Opening two cuda application using OpenMP threads in parallel causing a drop in performance of an application


I am having 2 applications one for doing ray tracing using RT cores and general cores, The second applications is mainly used to do DFT’s and other complex calculations using general cores. Both of the applications running fine while running separately meeting the expectations. Now consider the 2 applications like a producer consumer problem. for data to be available for consumer I have combined these 2 applications using openMP with one core of CPU assigned to producer and 1 core assigned to consumer. After doing this I am seeing a drop in performance especially in consumer applications. What could be the possible reason for this?