mpi-benchmark

Appendix A

Benchmark Programs

Several MPI performance measurement programs are installed from the mpi-benchmarkRPM. This Appendix describes these useful benchmarks and how to run them. These programs are based on code from the group of Dr. Dhabaleswar K. Panda at the Network-Based Computing Laboratory at the Ohio State University. For more information, see:

http://nowlab.cis.ohio-state.edu/

These programs allow you to measure the MPI latency and bandwidth between two or more nodes in your cluster. Both the executables, and the source for those executables, are shipped. The executables are shipped in the

RPM, and installed under /usr/bin. The source is shipped in the mpi-develRPM and installed under

/usr/share/mpich/examples/performance.

The examples given below are intended only to show the syntax for invoking these programs and the meaning of the output. They are NOT representations of actual InfiniPath performance characteristics.

A.1

Benchmark 1: Measuring MPI Latency Between Two Nodes

In the MPI community, latency for a message of given size is defined to be the time difference between a node program’s calling MPI_Send and the time that the corresponding MPI_Recv in the receiving node program returns. By latency, alone without a qualifying message size, we mean the latency for a message of size zero. This latency represents the minimum overhead for sending messages, due both to software overhead and to delays in the electronics of the fabric. To simplify the timing measurement, latencies are usually measured with a ping-pongmethod, timing a round-trip and dividing by two.

The program osu_latency, from Ohio State University, measures the latency for a range of messages sizes from 0 to 4 megabytes. It uses a ping-pong method, in which the rank 0 process initiates a series of sends and the rank 1 process echoes them back, using the blocking MPI send and receive calls for all operations. Half the time interval observed by the rank 0 process for each such exchange is a measure of the latency for messages of that size, as defined above. The program uses a loop, executing many such exchanges for each message size, in order to get an average. It defers the timing until the message has been sent and received a number of times, in order to be sure that all the caches in the pipeline have been filled.

IB6054601-00 D

A-1

Page 64 Page 66
Page 65
Image 65
Q-Logic IB6054601-00 D manual Benchmark Programs, Benchmark 1 Measuring MPI Latency Between Two Nodes
Page 64 Page 66

IB6054601-00 D specifications

The Q-Logic IB6054601-00 D is a high-performance InfiniBand adapter card designed for data centers and enterprise applications requiring robust connectivity and low-latency communication. This adapter is part of QLogic's extensive portfolio of networking solutions, catering to the needs of high-performance computing (HPC), cloud computing, and virtualization environments.

One of the standout features of the IB6054601-00 D is its capability to support data transfer rates of up to 56 Gbps. This makes it ideal for applications demanding large bandwidth and quick data processing. The adapter is optimized for RDMA (Remote Direct Memory Access) technology, which allows data to be transferred directly between the memory of different computers without involving the CPU. This reduces latency and CPU overhead, leading to enhanced overall system performance.

The architecture of the IB6054601-00 D includes support for a dual-port design, which offers increased bandwidth, redundancy, and fault tolerance. This dual-port configuration is especially advantageous for environments that require high availability and reliability, such as financial services and mission-critical applications.

The adapter utilizes advanced error detection and correction mechanisms, ensuring that data integrity is maintained during transmission. With features like adaptive routing and congestion management, the IB6054601-00 D is capable of optimizing the handling of data flows, thereby enhancing performance even under heavy loads.

In terms of compatibility, the Q-Logic IB6054601-00 D supports a wide range of operating systems and virtualization technologies, making it easy to integrate into diverse IT environments. It also includes drivers and software packages that facilitate seamless deployment and management.

In addition to high-speed connectivity, the adapter is designed with power efficiency in mind. It adheres to Energy Star regulations, helping organizations lower their operational costs while minimizing their environmental footprint.

Overall, the Q-Logic IB6054601-00 D stands out for its high throughput, low latency, and reliability. Its combination of advanced features and technologies positions it as an excellent choice for organizations looking to enhance their data center performance and maximize the efficiency of their network infrastructure. With the growing demands for faster and more efficient data transfer, solutions like the IB6054601-00 D are essential in meeting the evolving needs of modern enterprises.
Top Page Image Contents