Deepthought2 - Revision history

Moulik: /* Sample Python environment */

2018-07-02T01:35:34Z

Sample Python environment

2017-09-08T13:31:16Z

Using sinteractive (preferred)

← Older revision		Revision as of 13:31, 8 September 2017
Line 70:		Line 70:
	and to request access to a GPU enabled node:		and to request access to a GPU enabled node:
	<code>sinteractive -c 1 -t 30 -g gpu:1</code>		<code>sinteractive -c 1 -t 30 -g gpu:1</code>

			To make sure you have accessed a node with NVIDIA GPUs, you can type to following command to list the GPU configuration:
			<source lang=bash>
			[moulik@compute-b17-4 ~/Slurm]> lspci \| grep -i nvidia
			03:00.0 3D controller: NVIDIA Corporation GK110GL [Tesla K20m] (rev a1)
			83:00.0 3D controller: NVIDIA Corporation GK110GL [Tesla K20m] (rev a1)
			</source>

	==Using salloc==		==Using salloc==

2017-08-31T20:37:02Z

Sample Python environment

← Older revision		Revision as of 20:37, 31 August 2017
Line 487:		Line 487:
	==Sample Python environment==		==Sample Python environment==
	Most modules were built using python/2.7.8 compiled with gcc/4.6.1 compiler. The GNU compiler (gcc/4.6.1) is what one gets by default if no other compiler (gcc, intel, pgi, sunsuite) is loaded.		Most modules were built using python/2.7.8 compiled with gcc/4.6.1 compiler. The GNU compiler (gcc/4.6.1) is what one gets by default if no other compiler (gcc, intel, pgi, sunsuite) is loaded.
	<source lang=~~bash~~>		<source lang=tcsh>
	#!/bin/tcsh		#!/bin/tcsh
	module load openmpi/1.6.5		module load openmpi/1.6.5

2017-08-31T20:36:40Z

Sample Python environment

← Older revision		Revision as of 20:36, 31 August 2017
Line 487:		Line 487:
	==Sample Python environment==		==Sample Python environment==
	Most modules were built using python/2.7.8 compiled with gcc/4.6.1 compiler. The GNU compiler (gcc/4.6.1) is what one gets by default if no other compiler (gcc, intel, pgi, sunsuite) is loaded.		Most modules were built using python/2.7.8 compiled with gcc/4.6.1 compiler. The GNU compiler (gcc/4.6.1) is what one gets by default if no other compiler (gcc, intel, pgi, sunsuite) is loaded.
	<source lang=~~csh~~>		<source lang=bash>
	#!/bin/tcsh		#!/bin/tcsh
	module load openmpi/1.6.5		module load openmpi/1.6.5

2017-08-31T20:36:27Z

Sample Python environment

← Older revision		Revision as of 20:36, 31 August 2017
Line 487:		Line 487:
	==Sample Python environment==		==Sample Python environment==
	Most modules were built using python/2.7.8 compiled with gcc/4.6.1 compiler. The GNU compiler (gcc/4.6.1) is what one gets by default if no other compiler (gcc, intel, pgi, sunsuite) is loaded.		Most modules were built using python/2.7.8 compiled with gcc/4.6.1 compiler. The GNU compiler (gcc/4.6.1) is what one gets by default if no other compiler (gcc, intel, pgi, sunsuite) is loaded.
	<source lang=~~tcsh~~>		<source lang=csh>
	#!/bin/tcsh		#!/bin/tcsh
	module load openmpi/1.6.5		module load openmpi/1.6.5

2017-08-31T20:36:05Z

← Older revision		Revision as of 20:36, 31 August 2017
Line 483:		Line 483:

	ulimit -s unlimited		ulimit -s unlimited
			</source>

			==Sample Python environment==
			Most modules were built using python/2.7.8 compiled with gcc/4.6.1 compiler. The GNU compiler (gcc/4.6.1) is what one gets by default if no other compiler (gcc, intel, pgi, sunsuite) is loaded.
			<source lang=tcsh>
			#!/bin/tcsh
			module load openmpi/1.6.5
			module load python/2.7.8
			module load cuda/7.5.18

			python ./YOURCODE.py
	</source>		</source>

2017-08-30T23:07:25Z

Running Jobs

← Older revision		Revision as of 23:07, 30 August 2017
Line 100:		Line 100:
	#SBATCH -N 2		#SBATCH -N 2
	#SBATCH -t 2:00:00		#SBATCH -t 2:00:00
	~~#SBATCH --share~~
	</source>		</source>
	indicating the number of nodes you want, and the maximum running time. The script is then placed in the queue with <code>sbatch -A <account> script_name</code>		indicating the number of nodes you want, and the maximum running time. The script is then placed in the queue with <code>sbatch -A <account> script_name</code>

2017-08-30T21:53:51Z

Running Jobs

← Older revision		Revision as of 21:53, 30 August 2017
Line 107:		Line 107:
	<source lang=bash>		<source lang=bash>
	#SBATCH --share		#SBATCH --share
			</source>

			Most of the nodes currently have at least 30GB of scratch space, some have as much as 250GB available, and a few have as little as 1GB available. Scratch space is currently mounted as <code>/tmp</code>. Scratch space will be cleared once your job completes. The following example specifies a scratch space requirement of 5GB. Note however that if you do this, the scheduler will set a filesize limit of 5GB. If you then try to create a file larger than that, your job will automatically be killed, so be sure to specify a size large enough for your needs. Note that the disk space size must be given in MB.
			<source lang=bash>
			#SBATCH --tmp=5120
	</source>		</source>

2017-08-30T21:49:43Z

Sample SLURM Scripts

← Older revision		Revision as of 21:49, 30 August 2017
Line 152:		Line 152:
	=Sample SLURM Scripts=		=Sample SLURM Scripts=

	Below are a number of sample scripts that can be used as a template for building your own SLURM submission scripts for use on Deepthought2. If you choose to copy one of these sample scripts, please make sure you understand what each line of the sbatch directives before using it to submit your jobs. Otherwise, you may not get the result you want and may waste valuable computing resources.		The Deepthought HPC clusters use a batch scheduling system called [https://slurm.schedmd.com Slurm] to handle the queuing, scheduling, and execution of jobs. This scheduler is used in many recent HPC clusters throughout the world. Below are a number of sample scripts that can be used as a template for building your own SLURM submission scripts for use on Deepthought2. If you choose to copy one of these sample scripts, please make sure you understand what each line of the sbatch directives before using it to submit your jobs. Otherwise, you may not get the result you want and may waste valuable computing resources.

	==Basic, single-processor job==		==Basic, single-processor job==

2017-08-30T21:47:45Z

Overview

← Older revision		Revision as of 21:47, 30 August 2017
Line 5:		Line 5:

	=Overview=		=Overview=
	The Deepthought2 supercomputer consists of 2 login nodes and >400 compute nodes. After logging into <code>login.deepthought2.umd.edu</code> you'll be assigned to either <code>login-1</code> or <code>login-2</code>. The login nodes are to be used to compile code, move data, launch jobs, etc, and they have all the same software (and more) than the compute nodes. '''''Please be sure''''' not to use these nodes to directly run a multi-core job~~... odds are their IT will yell at us eventually~~. Jobs run on 1 or more of the compute nodes, accessible only from the login nodes. These '''400''' compute nodes consist of:		The Deepthought2 supercomputer consists of 2 login nodes and >400 compute nodes. After logging into <code>login.deepthought2.umd.edu</code> you'll be assigned to either <code>login-1</code> or <code>login-2</code>. The login nodes are to be used to compile code, move data, launch jobs, etc, and they have all the same software (and more) than the compute nodes. '''''Please be sure''''' not to use these nodes to directly run a multi-core job. Jobs run on 1 or more of the compute nodes, accessible only from the login nodes. These '''400''' compute nodes consist of:
	* 20 cores		* 20 cores
	* 128 GB of memory		* 128 GB of memory