White Papers

Deep Learning Performance: Scale-up vs Scale-out
Architectures & Technologies Dell EMC | Infrastructure Solutions Group
40
7.3 Results Showing Training Time Vs Accuracy
These tests were run with 90 epochs to determine training time to achieve top-1% and top-5%
accuracy.
Figure 35 shows the results for the server 8X SXM2, POWEREDGE C4140-V100 in single and multi-
node mode, C4130-P100 in single and multi-node mode, and R740-P40.
Results Highlights
➢ The fastest training time was achieved by the system 8X SXM2 with 93% of accuracy
convergence in 6.6 hours.
➢ PowerEdge C4140 – Configuration K with SXM2 in multi-node configuration achieved 95%
of accuracy convergence in 7.2 hours.
➢ R740-P40 with 48.8hrs and 93% of accuracy convergence.
Figure 35: Longest tests to extract accuracy convergence and training time