Grey screen flickers during ITI

Programming Help
1

Dear friends,

I am running PTB for some drifting gratings. The gratings were generated via procedural texture mapping (as seen in DriftDemo4). During the ITI, the screen is flipped to grey (fillRect). However, we noticed during the ITI, at around 5-6 ms time-mark, there was a strange flickering of the screen. You can see an example in the attached GIF. Flicker occurred at around second 7 of the movie.

In my experiment program, during the ITI, there was nothing else being executed as the while loop is locked to a waiting step until the ITI is fulfilled.

Do you have any clue what might be the cause of this? I am using Matlab 2019b and below is the start-up report from the PTB.

Thanks all!

PTB-INFO: OpenGL-Renderer is NVIDIA Corporation :: GeForce GTX 960/PCIe/SSE2 :: 4.6.0 NVIDIA 388.13
PTB-INFO: VBL startline = 1080 , VBL Endline = 1079
PTB-INFO: Measured monitor refresh interval from beamposition = 16.667734 ms [59.996158 Hz].
PTB-INFO: Beamposition queries are supported, but disabled. Using basic timestamping as fallback:
PTB-INFO: Timestamps returned by Screen(‘Flip’) will be therefore less robust and accurate.
PTB-INFO: Measured monitor refresh interval from VBLsync = 16.659346 ms [60.026366 Hz]. (50 valid samples taken, stddev=0.079672 ms.)
PTB-INFO: Reported monitor refresh interval from operating system = 16.666667 ms [60.000000 Hz].
PTB-INFO: Small deviations between reported values are normal and no reason to worry.
PTB-WARNING: ==============================================================================================================
PTB-WARNING: WINDOWS SYSTEM IS RUNNING IN MULTI-DISPLAY MODE! ALL FLIP STIMULUS ONSET TIMESTAMPS WILL BE LIKELY UNRELIABLE!
PTB-WARNING: STIMULUS ONSET TIMING WILL BE UNRELIABLE AS WELL, AND GRAPHICS PERFORMANCE MAY BE SEVERELY REDUCED!
PTB-WARNING: DO NOT USE MULTI-DISPLAY MODE FOR RUNNING REAL EXPERIMENT SESSIONS WITH ANY REQUIREMENTS FOR ACCURATE TIMING!
PTB-WARNING: ==============================================================================================================

2

Looks like a graphics/display driver bug, probably in gpu dynamic power management. My hunch would be some gpu VRAM memory reclocking gone wrong/happening in the middle of a video refresh cycle instead of inside vblank, causing temporary scanout of pixel trash instead of pixels. On your multi-display setup reclocking shouldn’t happen at all unless all monitors are the same brand and model, with identical settings.

I’d try updating the NVidia display drivers. Or if that doesn’t help, setting gpu power management to some fixed setting instead of automatic switching.