Issue/Bug with Screen flip command

1

Hello,


I will try to be as detailed as possible. We are having difficulties with stimuli being presented on a dual monitor setup. Recently, at random times at the beginning of a task, the flip command seems to fail, producing 1 of 3 errors: a pure white screen, a pure black screen, or a visually distorted mix of a black screen and the desktop showing some icons and the wallpaper. We have used this particular code for years but the issue is only about 1.5 months old - however we recently (~6 months ago) had to upgrade our machines and MATLAB versions which could also be a factor. MATLAB/Psychtoolbox itself doesn't seem to produce an error in the command window nor does the program terminate. We have gone through this with our IT department but they have run out of ideas. Below is the software and hardware versions we are using. If you could provide any insight or ideas we would greatly appreciate it.


Thank you,

Dan



MacbookPro Retina Display, 13-inch, Mid-2014

MATLAB 2015a

Psychtoolbox 3.0.13



2
You didn't mention the OSX version, but this sounds like an operating system/graphics driver bug. Nothing in PTB could cause such a behavior. Maybe the latest OSX update introduced some funny new graphics system bugs? Downgrading the OS might help. Or upgrading to Linux.

Message 21056 sounds similar, in that an OSX systems gamma table handling just broke somewhere during the last few months.

-mario

XX---In PSYCHTOOLBOX@yahoogroups.com, <p7ewlfy4wlyen6uo4vftuo7v77oclidk2n3zgbvj@...> wrote :

Hello,


I will try to be as detailed as possible. We are having difficulties with stimuli being presented on a dual monitor setup. Recently, at random times at the beginning of a task, the flip command seems to fail, producing 1 of 3 errors: a pure white screen, a pure black screen, or a visually distorted mix of a black screen and the desktop showing some icons and the wallpaper. We have used this particular code for years but the issue is only about 1.5 months old - however we recently (~6 months ago) had to upgrade our machines and MATLAB versions which could also be a factor. MATLAB/Psychtoolbox itself doesn't seem to produce an error in the command window nor does the program terminate. We have gone through this with our IT department but they have run out of ideas. Below is the software and hardware versions we are using. If you could provide any insight or ideas we would greatly appreciate it.


Thank you,

Dan



MacbookPro Retina Display, 13-inch, Mid-2014

MATLAB 2015a

Psychtoolbox 3.0.13



3

Hi Mario,


Sorry about not including the OS before - the machine is running on El Capitan. This crossed my mind as a concern given the documentation online. Do you think this is the most likely source of the issue?


Thanks,

Dan

4
XX---In PSYCHTOOLBOX@yahoogroups.com, <p7ewlfy4wlyen6uo4vftuo7v77oclidk2n3zgbvj@...> wrote :

Hi Mario,


Sorry about not including the OS before - the machine is running on El Capitan. This crossed my mind as a concern given the documentation online. Do you think this is the most likely source of the issue?


-> Yes. OS bugs are almost the only potential source of such issues, so an OS upgrade is the most likely cause for something going from working to non-working. That's why full specs - exact hardware config, exact OSX version may at least help others to avoid that problem. In general El Crapitan is no worse than Yosemite though -- also as far as my experience goes not better at all in quality, so it might be a OS bug specific to your hardware, introduced by the latest os update you installed.


-> In theory, but less likely, it could also be some freak interaction/incompatibility between your version of Matlab and the version of the OS, so trying a different Matlab version, or trying with Octave-4, could change something. I just know that there was no change to PTB's display pipeline recently which could explain your trouble.


-mario


Thanks,

Dan