Micro benchmarking - Revision history

Robert-schade-e757@uni-paderborn.de at 11:10, 19 July 2024

2024-07-19T11:10:48Z

2019-09-04T06:26:06Z

2019-06-06T05:08:06Z

Links and further information

2019-06-05T12:07:34Z

Intel MPI Benchmarks

2019-06-05T12:06:56Z

Intel MPI Benchmarks

2019-06-05T12:05:16Z

IOR Parallel filesystem I/O benchmark

2019-06-05T12:02:09Z

Intel MPI Benchmarks

2019-06-05T11:57:58Z

IOR Parallel filesystem I/O benchmark

2019-06-05T11:55:28Z

DGEMM (Linpack) benchmark

2019-06-05T11:27:40Z

Intel MPI Benchmarks

← Older revision		Revision as of 11:10, 19 July 2024
Line 1:		Line 1:
	[[Category:HPC-Developer]]		[[Category:HPC-Developer]]
		+	[[Category:Benchmarking]]
		+
	Microbenchmarking is about measuring the time or performance of small to very small building blocks of real programs. This can be a common data access pattern, a sequence of operations or even a single instruction.		Microbenchmarking is about measuring the time or performance of small to very small building blocks of real programs. This can be a common data access pattern, a sequence of operations or even a single instruction.

← Older revision		Revision as of 06:26, 4 September 2019
Line 1:		Line 1:
		+	[[Category:HPC-Developer]]
	Microbenchmarking is about measuring the time or performance of small to very small building blocks of real programs. This can be a common data access pattern, a sequence of operations or even a single instruction.		Microbenchmarking is about measuring the time or performance of small to very small building blocks of real programs. This can be a common data access pattern, a sequence of operations or even a single instruction.

← Older revision		Revision as of 05:08, 6 June 2019
Line 51:		Line 51:

	== Links and further information ==		== Links and further information ==
		+
		+	* [https://moodle.rrze.uni-erlangen.de/pluginfile.php/12918/mod_resource/content/4/03_Microbenchmarking.pdf Slide set] on Microbenchmarking as part of the RRZE Node-level tutorial.

@@ Line 40: / Line 40: @@
 The [https://software.intel.com/en-us/articles/intel-mpi-benchmarks Intel MPI Benchmarks] perform performance measurements for point-to-point and global communication operations for a range of message sizes. The generated benchmark data characterizes the performance of a cluster system, including node performance, network latency, and throughput efficiency of the MPI implementation used.
-Other MPI micro benchmarks worth looking at are the [https://www.nersc.gov/users/computational-systems/cori/nersc-8-procurement/trinity-nersc-8-rfp/nersc-8-trinity-benchmarks/omb-mpi-tests/ OSU Micro-Benchmarks] and the [http://www.cs.sandia.gov/smb/index.html Sandia MPI Micro-Benchmark Suite (SMB)].
+Other MPI micro benchmarks worth looking at are the [http://mvapich.cse.ohio-state.edu/benchmarks/ OSU Micro-Benchmarks] and the [http://www.cs.sandia.gov/smb/index.html Sandia MPI Micro-Benchmark Suite (SMB)].
 === DGEMM (Linpack) benchmark ===

@@ Line 40: / Line 40: @@
 The [https://software.intel.com/en-us/articles/intel-mpi-benchmarks Intel MPI Benchmarks] perform performance measurements for point-to-point and global communication operations for a range of message sizes. The generated benchmark data characterizes the performance of a cluster system, including node performance, network latency, and throughput efficiency of the MPI implementation used.
-Other MPI micro benchmarks worth looking at are the [https://www.nersc.gov/users/computational-systems/cori/nersc-8-procurement/trinity-nersc-8-rfp/nersc-8-trinity-benchmarks/omb-mpi-tests/ Ohio Microbenchmark MPI tests] and the [http://www.cs.sandia.gov/smb/index.html Sandia MPI Micro-Benchmark Suite (SMB)].
+Other MPI micro benchmarks worth looking at are the [https://www.nersc.gov/users/computational-systems/cori/nersc-8-procurement/trinity-nersc-8-rfp/nersc-8-trinity-benchmarks/omb-mpi-tests/ OSU Micro-Benchmarks] and the [http://www.cs.sandia.gov/smb/index.html Sandia MPI Micro-Benchmark Suite (SMB)].
 === DGEMM (Linpack) benchmark ===

@@ Line 48: / Line 48: @@
 === IOR Parallel filesystem I/O benchmark ===
-The [https://www.nersc.gov/users/computational-systems/cori/nersc-8-procurement/trinity-nersc-8-rfp/nersc-8-trinity-benchmarks/ior/ IOR] IO benchmark measure parallel file system I/O performance at both the POSIX and MPI-IO level. It performs writes and reads to/from files under several sets of conditions and reports the resulting throughput rates.
+The [https://www.nersc.gov/users/computational-systems/cori/nersc-8-procurement/trinity-nersc-8-rfp/nersc-8-trinity-benchmarks/ior/ IOR] IO benchmark measure parallel file system I/O performance at both the POSIX and MPI-IO level. It performs writes and reads to/from files under several sets of conditions and reports the resulting throughput rates. [http://wiki.lustre.org/MDTest mdtest] is an additional tool to evaluate the metadata performance of a file system and has been designed to test parallel file system.
 == Links and further information ==

← Older revision		Revision as of 12:02, 5 June 2019
Line 39:		Line 39:

	The [https://software.intel.com/en-us/articles/intel-mpi-benchmarks Intel MPI Benchmarks] perform performance measurements for point-to-point and global communication operations for a range of message sizes. The generated benchmark data characterizes the performance of a cluster system, including node performance, network latency, and throughput efficiency of the MPI implementation used.		The [https://software.intel.com/en-us/articles/intel-mpi-benchmarks Intel MPI Benchmarks] perform performance measurements for point-to-point and global communication operations for a range of message sizes. The generated benchmark data characterizes the performance of a cluster system, including node performance, network latency, and throughput efficiency of the MPI implementation used.
		+
		+	Other MPI micro benchmarks worth looking at are the [https://www.nersc.gov/users/computational-systems/cori/nersc-8-procurement/trinity-nersc-8-rfp/nersc-8-trinity-benchmarks/omb-mpi-tests/ Ohio Microbenchmark MPI tests] and the [http://www.cs.sandia.gov/smb/index.html Sandia MPI Micro-Benchmark Suite (SMB)].

	=== DGEMM (Linpack) benchmark ===		=== DGEMM (Linpack) benchmark ===

← Older revision		Revision as of 11:57, 5 June 2019
Line 45:		Line 45:

	=== IOR Parallel filesystem I/O benchmark ===		=== IOR Parallel filesystem I/O benchmark ===
		+
		+	The [https://www.nersc.gov/users/computational-systems/cori/nersc-8-procurement/trinity-nersc-8-rfp/nersc-8-trinity-benchmarks/ior/ IOR] IO benchmark measure parallel file system I/O performance at both the POSIX and MPI-IO level. It performs writes and reads to/from files under several sets of conditions and reports the resulting throughput rates.

	== Links and further information ==		== Links and further information ==

← Older revision		Revision as of 11:27, 5 June 2019
Line 38:		Line 38:
	=== Intel MPI Benchmarks ===		=== Intel MPI Benchmarks ===

		+	The [https://software.intel.com/en-us/articles/intel-mpi-benchmarks Intel MPI Benchmarks] perform performance measurements for point-to-point and global communication operations for a range of message sizes. The generated benchmark data characterizes the performance of a cluster system, including node performance, network latency, and throughput efficiency of the MPI implementation used.

	=== DGEMM (Linpack) benchmark ===		=== DGEMM (Linpack) benchmark ===