Interactive Environments & Data Exploration

Why Interactive Environments?

  • Debugging
  • Visualization
  • Scripts/command generator
  • Submit job directly to cluster without writing SLURM scripts
  • Dedicated node for web services
  • And more ..

Interactive sessions with BioHPC

Reserve Session with command

  • srun --partition=super --pty --time=00:10:00 /bin/bash

srun              create a resource allocation       

--partition     select interactive partition

--pty             run task zero in pseudo terminal mode

--time           setup time limit

/bin/bash    gives you a pseudo terminal that runs bash


  • vncpasswd/remoteGPU/remoteGUI

vncpasswd           setup your primary VNC password and view-only (optional) password, password has to be longer than 5 characters, password file saved at ~/.vnc/passwd

remoteGUI            reserve a CPU node from super partition for 20 hours

remoteGPU           reserve a GPU node from GPU partion for 20 hours


How to access with TurboVNC client?

VNC server:<port number>

<port number> will be 3+node number+9

For example, if you reserve a vnc session on Nucleus082, the port number will be 30829 and in the VNC server box you may enter:


Cloud Services

WebGUI            312 nodes available from partition super & 32GB, Minimum 32GB RAM
WebGUI256      126 nodes available from partition 256GB & 256GBv1, Minimum   256GB RAM
WebGPU            20 nodes available from partition GPU, Minimum 1 GPU card, Tesla   K20/K40 or 2 Tesla P100 cards
WebGPUv1        12 nodes available from partition GPUv1, 2 Tesla P100 GPU cards
WebWinDCV      20 nodes available from partition GPU, Virtual windows environment   (32 cores, 192GB RAM) running on a GPU node

What if you’ve been assigned to a 32GB partition?

How do I know?

$ sinfo -p 32GB


32GB         up   infinite      1  drain NucleusA013

32GB         up   infinite      9  alloc NucleusA[003,006-007,014,077-081]

32GB         up   infinite     70   idle NucleusA[002,004-005,008-012,015-076]


How to reduce memory reservation for applications (e.g: ImageJ)?

$ ImageJ --mem 16g


           A Dedicate Compute Node + No time limt + reserved for certain group/lab

          This service allows certain BioHPC groups to launch long-running interactive sessions on a dedicated GPU visualization server. There is no time limit for these sessions, but the server is shared among multiple users in the group.

BioHPC On-Demand