“Try before you buy”
is an integral part of the core culture at Aspen. We take pride in
maintaining cutting edge systems for your use to compile, execute, and evaluate your code.
Our benchmark clusters are configured to allow you to evaluate multiple
combinations of architectures, compilers, network interconnects, files systems, and memory
configurations. On any given day, our benchmark systems range from pre-release engineering
samples to current mass production chips from AMD, Intel and nVidia, all connected with
Infiniband, Myrinet, and Gigabit Ethernet interconnects.
Click the icon to expand a
section, and the icon beside a link to open the reference
link as a pop up.
There are three simple steps to benchmarking at Aspen.
After your account is approved, you will receive more information about how to log in to the benchmark
cluster you need. The account name you select for your Aspen Support Account and the
password you enter will be used for SSH and ABC web access to your selected benchmark cluster.
Review the available nodes below to select your benchmarking nodes.
Our benchmark clusters are constantly in flux. Nodes are often removed for on-site customer evaluation tests,
re-configured for dedicated tests in specific configurations, or upgraded to newer technology. Entire clusters
come and go based on our customers needs. Take a look at the available benchmarking systems shown below
to see what might interest you today. Talk to your sales engineer if you need a specific architecture or
configuration that isn't represented.
Work with your assigned Aspen engineer to benchmark your code.
After your account has been approved, an Aspen engineer will use the ASA information you provide
to contact you. Your production engineer may initiate a
Shared Session
to familiarize you with the benchmark cluster environment and help you through the process of
selecting your environment and desired test configuration(s). Your Aspen engineer can also help
you build, install, or configure your application.
Much like other large multi-user systems, you will be ultimately
responsible for scheduling your jobs and retrieving your results. Unlike many other
large systems, your Aspen production engineer will be standing by to help you should
you encounter difficulties.
Note: If you are benchmarking a commercial application that you currently own,
Aspen may need to utilize your license, or acquire a demonstration or evaluation license
for your benchmark runs on our clusters. We may need to access your current license
and support information to download and install the latest version of your application,
or conference with you and your vendor contact to allow your application runs on our
benchmark clusters. Your sales engineer will help you through this process.
Available Benchmarking Nodes - Configuration
If for some reason we do not have a benchmark cluster configuration that
meets your needs or desires, give us a call or email us and we will see if we have the parts
in stock to build the systems necessary for your specific benchmark.
Our culture does not stop at letting you log in to one of our many benchmark
clusters. We will also help you in getting your code compiled with GCC or any of the high
performance compilers from Intel, PathScale, nVidia (CUDA), PGI, or IBM. We also maintain a
large repository of MPI Libraries, Schedulers, and other tools.
We can help you optimize your code through both run-time
and compile-time options to achieve the best possible results. Our Engineers are
standing by to help through your benchmarking process today!
icon to expand a
section, and the 
icon beside a link to open the reference
link as a pop up.
