Tuesday Jun 23, 2009

Building Optimized Configurations for MD Nastran

MD Nastran is a very high compute and I/O intensive application. Jobs can run for days and generate terabytes of data.  Any processing or I/O optimization will help running more jobs concurrently and for shorter periods of time. From storage (Internal? Network Attached? SAS? HDD? SSD?) to memory (how much per node? per job?), the Sun BluePrints article "Sun Business Ready HPC for MD Nastran" provides interesting suggestions to help designing optimized configurations and avoiding processor or I/O bottlenecks. Detailed configuration examples and performance test results are provided.

To stay up to date when new Sun BluePrints are published, join the Sun BluePrint Community on Facebook or subscribe to the Sun BluePrints Community RSS Feed.

Thursday Jun 18, 2009

How To Increase HPC Productivity over 54%

The performance of some of the most popular High-Performance Computing (HPC) applications can be dramatically increased by simply swapping out drives. This is one of the findings by Larry McIntosh and Michael Burke, authors of the BluePrint "Solid State Drives in HPC: Reducing the I/O Bottleneck".

In one of the performance tests, by simply replacing the traditional 15K SAS  Hard Disk Drives (HDDs) with SATA Solid State Drives (SSDs), the increase in productivity reached 54.42% (xxocmd2 and xlotdf1 NASTRAN benchmark tests running on 8 processing cores).

Similar tests are run with other HPC finite-element analysis applications such as ABAQUS or ANSYS as well as with the Lustre parallel file system (using IOZone) and they all demonstrated significant performance improvements.

This Sun BluePrints article is a must read if you are running HPC applications and I/O bottleneck is what limits your productivity.

To stay up to date when new Sun BluePrints are published, join the Sun BluePrint Community on Facebook or subscribe to the Sun BluePrints Community RSS Feed.

About

PierreReynes

Search

Categories
Archives
« April 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
   
       
Today