Running a test where I am trying to determine how many video calls our code can run using a M2000 card. Each time the test fails because CuCtxCreate returns a 304 error (Operating System call failed) around the 30th creation. Is there a limit on the number of CUDA contexts that can be created? Using nvidia-smi I see the free memory is around 45% when the crash occurs. Any logs that can help me track down what is causing the 304?