Improving MPI communication latency on euroben kernels - Revision history

Torben Rasmussen (NSC) at 10:52, 23 April 2013

2013-04-23T10:52:22Z

Joel Hedlund (NSC) at 09:46, 13 March 2012

2012-03-13T09:46:37Z

Chandan Basu (NSC) at 13:12, 9 March 2012

2012-03-09T13:12:41Z

Torben Rasmussen (NSC) at 11:01, 9 March 2012

2012-03-09T11:01:01Z

Chandan Basu (NSC) at 14:16, 8 March 2012

2012-03-08T14:16:35Z

Chandan Basu (NSC) at 14:12, 8 March 2012

2012-03-08T14:12:18Z

Chandan Basu (NSC) at 13:48, 8 March 2012

2012-03-08T13:48:02Z

Chandan Basu (NSC) at 15:28, 6 March 2012

2012-03-06T15:28:43Z

Joel Hedlund (NSC) at 13:45, 8 February 2012

2012-02-08T13:45:01Z

Joel Hedlund (NSC) at 13:40, 8 February 2012

2012-02-08T13:40:01Z

@@ Line 3: / Line 3: @@
 |fields=Computational science
 |financing=PRACE
-|active=Yes
+|active=No
 |start date=2010-07-01
 |end date=2012-06-30

@@ Line 1: / Line 1: @@
 {{project info
 |description=Improving the MPI collective performance by network aware communication
 |financing=PRACE
 |active=Yes

@@ Line 18: / Line 18: @@
 * The tests are run on Curie. The standard MPI_Alltoallv calls are repalced by topology aware MPI_Alltoallv_tuned calls. [16/01/2012]
-* Contributed to deliverable D7.5 [28/02/2012] [[media:7.5.D_Guidelines_for_D7.5.pdf|7.5.D_Guidelines_for_D7.5.pdf]]
+* Contributed to deliverable D7.5: [[media:7.5.D_Guidelines_for_D7.5.pdf|report]] [28/02/2012]
 * Writing final report

@@ Line 18: / Line 18: @@
 * The tests are run on Curie. The standard MPI_Alltoallv calls are repalced by topology aware MPI_Alltoallv_tuned calls. [16/01/2012]
-* Contributed to deliverable D7.5 [28/02/2012] [[File:7.5.D_Guidelines_for_D7.5.pdf]]
+* Contributed to deliverable D7.5 [28/02/2012] [[media:7.5.D_Guidelines_for_D7.5.pdf|7.5.D_Guidelines_for_D7.5.pdf]]
 * Writing final report

@@ Line 18: / Line 18: @@
 * The tests are run on Curie. The standard MPI_Alltoallv calls are repalced by topology aware MPI_Alltoallv_tuned calls. [16/01/2012]
-* Contributed to deliverable D7.5 [28/02/2012]  {{pdf|7.5.D_Guidelines_for_D7.5.pdf‎|Contributed to deliverable D7.5}}
+* Contributed to deliverable D7.5 [28/02/2012] [[File:7.5.D_Guidelines_for_D7.5.pdf]]
 * Writing final report

@@ Line 17: / Line 17: @@
 * The tests are run on Curie. The standard MPI_Alltoallv calls are repalced by topology aware MPI_Alltoallv_tuned calls. [16/01/2012]
-* Writing final report. [28/02/2012]
 == Members ==

@@ Line 10: / Line 10: @@
 The goal of this project is to improve the performance of some euroben kernels. This project is part of PRACE-1ip, WP7.5. The euroben kernel is a set of synthetic benchmark codes which is developed by PRACE. This kernel is used for benchmarking PRACE systems. Some of these benchmarks, e.g., mod2f uses MPI_Alltoall based collective for doing fast Fourier transform. Generally the MPI_Alltoall scaling on many nodes is not good. The performance of alltoall operation can be improved by introducing network awareness in alltoall operation. This can be done by writing a modified alltoall and then calling that from the application code. We tested our modified alltoall on mod2f benchmark with good results
-status
+Status
 * Started code modification. The basic algorithm will remain same. Will map the communication such that there is more communication within node than outside. [30/03/2011]

← Older revision		Revision as of 15:28, 6 March 2012
Line 9:		Line 9:

	The goal of this project is to improve the performance of some euroben kernels. This project is part of PRACE-1ip, WP7.5. The euroben kernel is a set of synthetic benchmark codes which is developed by PRACE. This kernel is used for benchmarking PRACE systems. Some of these benchmarks, e.g., mod2f uses MPI_Alltoall based collective for doing fast Fourier transform. Generally the MPI_Alltoall scaling on many nodes is not good. The performance of alltoall operation can be improved by introducing network awareness in alltoall operation. This can be done by writing a modified alltoall and then calling that from the application code. We tested our modified alltoall on mod2f benchmark with good results		The goal of this project is to improve the performance of some euroben kernels. This project is part of PRACE-1ip, WP7.5. The euroben kernel is a set of synthetic benchmark codes which is developed by PRACE. This kernel is used for benchmarking PRACE systems. Some of these benchmarks, e.g., mod2f uses MPI_Alltoall based collective for doing fast Fourier transform. Generally the MPI_Alltoall scaling on many nodes is not good. The performance of alltoall operation can be improved by introducing network awareness in alltoall operation. This can be done by writing a modified alltoall and then calling that from the application code. We tested our modified alltoall on mod2f benchmark with good results
		+
		+	status
		+
		+	* Started code modification. The basic algorithm will remain same. Will map the communication such that there is more communication within node than outside. [30/03/2011]
		+
		+	* Some initial results are obtained and some improvements are achieved [25/05/2011]
		+
		+	* The tests are run on Curie. The standard MPI_Alltoallv calls are repalced by topology aware MPI_Alltoallv_tuned calls. [16/01/2012]
		+	* Writing final report. [28/02/2012]

	== Members ==		== Members ==

@@ Line 5: / Line 5: @@
 |start date=2010-07-01
 |end date=2012-06-30
-|border=1px solid black collapsed
+|border=1px solid black
 }}