I am using a i5 multi core machine w. the latest Ubuntu install, and am testing w. the current graphics card (AMD Radeon Pro WX3200) and a ASUS monitor capable of 144Hz (I’m trying to run it at 120Hz). I am testing it w. the VBLSync call from PTB, and it misses 6xx/1000 flip calls, and reports that it cannot get a beam position response. Calling PsychLinuxConfiguration does not fix this.
I switched from Win10 because, even though VBLSync reported that missed flips were 3 or so/1000, it was clearly tearing when I ran an experiment, and the SSVEP data were garbage (but single-event ERP data are fine). BTW - running an SSVEP experiment in Linux at full speed produces obvious tearing; either a software issue is present or the card cannot deal w. the demand; this is the case even though the fastest that the experiment calls for a change is every 4 refreshes (33 ms). Timing tests strongly suggest that the (fairly minimal) rendering demands from the script are not likely to be the problem. The presence of tearing is not a surprise; the VBLSynchTest figures show many missed deadlines, some by more than 15 ms.
Using Matlab 2021. I think I installed the drivers correctly (but maybe not - is there a good FAQ for this on the forum?). My question; is this a card that should work? Or do I need to swap it out for a newer/ more powerful card (like the AMD Pro RX 6600?)
Running a 144 Hz panel at 120 Hz could trigger temporal resampling in the panel and cause “tearing-like” effects, depending on specific panel type and capabilities. Baseline test is always native resolution and refresh rate.
Sounds like configuration or software problem.
No need to install any drivers for AMD on Linux, the default drivers should be perfect. If you did install proprietary amdgpu-pro drivers from AMD against clear recommendations on our website, that could be related to your problems.
Or some Matlab interference triggering software rendering? - our installation instructions do have info about proper setup of Matlab to work around problems caused by Matlab.
I’d assume most likely some misconfiguration, not a problem with your card. help PsychPaidSupportAndServices will tell you how to buy further advice from myself if above pointers are insufficient.
OK - the test at 144Hz and at 120Hz is now as expected after removing the amdgpu drivers. I will still need some help w. the screen configuration, but I’ll pursue that w. paid support so I don’t waste time on it. Thanks