White Papers

20 Dell HPC System for Manufacturing—System Architecture and Application Performance
Figure 7 ANSYS Fluent Performance—Explicit BB (4)
Figure 8 through Figure 10 present the same performance data, but plotted relative to the “32-cores (1
node)” result. It makes it easy to see the scaling of the solution—the performance improvement as more
cores are used for the analysis. Problem scalability depends on the cell count and physics being simulated
for the specific benchmark problem. For many of the problems, scaling is nearly linear. The results shown
in Figure 8 are for relatively small problems. Therefore, some of these cases scale well and some do not. In
Figure 10, Ice_2m does not scale as well beyond 128 cores or 4 nodes; this is expected because of the
physics of this problem.
Figure 11 presents relative scaling data for the three larger benchmark problems. For these cases, results
are plotted relative to the first valid data point for each case.
0
50
100
150
200
250
300
64(2) 128(4) 192(6) 256(8)
SolverRating(higherisbetter)
NumberofCores(NumerofNodes)
ANSYSFluentPerformance—ExplicitBB
combustor_71m f1_racecar_140m open_racecar_280m