hardware, Linux kernels, and proprietary OpenGL drivers to handle the vast quantity of immediate mode rendering geometry that PyMOL generates. It seems to be either a sporadic buffer overflow or a race condition within the kernel layer. > > The proprietary nature of the OpenGL drivers prevents anyone other than the hardware vendor from being able to actually understand and solve the problem. Plus, the problem is sporadic -- most of the time everything works fine, but sometimes PyMOL hangs, sometimes all of X11 freezes, and sometimes the whole system comes crashing down. The intractable nature of these Linux stability problems is one of the primary reasons why I recommend Mac OS X over Linux for unix-based PyMOL visualization. > > Note that the settings "nvidia_bugs" and "use_display_lists" may reduce the frequency of the problem on your system. Other than that, my suggestion would be to try different Linux kernels and OpenGL driver versions. > > The combinatorial matrix of possibilities is huge, but maybe we should create a table on the Wiki for people to post working and non-working combinations of Linux kernels, cpu and graphics hardware, and opengl drivers? That's how we used to do things in the old days. I suspect only 5-10% of configurations are truly unworkable, and once enough data is collected from enough people, meaningful patterns would likely emerge. > > Cheers > Warren > > -- > Warren L. DeLano, Ph.D. > Principal Scientist > > . DeLano Scientific LLC > . 400 Oyster Point Blvd., Suite 213 > . South San Francisco, CA 94080 USA > . Biz:(650)-872-0942 Tech:(650)-872-0834 > . Fax:(650)-872-0273 Cell:(650)-346-1154 > . mailto:wa...@de... > > >> -----Original Message----- >> From: pym...@li... >> [mailto:pym...@li...] On Behalf Of >> Peter Adrian Meyer >> Sent: Thursday, March 23, 2006 9:32 AM >> To: pym...@li... >> Subject: [PyMOL] sporadic pymol/X-server freezes? >> Hi everyone, >> Has anyone else run into cases where pymol will freeze the >> X-server (pymol doesn't repond to mouse or keyboard input; >> keyboard non-responsive to caps-lock on/off)? The system is >> still stable, I'm able to login remotely and see that pymol >> is running at 100% CPU. Killing pymol results in xorg >> running at 100% CPU (killing the xorg process as root doesn't >> do anything). >> I've only seen this while looking at several maps/masks (but >> that's most of what I use pymol for, so I'm not sure if >> that's the cause). I've seen this error on both suse 10.0 >> and kubuntu (3.10? whatever the lastest apt-get dist upgrade >> got me; two different machines) using source builds of >> pymol-0.98, and 0.99rc6 with ext-0.98,ext-0.99rc6, and ext-0.95. Any suggestions on how to fix this (or get debugging >> information out of pymol to narrow down the problem)? >> Thanks, >> Pete >> Pete Meyer >> Fu Lab >> BMCB grad student >> Cornell University >> ------------------------------------------------------- >> This SF.Net email is sponsored by xPML, a groundbreaking >> scripting language that extends applications into web and >> mobile media. Attend the live webcast and join the prime >> developer group breaking into this new coding territory! >> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720& >> dat=121642 >> _______________________________________________ >> PyMOL-users mailing list >> PyM...@li... >> https://lists.sourceforge.net/lists/listinfo/pymol-users > Pete Meyer Fu Lab BMCB grad student Cornell University |
From: Tim F. <fe...@st...> - 2006-03-24 08:30:40
|
On Thu, Mar 23, 2006 at 08:56:27PM -0500, Peter Adrian Meyer wrote: > > I guess it shouldn't supprise me that there are issues with proprietary > OpenGL on linux (FYI the two machines I've seen this on are using ATI and > intel graphics cards). I'll try out the options you mentioned. > So the ATI card is using the fglrx driver, I'm assuming (right?), and is the intel card using the Intel graphics drivers from intel.com? I've been told i810 that comes with XOrg is actually better (besides having full 3d support)... but what intel chipset are you using? Have you tried the NON-proprietary drivers? > Your idea about known-working (and non-working) combinations of hardware > seems like a very good thing to do. > If its a proprietary driver, my money is on a buggy driver that doesn't cooperate with the kernel. > > > > > >> Has anyone else run into cases where pymol will freeze the > >> X-server (pymol doesn't repond to mouse or keyboard input; > >> keyboard non-responsive to caps-lock on/off)? The system is > >> still stable, I'm able to login remotely and see that pymol > >> is running at 100% CPU. Killing pymol results in xorg > >> running at 100% CPU (killing the xorg process as root doesn't > >> do anything). So xorg quits, but the monitor stays frozen? Or the xorg process doesn't die? If its the latter, can you just send it a KILL (try an INT signal first, of course)? Anything in dmesg or the XOrg logs? Can you grab the XOrg pid with gdb and get a stack trace? Can you switch over to a console (ctrl-alt-f1, typically)? -Tim -- --------------------------------------------------------- Tim Fenn fe...@st... Stanford University, School of Medicine James H. Clark Center 318 Campus Drive, Room E300 Stanford, CA 94305-5432 Phone: (650) 736-1714 FAX: (650) 736-1961 --------------------------------------------------------- |