It is exciting to observe the rapid evolution of the cloud high-performance computing market and to think of what it can mean for customer innovation. Just a year ago, Microsoft Azure was the first to run a 10,000 core Simcenter STAR-CCM™+ job in the cloud with AMD EPYC processor-based HB-series VMs. This run proved the viability of large-scale cloud HPC while showcasing impressive performance that rivals on-premises HPC clusters. Azure customers shared resoundingly positive feedback about how this newfound scale helped them to accelerate research and be more productive.
This year, Microsoft upped the ante when they published results of the 2nd Gen AMD EPYC processor-powered Azure HBv2 running a 57,000 core “Le Mans” computational fluid dynamics (CFD) model on Simcenter STAR-CCM+ with 10x the mesh resolution of last year’s test. This was among the largest Simcenter STAR-CCM+ jobs ever validated by Siemens*-- not just in the cloud, but in any datacenter environment. Until recently, only the most powerful computing environments were equipped to run simulations of such scale and accuracy. Now, small and large companies alike possess the computing power to drive world-changing innovation at supercomputer-scale on-demand for as little as $0.6614/hr per virtual machine using spot pricing.
Figure 1: Simcenter STAR-CCM+ Speedup on HBv2 from 1 to 640 virtual machines (57,600 cores)
Note: A given scaling point may achieve optimal performance with 90, 112, 116, or 120 parallel processes per VM. ###Plotted data below shows optimal performance figures. All tests were run with HPC-X MPI ver. 2.50.
Why AMD EPYC Processors for Simcenter STAR-CCM+?
As with a wide variety of enterprise and HPC workloads, 2nd Gen AMD EPYC processors offer leading performance for CFD, with up to 95% faster simulations than competitive alternatives on Simcenter STAR-CCM+[1]. AMD works with major software vendors like Siemens to help ensure your software is optimized for performance with AMD EPYC processors. The broad ecosystem of open tools and libraries are more reasons why Microsoft Azure trusts AMD EPYC processors for its most demanding services.
“There is a constant pressure to accelerate the speed of product design and today, our customers are looking to turn around high-fidelity simulations in hours, not weeks. We are very impressed with the scalability and performance of Simcenter STAR-CCM+ on the AMD EPYC processor-based Azure Virtual Machines. It will enable our customers to scale-up their simulations quickly, so that they can get the necessary insight to make better engineering decisions faster.”
Keith Foston, Senior Product Manager - Cloud and SaaS at Siemens
AMD Technology Powers HBv2 Leadership
Azure HBv2 extends leadership AMD EPYC performance to the cloud with a fully optimized virtual machine featuring the latest HPC technologies. This makes HBv2 an ideal extension of an on-premises environment and helps provides customers the capability to deliver net new workloads in the cloud on a scale previously unimaginable for some companies.
Specific features include:
- 2nd Gen AMD EPYC processors with leadership core counts and 45% more memory bandwidth than competing alternatives*
- Cloud’s only 200Gbps HDR InfiniBand* running PCIe Gen 4.0, providing excellent application scaling efficiency and low latencies
- Elastic scale ranging from 120-80,000 cores for MPI workloads
Azure HBv2 VM Specs
|
CPU Cores |
Memory |
Memory per CPU Core |
|
|
120 |
480 GB |
4GB |
|
|
Local SSD: GiB |
RDMA Network |
Azure network |
|
|
1.6TB |
200 Gbps |
|
Learn more about the AMD-powered HBv2 solution for Simcenter STAR-CCM+
Azure and AMD make it easy to set up an AMD EPYC processor based VMs and run the benchmarks you see above. Learn more about manufacturing solutions with HBv2 and 2nd Gen AMD EPYC processors.
Shrinking time-to-results with CFD at extreme scale with STAR-CCM+ on Azure HBv2 VMs
Run STAR-CCM+ In Azure HPC Cluster
AMD EPYC for STAR-CCM+ Solution Brief
[1] Based on AMD internal testing of Siemens PLM STAR-CCM+ 14.02.009, kcs_with_physics benchmark, as of July 17, 2019 using a 2P EPYC 7742 powered reference server versus a 2P Xeon Platinum 8280 powered server. Results may vary. ROM-70
*Sources:
https://azure.microsoft.com/en-us/blog/azure-hbv2-virtual-machines-eclipse-80000-cores-for-mpi-hpc/
https://azure.microsoft.com/en-us/pricing/details/virtual-machines/linux/
Sean Kerr is Product Marketing Manager, Cloud at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites or use of third-party names/marks are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.