This is an unofficial quick start tutorial for NYU's UltraViolet HPC, intended for statistical analysts with no prior experience with remote SSH or Linux.
- Official guide: https://hpcmed.org/guide
- Wiki: http://bigpurple-ws.nyumc.org/wiki/index.php/BigPurple_HPC_Cluster
- MobaXTerm (console access + file transfer + light editing): https://mobaxterm.mobatek.net/
Also consider:
- WinSCP (file transfer): https://www.winscp.net/
- Terminal (console access): Included
Also consider:
- Filezilla (file transfer): https://filezilla-project.org/
- XQuartz (enable graphic user interface for Stata and SAS): https://filezilla-project.org/
OnDemand: https://ondemand.hpc.nyumc.org/
Navigate your directories and submit batch jobs on your web browser. VPN needed.
Consider Sublime Text or VSCode to edit Stata or R scripts.
VPN access is needed when you’re off-campus: https://atnyulmc.org/help-documentation/NYU-Langone-Advanced-Access-App
Read this first: https://hpcmed.org/guide/bigpurple
Once you log in, you'll see:
__ __ __ __ _ __ _ __ __
/ / / / / / / /_ _____ _____ | | / / (_) ____ / / ___ / /_
/ / / / / / / __/ / ___/ / __ | | | / / / / / __ \ / / / _ \ / __/
/ /_/ / / / / /_ / / / /_/ | | |/ / / / / /_/ / / / / __/ / /_
\____/ /_/ \__/ /_/ \____/\_\ |___/ /_/ \____/ /_/ \___/ \__/
NYU Langone Health HPC
Use the following commands to adjust your environment:
module avail - show available modules
module add <module> - adds a module to your environment for this session
module initadd <module> - configure module to be loaded at every login
BigPurple User Guide available at: http://bigpurple-ws.nyumc.org/wiki
New HPC Portal: https://hpcmed.org/
HPC Community town hall is held every Thursday from 12 to 1 PM.
Meeting link: https://nyumc.webex.com/meet/siavoa01
You may email <[email protected]> for any further assistance.
Quarterly maintenance, schaduled for June 9th, 2024 IS POSTPONED.
The new date will be announced well in advance to reduce the
impact to computational researches.
Loading default-environment
Loading requirement: slurm/current
[baes03@bigpurple-ln2 ~]$
You are in the login node. Your command prompt shows ln2.
Use this workflow only for development purposes. Your final results must come from a batch job.
Read this first: https://hpcmed.org/guide/slurm#headings2
Run this command: srun -p cpu_short --mem-per-cpu=4G -t 00-02:00:00 --pty bash
[baes03@bigpurple-ln2 ~]$ srun -p cpu_short --mem-per-cpu=4G -t 00-02:00:00 --pty bash
srun: job 48302015 queued and waiting for resources
srun: job 48302015 has been allocated resources
[baes03@cn-0012 ~]$
Now your command prompt shows cn-0012.
You can request more memory by increasing --mem-per-cpu and longer runtime by increasing -t.
Loading stata: module load stata
Loading R: module load r
List of available modules: https://hpcmed.org/guide/modules
Stata (CLI): stata
Stata (GUI): xstata (You need X11. Use MobaXTerm or XQuartz.)
R (CLI): R
Go to /gpfs/data/massielab/data/srtr. You’ll see many subdirectories named srtrYYMM.
YY and MM are the year and month of the release. SRTR releases datasets quarterly, plus when necessary. We will use the latest release with the standard analysis files (SAFs) in most cases.
You’ll probably start with the TX_KI dataset.
Data dictionary available here: https://www.srtr.org/requesting-srtr-data/saf-data-dictionary/
The latest release available here: /gpfs/data/easelab/data/USRDS/stata
Previous versions (e.g. 2022 release) here. You probably won’t need this: /gpfs/data/easelab/data/USRDS/USRDS2022
USRDS researcher’s guide has data file descriptions, data dictionary, and more: https://www.niddk.nih.gov/about-niddk/strategic-plans-reports/usrds/for-researchers/researchers-guide
Suppose you want to run an R script called demo.r, which is stored at ~/demo_project.
Create a text file called run_demo.sh, with the script below. Make sure you update [YOUR EMAIL]
#!/bin/bash
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --time=10:00:00
#SBATCH --mem=8GB
#SBATCH --job-name=demo
#SBATCH --mail-user=[YOUR EMAIL]
#SBATCH --mail-type=END
#SBATCH --output=/dev/null
#SBATCH --error=/dev/null
module load r
cd ~/demo_project
R CMD BATCH --no-restore --no-save demo.r
Now, go to the directory where you stored run_demo.sh, and run
sbatch run_demo.sh
Suppose you want to run a Stata script called demo.do, which is stored at ~/demo_project.
Create a text file called run_demo.sh, with the script below. Make sure you update [YOUR EMAIL]
#!/bin/bash
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --time=10:00:00
#SBATCH --mem=8GB
#SBATCH --job-name=demo
#SBATCH --mail-user=[YOUR EMAIL]
#SBATCH --mail-type=END
#SBATCH --output=/dev/null
#SBATCH --error=/dev/null
module load stata
cd ~/demo_project
stata-mp -b do demo.do
Now, go to the directory where you stored run_demo.sh, and run
sbatch run_demo.sh
squeue -u $USER
Basic command: scancel [JOBID]
In action:
Check your JOBID first
[baes03@cn-0032 bin]$ squeue -u $USER
JOBID PARTITION NAME USER ST TIME NODES NODELIST(REASON)
48302494 cpu_short bash baes03 R 0:27 1 cn-0032
Then cancel it!
scancel 48302494
- Go to
~/bin. If you don't have one, create one.
cd ~/
mkdir bin
cd ~/bin
- Create two files.
First, create srun_r (no extension). Open the file in any text editor (like the MobaXTerm internal editor). Put the script below into srun_r and save it.
echo '#!/bin/bash
module load stata
stata-mp -b do '$1 | sbatch --output=/dev/null --error=/dev/null --job-name=$1 --time=24:00:00 --ntasks=1 --cpus-per-task=1 --mem-per-cpu=6G --mail-user=[YOUR EMAIL HERE] --mail-type=END,FAIL
Second, create srun_stata (no extension). Put the script below.
echo '#!/bin/bash
module load r
R CMD BATCH --no-restore --no-save '$1 | sbatch --output=/dev/null --error=/dev/null --job-name=$1 --time=24:00:00 --ntasks=1 --cpus-per-task=1 --mem-per-cpu=6G --mail-user=[YOUR EMAIL HERE] --mail-type=END,FAIL
All done! Let's try these.
Go to your project directory, and run your script. Let's say its name is demo.r
cd ~/demo_project
srun_r demo.r
If Stata,
cd ~/demo_project
srun_stata demo.do
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent