Would you like to increase the flexibility, efficiency and effectiveness of your HPC infrastructure, while reducing time-to-market? If so, Advania Data Centers’ (ADC) HPCFLOW HPC as a Service platform has your name written all over it.
We have run OpenFOAM using 3 different configurations of HPCFLOW (1, 2 and 4 nodes) against comparable Azure configurations. Our tests have shown that HPCFLOW is up to 50% faster than Azure’s HBv2, depending on the complexity and size of the project.
ADC HPCFLOW vs Azure HBv2 – ClockTime, lower is better
|1 node||2 nodes||4 nodes|
|Run 1||1036 sec||1566 sec||523 sec||766 sec||252 sec||374 sec|
|Run 2||1035 sec||1568 sec||521 sec||767 sec||252 sec||382 sec|
|Run 3||1035 sec||1565 sec||522 sec||766 sec||253 sec||376 sec|
Linear progression for your scaling needs
Although we’ve taken only a small sample into consideration for this benchmark, the solution scales linearly. In other words, if your project took 10 times longer to complete, the number of seconds would increase tenfold.
What this means, generally speaking, is that if your project takes weeks to complete, you’re looking at wrapping it up a few days earlier.
Whether you are using open source applications like OpenFoam or you rely on commercial applications such as ANSYS, Altair, NUMECA, or any other compute intensive application, put us to test and benchmark your applications using HPCFLOW .
We provide better performance and higher flexibility than public cloud services, helping you reducing your time to market and increasing your efficiency.
Big or small, everyone wins
Since HPCFLOW is based on Intel® Xeon® Platinum 9200 processors, we’re able to provide unparalleled performance and platform compatibility. As a result, customers gain speedier insights with HPC workloads in a range of applications: finance, weather, satellite, manufacturing and AI, among others.
Book your benchmark tests with us to see how your application runs with HPC Flow.
Advania: Test as of 06/04/2020. 4‑nodes, Intel® Xeon® Platinum 9242 CPU @ 2.30GHz Processor, 2 sockets, 48 cores/socket, 1Threads/core, 96 cpus, Total Memory 384 GB (24 slots/16GB/ 2933 MHz), CentOS Linux release 7.6.1810 (3.10.0 – 957.5.1.el7.x86_64) ICC compiler version 2019.4.243, IMPI Version 2019 Update 7 Build 20200312, Openfoam version 7/42M Cell Motorbike , score=254s (lower is better),
Azure: Test as of 06/12/2020. 4‑nodes, AMD EPYC 7V12 64-Core Processor, 2 sockets, 60 cores/socket, 1Threads/core, 120vcpus, 30 Numa nodes, Total Memory 448 GB, CentOS Linux release 7.7.1908 (3.10.0 – 1062.12.1.el7.x86_64) , ICC compiler version 2019.4.243, IMPI Version 2019 Update 7 Build 20200312, Openfoam version 7/42M Cell Motorbike , score=374s (lower is better)