[Q] Nodes and parallel cpu no.

Whitebear · September 23, 2014, 02:06

Global No. of Nodes=598307, serial run is faster than 6 core MPI parallel CPU. CPU:Intel(R) Zeon(R) CPU E5-2643 v2@3.50GHz, 64.0GB RAM, 64bit Win 7 OS

I am solving coal combustion problem. I used to use maximum core number for parallel above half million nodes number.

Does anyone suggest me how to set proper nodes number per CPU no.(core)?

JuPa · September 23, 2014, 07:11

Read the release notes for your Ansys version. It should be in there. There is such a thing as parallel slowdown.

Opaque · September 23, 2014, 11:03

From Intel website,

Quote:

http://ark.intel.com/products/64587/...-GTs-Intel-QPI

Processor Number E5-2643
# of Cores 4
# of Threads 8

Therefore, you should not run anything beyond 4 partitions on this machine.

Whitebear · October 6, 2014, 22:14

Thank you very much all of you.

evcelica · October 10, 2014, 05:58

The E5-2643v2 is a 6 core processor. This is different from the E5-2643, which is a quad core. Either way, I usually only run 4 Processors on my 6 core as well. But I've never had it slow down when running 6 cores. Are you comparing wall time, or CPU time?

September 23, 2014, 02:06	[Q] Nodes and parallel cpu no.	#1
Whitebear Member Jinwhan Ryuk Join Date: Feb 2013 Location: South Korea Posts: 91 Rep Power: 13	Global No. of Nodes=598307, serial run is faster than 6 core MPI parallel CPU. CPU:Intel(R) Zeon(R) CPU E5-2643 v2@3.50GHz, 64.0GB RAM, 64bit Win 7 OS I am solving coal combustion problem. I used to use maximum core number for parallel above half million nodes number. Does anyone suggest me how to set proper nodes number per CPU no.(core)?

October 6, 2014, 22:14	Thank you very much.	#4
Whitebear Member Jinwhan Ryuk Join Date: Feb 2013 Location: South Korea Posts: 91 Rep Power: 13	Thank you very much all of you.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[ICEM] Assigning different number of nodes to parallel edges	Rohit	ANSYS Meshing & Geometry	3	September 22, 2017 09:16
Something weird encountered when running OpenFOAM in parallel on multiple nodes	xpqiu	OpenFOAM Running, Solving & CFD	2	May 2, 2013 04:59
identify interface boundary nodes in parallel	emdoran	Fluent UDF and Scheme Programming	0	September 29, 2010 23:11
CFX-5.7.1(Linux) Parallel - 4 CPU Machine	James Date	CFX	6	June 14, 2005 18:03
Minimum number of nodes to run CFX in parallel	Rui	CFX	3	April 11, 2005 20:46

September 23, 2014, 07:11		#2
JuPa Senior Member Mr CFD Join Date: Jun 2012 Location: Britain Posts: 361 Rep Power: 14	Read the release notes for your Ansys version. It should be in there. There is such a thing as parallel slowdown.

October 10, 2014, 05:58		#5
evcelica Senior Member Erik Join Date: Feb 2011 Location: Earth (Land portion) Posts: 1,167 Rep Power: 23	The E5-2643v2 is a 6 core processor. This is different from the E5-2643, which is a quad core. Either way, I usually only run 4 Processors on my 6 core as well. But I've never had it slow down when running 6 cores. Are you comparing wall time, or CPU time?