- What does it mean that Big Red II is the first US university-owned, university-funded supercomputer capable of 1 petaflop?
One petaflop is a thousand trillion mathematical operations per second – 1.006 petaflops is the peak theoretical capability of Big Red II. Big Red is the first announcement of a system capable of at least 1 petaflop that is owned by and operated solely for the benefit of a single university. There are other, larger systems to which a US university technically has legal title. These include:
- Kraken (University of Tennessee, Knoxville)
- Blue Waters (University of Illinois at Urbana-Champaign)
- Stampede (University of Texas at Austin)
However, the above systems are funded under grant awards by the National Science Foundation (NSF), and are assets for the national research community. Access to these systems is not controlled by the universities that hold legal title to them, but rather by allocation processes specified and organized under contracts with the NSF.
- Is Big Red II faster than Big Red? How much faster?
Yes. However, there are different ways to define “faster.” The peak theoretical performance of Big Red II will be more than 25 times larger than Big Red. That said, the number of available compute nodes and the capability of each compute node is usually more important to users. Big Red II will feature roughly the same number of compute nodes as Big Red, but each compute node will have many more processing cores. The CPU-only compute nodes will have two processors with a total of 32 cores, compared to Big Red’s two processors with a total of four cores. The GPU-enabled compute nodes will have one processor with 16 cores, plus one NVIDIA GPU.
- How can I take advantage of the GPU-enabled nodes in Big Red II?
There are a couple of ways to exploit the power of the NVIDIA Kepler GPUs.
- First, you can run applications that have been compiled with GPU support (NAMD, GROMACS, and others will be available on Big Red II).
- Second, you can link your applications to libraries that take advantage of the GPU (such as cuBLAS or cuFFT).
- Third, you can install your own applications that take advantage of the GPU. NVIDIA developer tools will be available on Big Red II to compile CUDA codes. We will also support OpenACC.
- Is there anything I can do to prepare for the change?Watch this website – we will provide updates here as more information becomes available. (If you have specific questions, feel free to contact us using the link below.)
- Do I need to rebuild my applications so they will run on Big Red II?
Since Big Red II will use the X86-64 architecture, applications that you have built on Mason or Quarry will also run on Big Red II after a recompilation. We will support the Intel and PGI compilers on Big Red II, just as on Quarry. In addition, the GCC and the Cray Compiler will be available. The software environment on Big Red II will be very similar to other Cray systems. If you have access to other Cray systems through XSEDE or other organizations, you can use those systems as testbeds for Big Red II.
- Can I still access my data on DC or DC-WAN?
Yes, DC and DC-WAN will be available on Big Red II. They will also continue to be available on Quarry and Mason, allowing for easy filesharing between the systems.
- Can I access the Scholarly Data Archive (SDA) from Big Red II?
- Will Big Red II support Microsoft Windows applications?
Yes. The system itself will not run Windows or Windows applications; it will run a flavor of Linux. However, you can use a Windows interface such as Putty or Cygwin.
- Can I compile my own codes on Big Red II?
Yes, we encourage you to install your own applications. We will support the following four compilers on Big Red II: GCC, Intel, PGI, and Cray.
- What software packages will be available on Big Red II?
When Big Red II becomes available to general users, popular software packages will already be installed. As with Big Red and all our other HPC systems, you will be able to install your own applications and ask for help installing software. In addition, you will be able to request software packages used by a large number of researchers.
- Will open source and commercial applications run on Big Red II?
Applications designed to take advantage of the Cray Extreme Scalability Mode (ESM) will run on Big Red II. In addition, most applications that you can recompile or that are installed in a publically accessible location will use ESM.
Applications specifically designed to take advantage of commodity clusters will be able to run in Big Red II’s Cluster Compatibility Mode (CCM), an execution environment that supports major commercial applications like Ansys or Matlab. This environment will support any Linux application that can also run on a standard cluster.
- What is the difference between ESM and CCM?
ESM targets large-scale parallel applications, and offers a low-noise kernel and an optimized MPI library for use with the Gemini interconnect.
CCM targets applications that have not been optimized for ESM, and offers full compatibility with a standard X86 Linux environment. In addition to use with different MPI libraries (such as OpenMPI), CCM supports all services generally available on a Linux cluster.
- Who can get access to Big Red II?
Big Red II will be available for use by all members of the IU research community as well as by students. Further information will be posted.
The UITS High Performance Systems group will support Big Red II. If you have system-specific questions about Big Red II, email High Performance Systems.
If you have questions about compilers, programming, scientific/numerical libraries, or debuggers, email Scientific Applications and Performance Tuning.
Please feel free to ask any other questions about Big Red II or other Research Technologies resources by using this form.