Some workloads generate a LOT of output files and sometimes quite suddenly. For codes like OpenFOAM, this is data that you may not need until later when you run a post-processing job. Given the amount of data isn’t always predictable, there …
Read More
In the search for HPC application performance, there’s more than one way to build a network. HPC applications (or “codes” as we usually call them) are often at the mercy of the network underpinning an HPC cluster. If your CPUs aren’t busy, …
Read More
By popular request, we’re looking today at the EFA software stack and environment: how to make sure it’s set up correctly (so you get great performance from your codes), how to tell if it’s not, and how to fix that. Austin Cherian, our …
Read More
(A complete teardown of WRF performance on x86 and AWS Graviton, from memory subsystems, compilers and MPI stacks). Weather simulation is a reliably difficult workload for almost any HPC architecture and is often used as a litmus test by …
Read More
AWS’s compute infrastructure is very much not like a ‘normal’ supercomputer (whatever that is). We don’t start with a blank page every few years and design the next big system. It’s more like a city where we have to build on what’s there …
Read More
DCV was originally created for scientists and engineers who run thing on supercomputers. They needed to visualize massive datasets produced by really large compute workloads crunching petabytes of data. That takes some fast machines and …
Read More
We just published a blog post last week with a deep dive into what makes GROMACS tick. The blog post talked about software stacks and EC2 instances that will deliver the best possible performance for people trying to do some molecular …
Read More
Christian Kniep (our senior developer relations engineer) from HPC Engineering is back to finish the conversation we started about containers in HPC. Christian is leading the cause for containerization in HPC, and helping our engineering …
Read More