There's a variety of things you can do to help get some reliability and
performance out of your system. As odd as it may sound, look up how
gamers tune thier systems to run games with a minimum of hickups.
Some techinques include making a hardware profile, using msconfig to
disable startup programs, turning off the anti virus software while
running a performance app (the single greatest performance hole for most
computers these days). You can also use batch files to automatically
push your program to high priority or dedicate a core to it by taking
all other processes off that core (this however can lead to instability
if certain processes are taken off the core). there's tons of
information online for tricks that different tweaking communities have
come up with to reduce the 'kruft' processing in windows systems.
-Garrett
d'Acremont Mathieu wrote:
> Thanks for your answers so far. The interest of using Java is that we can re-use the code to make behavioral group experiments in our lab (in a client-server configuration).
>
> The motivation of my post was mainly about the *synchronization* with the scanner. Can Java talk with the scanner and how should we implemented it (e.g. with Java Native Interface, JNI). Is there any piece of code which has been written for that?
>
> For time accuracy, it appears from your posts that Java is not so good at it. But do we really need 1/1000 [s] accuracy when the typical TR in fMRI is 2 [s]? And there is always the law of large number that make us converge to the mean by repeating measures. So 1/100 [s] time accuracy is probably enough.
>
> Of course we don't want the OS to start scanning the hard drive during the experiment. I wonder if this can be controlled in Windows (which is the OS installed in the brain imaging lab) by prioritizing the java process.
>
> Cheers,
>
>
|