16 Jun
2013
16 Jun
'13
1:55 p.m.
I think the problem is that the 2mm fsaverage volume is 76 76 93 = 537168 voxels. Eventhough most of these are 0s, the full 4D volume (after decompressing) must be represented at some point as it gets read in and written out. doug
On 6/15/13 11:56 AM, Marcel Falkiewicz wrote:
Hello, actually, I've noticed something weird. The surface analysis runs correctly with ~150k in the mask. However, there are only ~50k voxels in the subcortical mask, so it should require 3x less memory to process, but it crashes with OUT OF MEMORY message.
Best regards, Marcel
2013/6/13 Douglas Greve <greve@nmr.mgh.harvard.edu mailto:greve@nmr.mgh.harvard.edu>
Good idea. Actually, I need to fix selxavg3-sess to handle this better. When I wrote it, I never tested it on runs with so many time points, but they are becoming more common. doug On 6/13/13 5:26 PM, Marcel Falkiewicz wrote:Thanks Jacek for the advice, I'll try to upgrade to newer MATLAB version, maybe it will manage the memory better. I've got 64 bit versions of everything, so this doesn't really help. I've never considered a 800 timepoint timeseries as a large dataset, other software packages do not seem to be bothered by this, even SPM which is Matlab-based. Doug, I will try to split the dataset as you advised, it will help for sure, but given the fact that this problem occurs often, maybe dataset splitting should be automated in fsfast in order to avoid such problems in the future. Thanks again for your help guys. 2013/6/13 Jonca, Jacek <jacek.jonca@ttu.edu <mailto:jacek.jonca@ttu.edu>> In the past (for other people) selxavg3 failed with as much as 16GBs although as always much depends on your dataset. Most likely the problem is not with the fact that you don't have enough memory but rather that not enough memory is being allocated to Matlab. Are you running it on a 64-bit machine with 64-bit version of Ubuntu installed and 64-bit version of Matlab? If not this may be your best bet right now. In addition, as always having most up-to-date version of both Ubuntu and Matlab may be helpful as well. I noticed that you are running R2011a, that is a tad bit old. I think that R2013a has been out for some time. Do you have access to a cluster where you can deploy freesurfer, Sunsite? In my experience those tended to work better for running computationally demanding analyses. I simply rsynced my remote directory with my laptop to examine results from the laptop but run the actual anayses on the mainframe. Cheers, Jacek cid:image003.png@01CC924F.6F547760 Jacek Jon'ca-Jasin'ski, Ph.D. | The Graduate School | Texas Tech University Administration Bldg 327 / 40D | 2625 Memorial Circle | Box 1030 | Lubbock, TX 79409-1030 ph.: 806.834.3348 <tel:806.834.3348> | fax: 806.742.4038 <tel:806.742.4038> | http://www.gradschool.ttu.edu <http://www.gradschool.ttu.edu/> Capture *From:*freesurfer-bounces@nmr.mgh.harvard.edu <mailto:freesurfer-bounces@nmr.mgh.harvard.edu> [mailto:freesurfer-bounces@nmr.mgh.harvard.edu <mailto:freesurfer-bounces@nmr.mgh.harvard.edu>] *On Behalf Of *Marcel Falkiewicz *Sent:* Thursday, June 13, 2013 1:19 PM *To:* freesurfer@nmr.mgh.harvard.edu <mailto:freesurfer@nmr.mgh.harvard.edu> *Subject:* [Freesurfer] Fwd: Out of memory problem during selxavg3-sess Hello, I know this problem has been discussed earlier, but I think in my case the source of the problem might be quite different. I'm running FS 5.3.0 and MATLAB R2011a on Ubuntu machine with 12 GB RAM. selxavg3-sess runs perfectly for both hemis, but I get the following error when trying to run it on subcortical volume: $Id: fast_selxavg3.m,v 1.100.2.2 2012/11/30 18:40:38 greve Exp $ /home/marcel/freesurfer/fsfast/toolbox/fast_selxavg3.m /home/marcel/freesurfer/fsfast/toolbox/fast_ldanaflac.m /home/marcel/freesurfer/matlab/MRIread.m ------------------------- outtop = /media/2TB/SDT_MRI/FS Extension format = nii.gz 1 DC_vs_HC.mat 2 HI_vs_DC.mat 3 HI_vs_HC.mat 4 NAT_vs_PAL.mat 5 NAT_vs_PAT.mat 6 PAL_vs_NAT.mat 7 PAL_vs_PAT.mat nruns = 1 autostimdur = outanadir = /media/2TB/SDT_MRI/FS/AC1004/bold/sdt.sm8.mni Found 52644/537168 ( 9.8) voxels in mask Creating Design Matrix ... creation time = 0.007 sec DoMCFit = 1 ntptot = 851, nX = 19, DOF = 832 Saving X matrix to /media/2TB/SDT_MRI/FS/AC1004/bold/sdt.sm8.mni/Xtmp.mat XCond = 31.0456 (normalized) Computing compensation for resdual AR1 bias 1 -0.5 -0.500489 (t=0.097569) 2 -0.25 -0.255326 (t=0.427178) 3 0 -0.011955 (t=0.516085) 4 0.25 0.229273 (t=0.839369) 5 0.5 0.466983 (t=0.928175) AR1 Correction M: 0.0147777 1.03322 Computing contrast matrices OLS Beta Pass run 1 t= 0.0 Global Mean 1120.98 Global In-Mask Mean = 1120.98 (983.432) Rescale Target = 100 RescaleFactor = 0.0892077 OLS Residual Pass run 1 t= 0.0 ??? Out of memory. Type HELP MEMORY for your options. Error in ==> fast_selxavg3 at 489 rsserun = sum(rrun.^2); >> ------------------------------------------ I observed memory consuption with top(8) and noticed that around 11 GBs are eaten after about 30 second of running OLS residual pass. Do you think adding more memory to the PC would solve the problem, or is it some MATLAB/FS-related bug? Best regards, Marcel _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
