Cluster Specs
Hardware
The University of Vermont Enterprise Technology Services (ETS) has complete the third major upgrade to the high performance computing (HPC) cluster. More than doubling the total number of computing cores available to faculty and student users, the upgrade also showcases a continued investment at UVM in "green computing," with next generation IBM hardware chosen for the upgrade bringing up to 30% more energy efficiency than equivalent HPC systems. Multi-year grants from NASA, championed by US Senator Patrick Leahy, give strong support to the VACC.
The IBM Bluemoon cluster has 364 nodes providing roughly 3,000 compute cores with the following configurations:
- 128 dual-processor, six-core (Intel Xeon X5650) IBM iDataPlex, with 24GB of ram each, Myrinet-10G connected.
- 98 dual-processor, quad-core (Opteron 2356) IBM x3455s, with 12GB of ram each, Ethernet-connected only.
- 130 dual-processor, dual-core (Opteron 2220) IBM x3455s, with 6GB of ram each, Ethernet-connected only.
- 1 dual-processor, eight-core (Intel Xeon E7-8837) IBM x3690 shared memory machine with 512GB of ram.
- 1 quad-processor, dual-core (Opteron 8220) shared memory machine with 128GB of ram.
- 4 NVIDIA Tesla M2090 GPGPUs, each with 512 CUDA cores and 6GB of memory.
- 4 user nodes (IBM x3455s, and iDataPlex))
- 4 I/O nodes (IBM x3655s, 10G ethernet connected)
- connected to an IBM DS4800 providing 120 terabytes of raw storage to GPFS.
System Software
- RedHat Enterprise Linux 5
- Oscar (ver 4.2) - Open Source Cluster Application Resource cluster management
- TORQUE (ver 1.2) - (Terascale Open-Source Resource and QUEue Manager)
- Cluster Resources' MOAB scheduler.
Application Software
In addition to the list below, we offer a variety of applications software, community codes, and tools for use on the VACC system. Science areas include chemistry, molecular dynamics and image processing, including various libraries and support packages for building high-performance software applications. An assortment of compilers, debuggers and performance tools are also available.
MPI Libraries
- Lam-7.1.4
- Mpich-ch_p4-gcc-1.2.7
- Mpich-mx-pgi-1.2.7.5
BLAS, LAPACK
- ACML 3.1.0
FFTW
- FFTW 2.1
- FFTW 3.0.1
- FTW 3.1
GSL
- GSL 1.9



