SIGN-IN

Accelerator angel.sharcnet.ca

Links System documentation in the SHARCNET Help Wiki

Manufacturer HP
Operating System CentOS 6.x
Interconnect InfiniBand
Total processors/cores 192
Nodes
angel: 1‑22
8 cores
2 sockets x 4 cores per socket
Intel Xeon @ 3.0 GHz
Type: Compute
Notes: Two Geforce 750Ti GPUs with 2GB memory each.
Memory: 8.0 GB
Local storage: 1000 GB
angel: 23‑24
8 cores
2 sockets x 4 cores per socket
Intel Xeon @ 3.0 GHz
Type: Login
Notes: Login/Admin node.
Memory: 8.0 GB
Local storage: None
Total attached storage 0 Bytes
Suitable use

Parallel computing, rendering.

Notes

The GPU cluster is composed of 11 NVIDIA Tesla S1070 GPU servers, each containing 4 GPUs. The GPUs currently available are 750Ti GPUs, with 640 CUDA cores, 1.085 Ghz, 2 GB memory, and 86 GB/s memory bandwidth. Each of the S1070 servers are connected to 2 HP DL160G5 CPU servers. The DL160G5s contain 2 quad core Intel E5430 Xeon chips, 8GB of FBD PC 5300 RAM and 4x250GB SATA disks. As such, there will be 1 GPU per quad core CPU, and a 1:1 memory ratio between the GPUs and CPUs. All of the DL160G5 servers are connected via 4X DDR Infiniband. There is also a login node and a management node with the same specs, but not attached to GPU servers, with a 10Gb uplink to the SHARCNET WAN.

Software available

MPFUN2015, GCC, UTIL, GSL, GMP, MPFR, R, PETSC_SLEPC, SQ, SAMTOOLS, EMACS, CUDA, OPENJDK, VIM, NETCDF, OPEN64, IPM, MPFUN90, NIX, BIOSAMTOOLS, INTEL, GNU , CDF, DAR, CPAN, PARI/GP, CMAKE, NCL, GHC, SIESTA, PNETCDF, PERL, BIOPERLRUN, MrBAYES, QD, SPRNG, NAMD, SUBVERSION, MPC, OPENMPI, ORCA, RUBY, CHARM++, YT, PGI, MKL, LDWRAPPER, BLAST, GNUPLOT, COREUTILS, TEXLIVE, VALGRIND, MERCURIAL, OPENCL, RLWRAP, HDF, BOOST, ACML, ARPACK-NG, GEANT4, TINKER, VMD, PROOT