new MARK build

[cooch]

Gary has been working on the MARK code -- and you'll want to download the newest version fairly soon, especially if you're working with 'large dimension' problems (lots of occasions x lots of groups). Thee code related to use of multiple processors has been tweaked, significantly improving execution times for some problems.

Also, Gary has added a new 'preference', wherein you can force MARK to use the 32-bit version of mark.exe, even if you're using a 64-bit machine (MARK normally 'looks' at the architecture of your computer, and picks the version of mark.exe which equals the bit-length of the OS on the machine). But, for some technical reasons, especially for some problems, 32-bit mark.exe will run faster than 64-bit mark.exe. So, you can no force MARK to use 32-bit mark.exe, if you want.

You can download latest version of MARK here:

http://www.phidot.org/software/mark/downloads/

(Note: Linux version of MARK has also been updated. Mac OS/X version will be updated next week sometime).

[cooch]

More details, and some even more recent changes (you'll want to update to versions posted December 28 or later). Meaning, even if you updated 2-3 days ago, you'll want to do so again.

Some particulars on the changes:

A rather severe problem was uncovered with the gfortran compiler when a "large" problem is optimized in the mark.exe code. The problem occurred on all machines running the gfortran code: PC, Cray, or Unix. Specifically, when a design matrix of dimension 460 X 460 was used, the cpu time to allocate thread-specific copies for the multiple threads was ~100 times what would take a run without multiple threads. Several changes were made to circumvent this problem. For analyses without individual covariates, only the original design matrix is needed -- not multiple copies -- so a test to determine this condition was added and processing without multiple copies then proceeds. Further, an option was added to not use parallel processing with a single thread, and thus avoid the overhead of the OpenMP with multiple threads. If you find that a "large" design matrix with individual covariates is taking an exorbitant amount of time, try specifying threads=1.

As you might surmise from the preceding, compiling for multiple threads/cores is non-trivial.

[cooch]

Mac OS/X version will be updated next week sometime)...

Update applied and upload 5 minutes ago...

[dhpstroud]

Is anyone else having trouble getting the new download up-and-running? I'm getting an error message that refers to line6 and line16.

[cooch]

Is anyone else having trouble getting the new download up-and-running? I'm getting an error message that refers to line6 and line16.

Runs fine for me on 5-6 different machines. Only suggestiojn is to uninstall, and re-download (using latest version):

http://www.phidot.org/software/mark/downloads/

[cooch]

Is anyone else having trouble getting the new download up-and-running? I'm getting an error message that refers to line6 and line16.

I'm assuming you mean the Windows version - yes?

[dhpstroud]

Hm. Yup, I've done that a few times now. I'm using windows 7. The issue must be on my end. This is a new machine and I haven't quite broken it in...

[dhpstroud]

Huh. I'm still having issues over here. It's working beautifully elsewhere, but not on my work laptop. Maybe you could give me some direction. It installs just fine, but when I try and open it I get a windows error message:
----------------------------------------------------
Error Code: 11 ()
Subsystem: BASE
Description: Dynamic Memory Low (*sidenote, I checked my memory and there's no real issue*)
CallStack:
APP:START (Line:16)
START (Line:6)
-------------------------------------------------
Any ideas what this is about?

[cooch]

Huh. I'm still having issues over here. It's working beautifully elsewhere, but not on my work laptop. Maybe you could give me some direction. It installs just fine, but when I try and open it I get a windows error message:
----------------------------------------------------
Error Code: 11 ()
Subsystem: BASE
Description: Dynamic Memory Low (*sidenote, I checked my memory and there's no real issue*)
CallStack:
APP:START (Line:16)
START (Line:6)
-------------------------------------------------
Any ideas what this is about?

Works elsewhere, works for me (and a lot of people), but doesn't work on your laptop.

Deep scientific conclusion: the problem is your laptop.

Sorry, can't do much more than that. I suppose it could be some weird interaction with Windows 8 (if that applies), but beyond that.

Try booting the machine into SafeMode, and try running it from there.

[gwhite]

This error sounds like you don't have enough memory on the notebook to run the version of MARK that I distribute. How much memory?

Gary