Reference Guide

Performance evaluation and analysis
17 Reference Architecture of Dell EMC Ready Solution for HPC Life Sciences | Document 309
Figure 7 Runtime comparisons of steps in variant calling pipeline
Total runtime, sum of all runtime from each step, provides better insight what we can choose for BWA-GATK
pipeline as well as other pipelines including aligning and sorting as sub steps. 6248 is the best choice in terms
of performance. However, it is important to note that the runtimes in Table 2 are drawn from single sample
test (one job on one compute node). The most economical choice could be 6230 due to the relatively low cost
and TDP among the tested 2nd Gen CPUs.
Table 2 Total runtime comparisons among 1
st
and 2
nd
Gen CPUs
CPU
Price
Spec
Total BWA-GATK
runtime (hours)
Skylake
6148
$3072.00 - $3078.00
2.4 GHz, 20 cores, 150 W
24.26
6154
$3,543.00
3.0 GHz, 18 cores, 200 W
23.47
6152
$3655.00 - $3661.00
2.1 GHz, 22 cores, 140 W
24.58
6138
$2612.00 - $2618.00
2.0 GHz, 20 cores, 125 W
24.83
Cascade Lake
6248
$3072.00 - $3078.00
2.5 GHz, 20 cores, 150 W
23.36
6252
$3655.00 - $3662.00
2.1 GHz, 24 cores, 150 W
23.82
6230
$1894.00 - $1900.00
2.1 GHz, 20 cores, 125 W
23.68
3.1.2 Single sample per node, multiple node test
In Figure 8, the runtime in a variety of samples and compute nodes with 50x whole genome sequencing data
are summarized. The tests performed here are designed to demonstrate performance at the server level, not
to compare individual components. The data points in Figure 8 are calculated based on the total number of
samples, one sample per compute node (X axis in the figure) that are processed concurrently. The details of
BWA-GATK pipeline information can be obtained from the Broad Institute web site (16). The maximum
number of compute nodes used for the tests are 64x C6420s. C6420s with Lustre/ME4 show a better scaling
behavior than Lustre/MD3.