You are here

New Compute Canada National Systems Available for Testing Use

Compute Canada’s newest national systems -- Cedar and Graham -- are now available for early-stage testing use by researchers.

Testing Phase

Please be aware that ongoing testing and component installations will continue over the next few weeks. However at this time users should be able to transfer in data, compile code, work on submission scripts, and explore the available modules. 

Please refer to the Known Issues page on Compute Canada’s User Documentation Wiki for further details and status updates.

How To Access

  • You can SSH to the new servers using the following login nodes:
    cedar.computecanada.ca
    graham.computecanada.ca
    ** Note: You will need to use your Compute Canada username and password to login (the same credentials you use for the CCDB). If you have troubles logging in or have forgotten your Compute Canada username or password, contact support@computecanada.ca.

  • You can transfer files to and from the new systems using the Globus endpoints:
    computecanada#cedar
    computecanada#graham

How To Use

The Compute Canada Documentation Wiki will be your go-to reference centre for step-by-step instructions and documentation for:

Getting Help & Reporting Issues

Mini-Webinars - Intros to the Basics

A series of short webinars (5-10 minutes each) are available for quick and basic overviews of the new systems. These are meant to complement the Documentation Wiki pages, which are more detailed.  

  • Getting an account: If you don’t have a Compute Canada account, you’ll need to register for one to access these new national systems.
  • General info: General information about the new national systems.
  • Getting help: Need help? This short video will show you how and where to get it.
  • Software environment: Learn more about the software environment of the national systems.
  • File systems: Discover the ins and outs of the file systems on the new national systems.
  • Managing jobs: Find out how to manage jobs on the new national systems, including basic slurm commands.
  • Common mistakes to avoid: Save yourself time and discover the most common mistakes to avoid.
  • Accessing the environment: This short video will tell you everything you need to know about accessing the new environment - from login to data transfer.

Other National System Training Events

  • Cedar Demonstration (as part of WestGrid July Town Hall)
    Wednesday, July 12
    10:00 - 11:00 am Pacific
    Online Webinar: Click here to register (connection details will be emailed)
    Description:
    Members of WestGrid's support team will provide a walk-through demonstration of using Cedar, including instructions for logging in, exploring the file systems, compiling basic programs, working with modules, running jobs via the SLURM scheduler, running interactive jobs, finding documentation and getting help. 

  • How Jobs are Scheduled to Run on Graham & Cedar
    Wednesday, July 19
    12:00 - 1:00 pm Eastern
    Online Webinar: Click here at time of webinar to connect & view
    Description:
    This webinar describes the configuration of the new national systems' Slurm scheduler and how it determines how jobs are dispatched to resources. Beyond describing the scheduler configuration, this webinar will also provide recommended best practices when submitting jobs and demonstrate tools for monitoring jobs and the job queue on Graham and Cedar.

  • What's New & Exciting About Graham's GPUs
    On June 21, SHARCNET presented a webinar outlining the newest and largest SHARCNET cluster and Compute Canada national system, Graham. Among its exciting features is a substantial number (320) of the newest HPC GPUs from NVIDIA, Pascal P100. (The other new National Compute Canada cluster, Cedar, will have 584 Pascal P100 GPUs.) This will be a very significant improvement for SHARCNET users in terms of GPGPU capabilities, both quantitatively (27x more GPU flops than our older GPU cluster, monk) and qualitatively (new GPU capabilities; three generations newer than the monk GPUs). This webinar will discuss all the new important features of the Graham GPUs, with a live demonstration. Some minimum familiarity with GPGPU programming (CUDA or OpenCL) is desirable.
    Click here to view the webinar recording.

RAC 2017 Allocations

The RAC 2017 allocations were implemented on WestGrid systems and the new Cedar system on June 30, 2017. Users should see their jobs running with their RAC 2017 priorities. WestGrid systems scheduled for defunding were not allocated, and will remain with their RAC 2016 priorities. Current jobs should continue to run during the RAC 2017 updates to the schedulers, however, due to complicated configurations required for the mix of new and old schedulers and the change of RACs some jobs may fail. Please contact us if there are any issues with current jobs terminating or new jobs failing to start.

Migrating to the New Systems

If you are currently using a WestGrid system but would like to move your compute and data storage use to Cedar or Graham, please refer to this General Directives for Migration page on the Documentation Wiki.