Application development (Hexagon)

2012-05-16T17:08:45Z

← Older revision		Revision as of 17:08, 16 May 2012
Line 31:		Line 31:
	Please avoid using version numbers unless strictly necessary since older versions of packages may be removed at a later time.		Please avoid using version numbers unless strictly necessary since older versions of packages may be removed at a later time.

-	If you want to change from the ~~PGI~~ compiler (default) to the Intel compiler you type:	+	If you want to change from the Cray compiler (default) to the Intel compiler you type:
-	module swap PrgEnv-~~pgi~~ PrgEnv-intel	+	module swap PrgEnv-cray PrgEnv-intel
-	You should also use swap if you want to load a different version of the same module, this will e.g. replace your current pgi version with 8.0.6:	+	You should also use swap if you want to load a different version of the same module, this will e.g. replace your current pgi version with 12.2.0:
-	module swap pgi pgi/8.0.6	+	module swap pgi pgi/12.2.0

	A complete list of subcommands can be found in the module man page or [http://modules.sourceforge.net/man/module1.html here].		A complete list of subcommands can be found in the module man page or [http://modules.sourceforge.net/man/module1.html here].
Line 43:		Line 43:
	== Compilers and programming languages ==		== Compilers and programming languages ==
	Four different compilers are available on Hexagon:		Four different compilers are available on Hexagon:
-	* ~~PGI~~ (default)	+	* Cray (default)
		+	* PGI
	* GNU		* GNU
	* Intel		* Intel
-	* Pathscale
	All compilation for compute nodes must be done using compiler wrappers.		All compilation for compute nodes must be done using compiler wrappers.
	To switch between compilers module command must be used:		To switch between compilers module command must be used:
	module switch PrgEnv-pgi PrgEnv-gnu		module switch PrgEnv-pgi PrgEnv-gnu
	By default the latest available version will be loaded. You can switch to another compiler version with e.g.:		By default the latest available version will be loaded. You can switch to another compiler version with e.g.:
-	module switch pgi pgi/8.0.6	+	module switch pgi pgi/12.2.0

	=== How to invoke the compiler ===		=== How to invoke the compiler ===
-	Compiling an application for use on the compute node should be done by the wrappers specified below. Running the command "module list" will give you one entry like "PrgEnv-###", where ### is either ~~pgi~~, ~~pathscale~~, gnu or intel.	+	Compiling an application for use on the compute node should be done by the wrappers specified below. Running the command "module list" will give you one entry like "PrgEnv-###", where ### is either cray, pgi, gnu or intel.

	====Compiling programs for compute nodes ====		====Compiling programs for compute nodes ====
Line 87:		Line 87:
	{\|		{\|
	! Compiler \|\| Module		! Compiler \|\| Module
		+	\|-
		+	\| Cray \|\| PrgEnv-cray
	\|-		\|-
	\| PGI \|\| PrgEnv-pgi		\| PGI \|\| PrgEnv-pgi
-	\|-
-	~~\| PathScale \|\| PrgEnv-pathscale~~
	\|-		\|-
	\| GNU \|\| PrgEnv-gnu		\| GNU \|\| PrgEnv-gnu
Line 104:		Line 104:
	{\| border="1" cellspacing="0"		{\| border="1" cellspacing="0"
	\| -mp=nonuma \|\| for the PGI compiler		\| -mp=nonuma \|\| for the PGI compiler
-	\|-
-	~~\| -mp \|\| for the Pathscale compiler~~
	\|}		\|}

Line 111:		Line 109:
	{\| border="1" cellspacing="0"		{\| border="1" cellspacing="0"
	\| -mp \|\| for the PGI compiler		\| -mp \|\| for the PGI compiler
-	\|-
-	~~\| -mp \|\| for the Pathscale compiler~~
	\|}		\|}

Line 120:		Line 116:
	In some cases you may need to use "--enable-static" during configure for running on compute nodes.		In some cases you may need to use "--enable-static" during configure for running on compute nodes.

-	'''Usefull optimization flags for ~~quadcore~~'''	+	'''Usefull optimization flags for the AMD "Interlagos"'''

-	When using PGI the "-tp ~~barcelona~~-64" flag will improve the performance of your code~~. For Pathscale the flag for optimizing for quadcore is "-march=barcelona"~~. These options are automatically provided by the module ''xtpe-~~barcelona~~''.	+	When using PGI the "-tp bulldozer-64" flag will improve the performance of your code. These options are automatically provided by the module ''xtpe-interlagos''. '''NOTE: To compile code that should run on the login nodes this module should NOT be loaded.'''

	=== Recommended environment variable settings ===		=== Recommended environment variable settings ===
-	We recommend you to have the module ''xtpe-~~barcelona~~'' loaded. It will automatically add recommended optimization flags.	+	We recommend you to have the module ''xtpe-interlagos'' loaded. It will automatically add recommended optimization flags.
		+	'''NOTE: To compile code that should run on the login nodes this module should NOT be loaded.'''

	Additionally, the "xt-libsci" module contains optimized versions of common scientific/math libraries (e.g. LAPACK, BLAS).		Additionally, the "xt-libsci" module contains optimized versions of common scientific/math libraries (e.g. LAPACK, BLAS).
Line 296:		Line 293:
	{\|		{\|
	\| -mp=nonuma \|\| for the PGI compiler		\| -mp=nonuma \|\| for the PGI compiler
-	\|-
-	~~\| -mp \|\| for the Pathscale compiler~~
	\|}		\|}

Line 303:		Line 298:
	{\|		{\|
	\| -mp \|\| for the PGI compiler		\| -mp \|\| for the PGI compiler
-	\|-
-	~~\| -mp \|\| for the Pathscale compiler~~
	\|}		\|}

-	In the batch-script set (replace "threads_per_node" with 1-4)	+	In the batch-script set (replace "threads_per_node" with 1-31)
	#PBS -l mppnppn=1,mppwidth=1,mppdepth=threads_per_node		#PBS -l mppnppn=1,mppwidth=1,mppdepth=threads_per_node
	export OMP_NUM_THREADS=threads_per_node		export OMP_NUM_THREADS=threads_per_node
Line 321:		Line 314:
	{\|		{\|
	\| -mp=nonuma \|\| for the PGI compiler		\| -mp=nonuma \|\| for the PGI compiler
-	\|-
-	~~\| -mp \|\| for the Pathscale compiler~~
	\|}		\|}

Line 328:		Line 319:
	{\|		{\|
	\| -mp \|\| for the PGI compiler		\| -mp \|\| for the PGI compiler
-	\|-
-	~~\| -mp \|\| for the Pathscale compiler~~
	\|}		\|}

Line 340:		Line 329:
	aprun ... -n number_of_nodes -d threads_per_node ...		aprun ... -n number_of_nodes -d threads_per_node ...

-	'''Note: the mppnppn and mppdepth values must be chosen such that mppnppn x mppdepth <= 4.'''	+	'''Note: the mppnppn and mppdepth values must be chosen such that mppnppn x mppdepth <= 32.'''

	== Checkpoint and restart of applications ==		== Checkpoint and restart of applications ==

-	To use the checkpointing feature the application must be compiled with blcr ~~and Cray MPT version 3.0.1 and up~~:	+	To use the checkpointing feature the application must be compiled with blcr:
	module load blcr		module load blcr
	With the module loaded, all necessary options will be automatically added by the compiler wrapper. Please recompile your application to include the blcr support. Note that only MPI and SHMEM programming models are supported.		With the module loaded, all necessary options will be automatically added by the compiler wrapper. Please recompile your application to include the blcr support. Note that only MPI and SHMEM programming models are supported.

Data (Hexagon)

2012-05-16T16:33:57Z

Transferring data to/from the system:

← Older revision		Revision as of 16:33, 16 May 2012
Line 31:		Line 31:
	Only ssh type of access is open to hexagon. Therefore to upload or download data only scp and sftp can be used. On special request it is possible to bring hard drives to the data center and have the data copied directly by the support staff. Please contact [[Support]] to arrange this.		Only ssh type of access is open to hexagon. Therefore to upload or download data only scp and sftp can be used. On special request it is possible to bring hard drives to the data center and have the data copied directly by the support staff. Please contact [[Support]] to arrange this.

-	To transfer data to and from hexagon use the ~~following~~ address:	+	To transfer data to and from hexagon use the address:
-	hexagon~~-ftp~~.bccs.uib.no	+	hexagon.bccs.uib.no
-	~~This address~~ has ~~nodes with~~ 10Gb network interfaces~~. Please do not use this address for compiling or running applications~~.	+	All login nodes on hexagon now has 10Gb network interfaces.

	=== Basic tools (scp, sftp) ===		=== Basic tools (scp, sftp) ===

Job execution (Hexagon)

2012-05-16T16:25:29Z

Show changes

Hpcdoc - Recent changes [en]

Application development (Hexagon)

Data (Hexagon)

Job execution (Hexagon)