In the http://cerebralvol.com service we are having 23.5 hours average with 5.1 and 23.8 hours average with 5.2-Beta
Notice that in order to minimize the cost to the user we are running it in m1.medium We have achieved a full recon-all in less than 4 hours (3.83 hours) with cg1.4xlarge But for a large amount of data I believe our 1024-core m1.medium is the best cost benefit. --------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441 On Sat, Feb 2, 2013 at 2:11 AM, Mehul Sampat <mpsam...@gmail.com> wrote: > Hi Bruce, > No I did not specify the # of open mp threads on the recon-all cmd line. > These run times were obtained by running one subject per core. for example > cc2.8xlarge has 8 cores and so we ran 8 subjects at once; > Thanks for the info about the # of open mp threads options; I will look > into it. > > One other note: Bruce, Nick did you improve the memory management in 5.2 ? > On our local machine we noticed we can run 6 subjects simultaneously even > though we only have 12gb of ram. > I thought some of the them might crash since we only have 2gb per subject > but no crashes so far over 30 subjects.. > Mehul > > > On Fri, Feb 1, 2013 at 7:12 PM, Bruce Fischl > <fis...@nmr.mgh.harvard.edu>wrote: > >> Hi Mehul >> >> did you specify the # of open mp threads on the recon-all cmd line? >> cheers >> Bruce >> >> On Fri, 1 Feb 2013, Mehul Sampat wrote: >> >> Hi Folks, >>> Just wanted to share our experience with running FS 5.2-beta on Amazon >>> Web >>> Services (AWS). >>> Basically, AWS has multiple instance types >>> (http://aws.amazon.com/ec2/**instance-types/<http://aws.amazon.com/ec2/instance-types/>) >>> and we were trying to figure out >>> the most cost-effective approach. >>> >>> We ran two subjects through FS 5.2-beta on M1 Large Instance (m1.large) >>> and Cluster Compute Eight Extra Large Instance (cc2.8xlarge). (same >>> subjects >>> run on both instance). We expected cc2.8xlarge to be faster (but it is >>> also >>> more expensive: $2.4 per hour; 8 cores); The run-times we got: >>> >>> instance-type subject start-time end-time run-time >>> m1.large subject-1 01:05:44 UTC 2013 15:40:45 UTC 2013 ~14hr-35mins >>> m1.large subject-2 01:06:06 UTC 2013 15:08:45 UTC 2013 ~14hr-02mins >>> cc2.8xlarge subject-1 01:26:38 UTC 2013 12:30:23 UTC 2013 ~11hr-04mins >>> cc2.8xlarge subject-2 01:27:28 UTC 2013 12:19:08 UTC 2013 ~10hr-52mins >>> >>> Although m1.large is a few hours slower, it seems to be the more cost >>> effective option since it is $0.24 per hour (2 cores). If you have run >>> Freesurfer on AWS, do you have a similar experience ? Any suggestions to >>> speed up the run-times on AWS would be very helpful. >>> >>> Thanks >>> Mehul >>> >>> >>> >>> >> >> 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<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 > > > 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 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.