Welfenlab - Leibniz 
                        Universität Hannover Welfenlab Leibniz Universität Hannover
Search:

Revision [71]

Last edited on 2009-08-19 12:22:34 by GuidoBoettcher
Additions:
=== Common Benchmarks ===


Revision [70]

Edited on 2009-08-19 12:22:13 by GuidoBoettcher
Additions:
=== Linpack ===
Deletions:
=== Linpack ==


Revision [69]

Edited on 2009-08-19 12:22:04 by GuidoBoettcher
Additions:
For HPC system a reference benchmark is the Linpack. You can download the benchmark
from the links below
[[http://netlib.org/benchmark/hpl/ Linpack Webpage]] (including patch for QS22)
For QS22 you will also need (also described in readme)
- ATLAS 3.8.1 (warning: long build, takes approx. 12 hours)
- OpenMPI 1.3.3 (custom version for infiniband)


Revision [68]

Edited on 2009-08-19 12:18:19 by GuidoBoettcher
Additions:
[[http://netlib.org/benchmark/hpl/IBM_README.txt Readme to QS22-optimized linpack]]
Deletions:
[[http://netlib.org/benchmark/hpl/IBM_README.txt] Readme to QS22-optimized linpack]


Revision [67]

Edited on 2009-08-19 12:18:09 by GuidoBoettcher
Additions:
=== Linpack ==
[[http://netlib.org/benchmark/hpl/IBM_README.txt] Readme to QS22-optimized linpack]


Revision [61]

Edited on 2009-08-19 12:07:43 by GuidoBoettcher
Additions:
You will have to start the so-called subnet manager **opensm** on one of the machines by typing
Deletions:
You will have to start the so-called subnet manager **opensm* on one of the machines by typing


Revision [60]

Edited on 2009-08-19 12:07:27 by GuidoBoettcher
Additions:
- MVAPICH 1.1.0
==== InfiniBand related problems ====
Q: I cannot reach the nodes by tcp/ip. The node fails to respond on a icmp ping.
For some reason, the InfiniBand network switches have no //logic// of detecting connected network adapters. In contrast to ethernet, the //lookup// similar to an ARP Cache is not provided by the switch. Thus, a host connected to the switch needs to run this logic part for tcp/ip connections.
You will have to start the so-called subnet manager **opensm* on one of the machines by typing
%%(language-ref)
service opensmd start
%%
Afterwards you should be able to //ping// a node.
typical results so far should be: ~1500MiB/s on rdma transfers
Advanced tests conducting on the mpi are also available
- OSU tests (from Ohio State University, a benchmark suite provided by mvapich)
- IMB (Intel MPI Benchmark)
Deletions:
- MVAPICH 1.0.3
Advanced tests conducting on the mpi


Revision [59]

The oldest known version of this page was created on 2009-08-19 11:59:37 by GuidoBoettcher
Page was generated in 0.0189 seconds