Difference between revisions of "GPUcluster"

Revision as of 22:51, 24 February 2011

All nodes runs CentOS 5.5, with openmpi-1.4.

To properly get a functional MPI environment, it is sufficient to create a file named ~/.mpi-selector with the content:
openmpi-1.4-gcc-x86_64

The current node list with their characteristics
Host name	Motherboard	Processor	#cores	Freq	RAM	#GPU	GPU model	CUDA vers.
agape0	X8DTG-D	X5570	8	2.93 Mhz	12280 MB	1	C1060	3.2
agape1	X8DTG-D	X5570	8	2.93 Mhz	12280 MB	1	C1060	3.2
agape2	X8DTG-D	X5650	12	2.67 Mhz	32768 MB	1	M2050	4.0 preview
agape3	X8DTG-D	E5620	8	2.40 Mhz	12280 MB	1	M2050	3.2
agape4	X8DTG-D	E5620	8	2.40 Mhz	12280 MB	1	M2050	3.2
agape5	X8DTG-D	E5620	8	2.40 Mhz	12280 MB	2	S2050	3.2
agape6	X8DTG-D	E5620	8	2.40 Mhz	12280 MB	2	S2050	3.2

@@ Line 2: / Line 2: @@
 All nodes runs CentOS 5.5, with openmpi-1.4.
-To properly get a functional MPI environment, it is sufficient to create a file named ~/.mpi-selector with the content:
+To properly get a functional MPI environment, it is sufficient to create a file named ~/.mpi-selector with the content:<br>
 openmpi-1.4-gcc-x86_64