White Papers

Deep Learning Performance: Scale-up vs Scale-out
Architectures & Technologies Dell EMC | Infrastructure Solutions Group
11
Figure 6: Frameworks Comparison
4 Test Methodology
The test methodology consists of 3 distinct phases.
Phase 1 is where we test the hardware performance of each server using NVidia supplied
p2pbandwidth and latency tests and Baidu Deep Bench. This is explained in section Phase 1.
Phase 2 we used TensorFlow framework and ran some of the well-known neural models to
compare performance in terms of throughput & training time. This is explained in section Phase
2.
Phase 3 we used TensorFlow but compared performance at multi-node level using Uber’s
Horovod implementation.