OSX Mavericks

[mcnelson.phd]

Hey guys,

Has anyone updated any of their Macs that they use for bioinformatics to Mavericks yet? We have a Mac Pro running 10.6.8 that we use for the heavy data analysis that we don't do on our linux clusters, as well as some iMacs that we use for lighter analyses and I'm considering whether or not it's worth it to upgrade them. The AppNap and compressed memory features in particular seem like they could give those systems a boost and help extend their usable life span.

Given that the upgrade is free, it's very tempting to just go ahead and try it, but my concern is that there will be unknown issues with some of the programs that we rely on for our analyses. Figure it was worth checking to see if others have upgraded yet and what their impressions were before I try it for myself.

[JohnN]

Because on my heavy reliance on MacPorts, I'm waiting for verification that MacPorts works with Mavericks. (I have not checked since Friday).

[rhinoceros]

I upgraded my non-critical home laptop to Maverics and didn't have any issues yet. I don't use MacPorts however..

[mdobeson]

There were issues which you could work around by compiling from sources, but i believe macports have released an official installer. as of yesterday i had issues with the mac-qt4 library not compiling. its used for quite a few GUI apps in macports and i'm waiting for them to update. so you've been warned.

[PeteH]

There are known problems with Mavericks and RStudio (http://blog.rstudio.org/2013/10/22/r...0-9-mavericks/). R itself is fine but it's something to bear in mind if you use RStudio and are considering moving to Mavericks.

[jimmybee]

Because on my heavy reliance on MacPorts, I'm waiting for verification that MacPorts works with Mavericks. (I have not checked since Friday).

Ive had no issues

[GenoMax]

Waiting for 10.9.1 to show up may be prudent (at the most a couple of months). Apple generally tackles most of the initial bugs with that minor release.

That said, commercial software we support seems to all work with Mavericks (just needed the java install which is still on 1.6). There is an Xcode update to go along as well so there should be compatible compiler/libraries in case you need to recompile things. Between the OS and Xcode you are looking at a ~7 GB download.

[mdobeson]

The todo list...
http://trac.macports.org/wiki/MavericksProblems

[sinnamone]

I have updated my macbook(mid 2010) and i CLC works fine.

[fahmida]

So far so good, no issues to report yet. GUI one's such as CLC, Tablet etc. seems coped well.Hoping for no dramas.

[mcnelson.phd]

Looks like my best bet may be to update one of our less crucial systems and see what works and what doesn't.

Thanks for the info guys!

[rhinoceros]

I updated my work computer and ran into problems.

Apparently, OS X Mavericks memory management sucks. I can no longer run some scripts I've written for 454-sequence analysis. Instead, in memory heavy phases of these scripts, a warning that I'm out of memory pops up. A few seconds later the computer freezes completely and has to be cold rebooted. Mind you, I never had these problems in Mountain Lion with the same exact parameters and data. The only workaround is to reduce parallelization.

Computer specs:

MacBook Pro
Retina, 15-inch, Early 2013
Processor 2,7 GHz Intel Core i7
Memory 16 GB 1600 MHz DDR3
Graphics NVIDIA GeForce GT 650M 1024 MB
Software OS X 10.9 (13A603)

Oh well, back to Mountain Lion. I'm somewhat sad as Mavericks handles multiple screens way better than its predecessor.

[mdobeson]

I can no longer run some scripts I've written for 454-sequence analysis. Instead, in memory heavy phases of these scripts, a warning that I'm out of memory pops up. A few seconds later the computer freezes completely and has to be cold rebooted. Mind you, I never had these problems in Mountain Lion with the same exact parameters and data. The only workaround is to reduce parallelization.

Bugger! What are your scripts written in? What libraries are you using? Are you hitting swap space?

My 454 perl scripts use ForkManager extensively and are pretty cpu intenstive but I'm careful to stay away from hitting swap hard for obvious performance reasons. I've had no memory issues thus far, just missing macports. Looks like we have the same spec macbook.

Mavericks does have that new memory compression feature which seems a possible culprit if this is indeed a mavericks bug. Who knows! YMMV

[jimmybee]

I updated my work computer and ran into problems.

Apparently, OS X Mavericks memory management sucks. I can no longer run some scripts I've written for 454-sequence analysis. Instead, in memory heavy phases of these scripts, a warning that I'm out of memory pops up. A few seconds later the computer freezes completely and has to be cold rebooted. Mind you, I never had these problems in Mountain Lion with the same exact parameters and data. The only workaround is to reduce parallelization.

Computer specs:

MacBook Pro
Retina, 15-inch, Early 2013
Processor 2,7 GHz Intel Core i7
Memory 16 GB 1600 MHz DDR3
Graphics NVIDIA GeForce GT 650M 1024 MB
Software OS X 10.9 (13A603)

Oh well, back to Mountain Lion. I'm somewhat sad as Mavericks handles multiple screens way better than its predecessor.

Do you use chrome by any chance?

I had similar issues with my system but has since cleared when I switched to safari. Ive been really throttling the system recently and its been handling parallelisation better than ever. Specially with high memory tasks

[rhinoceros]

Bugger! What are your scripts written in? What libraries are you using? Are you hitting swap space?

My 454 perl scripts use ForkManager extensively and are pretty cpu intenstive but I'm careful to stay away from hitting swap hard for obvious performance reasons. I've had no memory issues thus far, just missing macports. Looks like we have the same spec macbook.

Mavericks does have that new memory compression feature which seems a possible culprit if this is indeed a mavericks bug. Who knows! YMMV

These are shell scripts that are basically wrappers for QIIME and indeed, I'm hitting swap space. I think running pick_open_reference_otus.py script from QIIME with 8 threads against 97_greengenes stayed just a little bit under 16GB in ML, however, for whatever reason Mavericks goes way above it. The memory compression thingy was my first idea too..