SIGN-IN

Cluster copper.sharcnet.ca

Links System documentation in the SHARCNET Help Wiki

Manufacturer Dell
Operating System CentOS 6
Interconnect FDR Infiniband
Total processors/cores 716
Nodes
1‑8
16 cores
2 sockets x 8 cores per socket
Intel Xeon E5-2630 v3 @ 2.4 GHz
Type: Compute
Notes: Each node has four NVIDIA Tesla K80 GPU cards. Each card has two Kepler GK210 GPUs. Run time limited to four (4) hours for non-contribution users.
Memory: 96.0 GB
Local storage: 2 TB
9‑28
24 cores
2 sockets x 12 cores per socket
Intel Xeon E5-2690 v3 @ 2.6 GHz
Type: Compute
Notes: Run time limited to four (4) hours for non-contribution users.
Memory: 64.0 GB
Local storage: 500 GB
29‑32
24 cores
2 sockets x 12 cores per socket
Intel Xeon E5-2690 v3 @ 2.6 GHz
Type: Compute
Memory: 128.0 GB
Local storage: 500 GB
33
12 cores
2 sockets x 6 cores per socket
Intel Xeon E5-2620 v3 @ 2.4 GHz
Type: Compute
Memory: 768.0 GB
Local storage: 800 GB
Total attached storage 27.7 TB
Suitable use

Parallel applications with GPU acceleration.

Large memory applications.

Software available

GCC, GSL, OCTAVE, INTEL, MPFUN2015, PETSC_SLEPC, PYTHON, SAMTOOLS, OPENCV, EMACS, BIOPERL, FFTW, OPENMPI, NETCDF, BIOSAMTOOLS, BINUTILS, BOOST, CPAN, LAMMPS, NCL, GHC, R, PERL, CUDA, BIOPERLRUN, SYSTEM, NAMD, OPEN64, MPC, PNETCDF, BLAST, GNUPLOT, TEXLIVE, GIT, LLVM, OPENJDK, CMAKE, HDF, PROOT, ACML, GDB, MONO, IPM, AUTODOCKVINA, MPFUN90, NIX, SIESTA, NINJA, QD, RUBY, MKL, LDWRAPPER, SUBVERSION, NCBIC++TOOLKIT, ARPACK-NG, GEANT4, VMD

Recent System Notices

Status Status Notes
Oct 16 2019, 11:48AM
(1 day ago)

Goblin, dusky and copper are currently down after a brief unscheduled power outage. We are working on recovering them.

Jun 10 2019, 04:17PM
(4 months ago)

Electrical work will be happening between 12:00 and 13:00 Wednesday June 12 in the data centre housing dusky, goblin and copper. Service on these clusters is not expected to be affected, but we are posting this notice out of an abundance of caution.

Feb 21 2019, 02:48PM
(8 months ago)

Due to their age, the legacy clusters orca, windeee, goblin, shadowfax, copper, and monk will be decommissioned on March 29, 2019. Some nodes from goblin, shadowfax, and copper will be transferred to a new cluster, the affected contributors will be contacted separately.

After March 29 any data you have in /scratch on these clusters will become inaccessible; anything you want to keep should be copied elsewhere as soon as possible. For orca, data in /home and /project is shared with Graham and will be unaffected. For the other clusters, data in /home and /work will remain accessible for a few months via dtn.sharcnet.ca.

If you have not already done so we strongly recommend that you move all of your computing to the new national systems Graham, Cedar and Niagara. See this URL for help: https://docs.computecanada.ca/wiki/Getting_Started

Feb 05 2019, 12:05PM
(8 months ago)

One of the legacy global filesystems will be migrated to new hardware on Wednesday February 20th. To complete this we must unmount the filesystem from all clusters and prevent jobs from running during the outage.

All legacy clusters will be configured to avoid running any jobs after 3pm on February 19.

We expect all legacy clusters to return to service the following day at 10am.

This outage does not affect Graham or Orca.

Sign-in to get full status history