The total user density achieved for this test is 56 (1B vGPU profile) desktops per host, 5 x NVIDIA T4 cards installed out of 8 PCIe Slots (2 slots are used for Storage Controller and 1 slot is used for Network adapter) and Fixed share scheduler (vGPU scheduling policy) is used for testing.
CPU usage at start
CPU usage at steady state
CPU usage steadily increased during logon phase and converged at the defined threshold during steady state phase of the test. The compute nodes only showed slight discrepancies in usage during the logon phase which appears to be a normal distribution of usage. There were no anomalies seen during testing.