The following test results include the performance of the platform during the deletion and re-creation of the instant clone VMs after all users logged out when the test run was completed. The different phases of the test cycle are displayed in the test results graphs in Appendix A.
The following tables summarize the test results for the various workloads and configurations. For graphs and analysis of the detailed test results for the dual workload AI and VDI test cases, see Appendix A.
The user densities we tested with for the "VDI Medical Knowledge Worker GPU enhanced" workload were as follows:
- Single workload—72 users
- Dual workload—24 users
The following table summarizes the test results for single workload test cases:
Test case | AI duration | GPU A | GPU B | GPU C | Average host CPU utilization | Average GPU card utilization | Average vSAN read latency | Average vSAN write latency | LVSI UI metrics | Average AUC |
AI Training GPU enhanced | 1 hr 5 min | AI | AI | AI | 28.6% | GPU A 81% GPU B 81% GPU C 81% | 0.3 ms | 1.8 ms | N/A | N/A |
AI Validation GPU enhanced | 38 min | AI | N/A | N/A | 13% | GPU A 30% GPU B N/A GPU C N/A | 0.4 ms | 1.3 ms | N/A | 0.80 |
AI Training CPU only | 47 hr 24 min | N/A | N/A | N/A | 42% | N/A | 0.2 ms | 0.6 ms | N/A | N/A |
AI Validation CPU only | 1 hr 35 min | N/A | N/A | N/A | 30% | N/A | 0.2 ms | 1.0 ms | N/A | 0.72 |
VDI Medical Knowledge Worker GPU enhanced | N/A | VDI | VDI | VDI | 84% | GPU A 25% GPU B 25% GPU C 25% | 0.1 ms | 1.02 ms | Base: 806 Avg max: 1226 Threshold: 1806 | N/A |
VDI Medical Knowledge Worker CPU only | N/A | N/A | N/A | N/A | 73% | N/A | 0.2 ms | 1.0 ms | Base: 792 Avg max: 1216 Threshold: 1792 | N/A |
The following table summarizes the test results for dual workload test cases:
Test case | AI duration | GPU A | GPU B | GPU C | Average host CPU utilization | Average GPU card utilization | Average vSAN read latency | Average vSAN write latency | LVSI UI metrics | Average AUC |
Dual Workload AI Training and VDI Medical Knowledge Worker, both GPU enhanced | 4 hr 43 min | VDI | AI | AI | 38% | GPU A 28% GPU B 28% GPU C 29% | 0.2 ms | 0.8 ms | Base: 795 Avg max: 821 Threshold: 1795 | N/A |
Dual Workload AI Validation and VDI Medical Knowledge Worker, both GPU enhanced | 48 min 30 sec | VDI | AI | N/A | 38% | GPU A 27% GPU B 13% GPU C N/A | 0.1 ms | 0.77 ms | Base: 791 Avg max: 796 Threshold: 1791 | 0.75 |
Dual Workload AI Training CPU only and VDI Medical Knowledge Worker GPU enhanced | 50 hr 30 min | N/A | N/A | VDI | 71% | GPU A N/A GPU B N/A GPU C 27% | 0.25 ms | 1.30 ms | Base: 830 Avg max: 897 Threshold: 1831 | N/A |
Dual Workload AI Validation CPU only and VDI Medical Knowledge Worker GPU enhanced | 1 hr 44 min | N/A | N/A | VDI | 53% | GPU A N/A GPU B N/A GPU C 27% | 0.2 ms | 1.0 ms | Base: 920 Avg max: 894 Threshold: 1920 | 0.69 |