<div dir="ltr">I pinpointed the problem to being the access time of the second HDD mounted as a ntfs filesystem. Not having this HDD in my searchpath solves my problem. <div><br></div><div>Berry<br><br><div class="gmail_quote">
---------- Forwarded message ----------<br>From: <b class="gmail_sendername">Gio Piantoni</b> <span dir="ltr"><<a href="mailto:gio@gpiantoni.com" target="_blank">gio@gpiantoni.com</a>></span><br>
Date: 20 January 2014 14:01<br>Subject: Re: [FieldTrip] ft_statistics_montecarlo runs slow under ubuntu 13.10<br>To: berry van den berg <<a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a>><br>
<br><br>
sorry, I don't know much more than this, but it makes sense that Linux<br>needs some extra time to access a non-native filesystem. If I were<br>you, I'd just comment out the part in ft_statistics_montecarlo that<br>
checks for the toolbox, once you know that the toolbox is installed,<br>you don't need to check it every time.<br>Good luck!<br><br>On Mon, Jan 20, 2014 at 12:56 PM, berry van den berg<br><div><div>
<<a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a>> wrote:<br>> Yeah, you are right, having a folder on that harddisk added to the search<br>> path slowed those functions (which, exist) by A LOT! I copied fieldtrip to<br>
> the main SSD and removed everything from the path in matlab on the ntfs<br>> drive, which is mounted through ntfs-3g. Even though I dont actually use<br>> those functions, it slows up the process by a lot: 1 second versus 6 seconds<br>
> when fieldtrip is on the ssd with ext4....<br>><br>> I wonder if it is due to the drive being ntfs, or something else... Any<br>> ideas?<br>><br>><br>> On 20 January 2014 12:00, Gio Piantoni <<a href="mailto:gio@gpiantoni.com" target="_blank">gio@gpiantoni.com</a>> wrote:<br>
>><br>>> Hi Berry,<br>>><br>>> interesting debugging. Not sure exactly what's going on, but I noticed<br>>> that Linux might become slower if you have samba/cifs disks mounted.<br>>> Is that the case for you maybe?<br>
>><br>>> HTH,<br>>> -g<br>>><br>>> On Mon, Jan 20, 2014 at 11:45 AM, berry van den berg<br>>> <<a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a>> wrote:<br>
>> > Ok, I dove a bit deeper into the problem, using the matlab profiler I<br>
>> > was<br>>> > able to pinpoint the problem to ft_hastoolbox.m called by findcluster.m,<br>>> > and<br>>> > specifically the functions fileparts and exist.... Copy pasting those<br>>> > two<br>
>> > functions to ft_statistics_montecarlo solves the issue for me for now.<br>>> ><br>>> > The problem seems to be that matlab accessing my filesystem runs slow<br>>> > under<br>>> > linux compared to windows.. I have no idea why and how to solve it but<br>
>> > it is<br>>> > not related to fieldtrip. If anyone has suggestions what this might be I<br>>> > would be glad to hear them!<br>>> ><br>>> > Cheers,<br>>> ><br>>> ><br>
>> ><br>>> ><br>>> > On 16 January 2014 17:11, berry van den berg <<a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a>><br>>> > wrote:<br>>> >><br>
>> >> Dear Fieldtrip experts,<br>
>> >><br>>> >> This might be an odd question, but maybe someone has an idea where to<br>>> >> start.<br>>> >><br>>> >> I just got a fancy new laptop (gigabyte p34g) and dual boot with ubuntu<br>
>> >> and windows. I usually work in Ubuntu for analysis, so I ran a time<br>>> >> freq<br>>> >> statistics analysis and noticed that ft_statistics_montecarlo runs<br>>> >> extremely<br>
>> >> slow under Ubuntu.... In windows it runs at normal speed. The<br>>> >> difference is<br>>> >> huge, 97 seconds vs, 2 seconds for 100 iterations, 24 subjects.<br>>> >><br>
>> >> Speed also doesnt seem influenced by averaging over freq or/and time,<br>>> >> it<br>>> >> is just slow.<br>>> >><br>>> >> It is not the cpu clock speed (ubuntu nicely utilizes turbo mode,<br>
>> >> running<br>>> >> max 3ghz), the cpu is not fully utilized though (only 30 percent or<br>>> >> so)...<br>>> >><br>>> >> I run matlab 2013b, fieldtrip 20140115<br>
>> >><br>>> >> Specs are<br>>> >> 8gb ram; only 4gb utilized.<br>>> >> 4700HQ cpu<br>>> >><br>>> >> Any ideas, because I am clueless<br>>> >><br>
>> >> Cheers,<br>>> >><br>>> >> --<br>>> >> Berry van den Berg<br>>> >> <a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a><br>>> ><br>
>> ><br>
>> ><br>>> ><br>>> > --<br>>> > Berry van den Berg<br>>> > <a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a><br>>> ><br>>> > _______________________________________________<br>
>> > fieldtrip mailing list<br>>> > <a href="mailto:fieldtrip@donders.ru.nl" target="_blank">fieldtrip@donders.ru.nl</a><br>>> > <a href="http://mailman.science.ru.nl/mailman/listinfo/fieldtrip" target="_blank">http://mailman.science.ru.nl/mailman/listinfo/fieldtrip</a><br>
><br>><br>><br>><br>> --<br>> Berry van den Berg<br>> <a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a><br></div></div></div><br><br clear="all"><div><br></div>-- <br>
Berry van den Berg<br>
<a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a><br></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On 20 January 2014 11:45, berry van den berg <span dir="ltr"><<a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Ok, I dove a bit deeper into the problem, using the matlab profiler I was able to pinpoint the problem to ft_hastoolbox.m called by findcluster.m, and specifically the functions fileparts and exist.... Copy pasting those two functions to ft_statistics_montecarlo solves the issue for me for now. <div>
<br></div><div>The problem seems to be that matlab accessing my filesystem runs slow under linux compared to windows.. I have no idea why and how to solve it but it is not related to fieldtrip. If anyone has suggestions what this might be I would be glad to hear them!</div>
<div><br></div><div>Cheers,</div><div><br></div><div><br></div></div><div class="HOEnZb"><div class="h5"><div class="gmail_extra"><br><br><div class="gmail_quote">On 16 January 2014 17:11, berry van den berg <span dir="ltr"><<a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Dear Fieldtrip experts,<div><br></div><div>This might be an odd question, but maybe someone has an idea where to start.<br>
<div><br></div><div>I just got a fancy new laptop (gigabyte p34g) and dual boot with ubuntu and windows. I usually work in Ubuntu for analysis, so I ran a time freq statistics analysis and noticed that ft_statistics_montecarlo runs extremely slow under Ubuntu.... In windows it runs at normal speed. The difference is huge, 97 seconds vs, 2 seconds for 100 iterations, 24 subjects.</div>
<div><br></div><div>Speed also doesnt seem influenced by averaging over freq or/and time, it is just slow. </div><div><br></div><div>It is not the cpu clock speed (ubuntu nicely utilizes turbo mode, running max 3ghz), the cpu is not fully utilized though (only 30 percent or so)... </div>
<div><br></div><div>I run matlab 2013b, fieldtrip 20140115</div><div><br></div><div>Specs are </div><div>8gb ram; only 4gb utilized.</div><div>4700HQ cpu</div><div><br></div><div>Any ideas, because I am clueless </div><div>
<br></div><div>Cheers,</div><span><font color="#888888"><div><div><br></div>-- <br>Berry van den Berg<br><a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a>
</div></font></span></div></div>
</blockquote></div><br><br clear="all"><div><br></div>-- <br>Berry van den Berg<br><a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a>
</div>
</div></div></blockquote></div><br><br clear="all"><div><br></div>-- <br>Berry van den Berg<br><a href="mailto:berryv.dberg@gmail.com" target="_blank">berryv.dberg@gmail.com</a>
</div>