Updated WRF benchmarks (Q1 2022)
One of the great things about working with public CSPs is that they continue bringing the latest hardware options from different manufacturers to the benefit of all of us. We continue evaluating the performance of several HPC apps with the latest hardware so that our customers can make informed decisions. Our updated round of WRF benchmarks include measurements with Intel Ice Lake and AMD EPYC3 (Milan) processors from AWS and Azure.
The figure reflects overall WRF performance on computing and high-performance-computing instances from AWS and Azure.
The above WRF performance has been evaluated from an assemble of several test cases ranging from a single domain (1,500 x 1,500 points) to forecast with several nested domains. The overall performance, which is essentially inversely proportional to wall times, is weighted from these test cases and normalized with a maximum score of 100% for an instance being the best performer for every test case.
The measurements reflect performance on the following instances:
AWS 5th generation
AWS c5n.18xlarge – 36 cores (Intel Xeon 8124 – Skylake), 192 GBytes; price (N. Va. region): on-demand: $3.888/hr; spot: $1.1688/hr
AWS m5zn.12xlarge – 24 cores (Intel Xeon – Cascade Lake), 192 GBytes; price (N. Va. region): on-demand price: $3.9641/hr; spot price: $0.982/hr
AWS 6th generation
AWS c6g.16xlarge – 64 cores (AWS Graviton2 – Neoverse 1), 192 GBytes; price (N. Va. region): on-demand: $2.176/hr; spot: $1.0882/hr
AWS c6i.32xlarge – 64 cores (Intel Ice Lake), 256 GBytes; price (N. Va. region): on-demand: $5.44/hr; spot: $2.1764/hr
AWS m6a.48xlarge – 96 cores (AMD EPYC 7R13), 768 GBytes; price (N. Va. region): on-demand: $8.2944/hr; spot: $3.4279/hr
AWS hpc6a.48xlarge – 96 cores (AMD EPYC 7R13), 768 GBytes; price (Ohio region): on-demand: $*****/hr; spot: $3.4279/hr
Azure
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Ut elit tellus, luctus nec ullamcorper mattis, pulvinar dapibus leo.
HB120rs_v2 (AMD EPYC 7002) – 120 cores – 480 GB; on-demand price= , spot-price=$ /hr
HB120rs_v3 (AMD EPYC 7003) – 120 cores – 480 GB; on-demand price= , spot-price=$ /hr
As the number of cores available for each instance has a great variability, it is also feasible to evaluate WRF performance based on a per core basics as the following figure shows.