<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 10pt;
font-family:Verdana
}
--></style>
</head>
<body class='hmmessage'>
Hi Guys,<br><br>Great toolbox, keep up the good work. I have a question about speeding up the Searchlight algorithm.<br><br>When I run the algorithm, it tends to start off quickly and then progressively slow down to a snail's pace. Windows reports that pythonw.exe, at it's peak, is only consuming ~250mbs of memory (despite the availibility of up to 14gbs of RAM and a massive pagefile).<br><br>My question is, is there some way to increase the memory being used and thus speed up the process? Windows limits 32bit programs to 2gbs and provides memory upon request and there does not seem to be any reason that Windows would refuse python any memory?<br><br>Also, it seems counterintuitive to me that Searchlight would start off quick and then slow down, surely each voxel should take the same amount of time? Can you think of anything that might be causing this?<br><br>
OS: Windows Vista Business 64bit (running Python in 32bit mode)<br>
Xeon E5440 processor<br>
16gb RAM<br>
<br>
#dataset is the accompanying Haxby study<br>
#nifti images and mask load fine<br>
clf = LinearNuSVMC(nu=0.1)<br>
cv = CrossValidatedTransferError(TransferError(clf),<br>
NFoldSplitter())<br>
sl = Searchlight(cv, radius=3)<br>
sl_map = sl(dataset)<br><br>Regards,<br><br>Dan<br><br>                                            <br /><hr />Hotmail: Trusted email with Microsoft’s powerful SPAM protection. <a href='https://signup.live.com/signup.aspx?id=60969' target='_new'>Sign up now.</a></body>
</html>