Cluster Documentation Project - User contributions [en]

Main Page

2026-09-29T20:37:45Z

Deadline: /* Cluster Documentation Project™ */

__NOTOC__
= Cluster Documentation Project™ =
[[File:cdp_logo_1.png|left|250x250px]]
Looking for Cluster Documentation? Jump right to [[Cluster Topics]].

Computational Clusters have grown from a back room curiosity to a driving force in the HPC community. The roll your own nature of cluster computing has also created a wide community of users, vendors, and applications. The Cluster Documentation Project (CDP) is an effort to help document best practices, tutorials, projects, benchmarks, products, and anything else that helps facilitate HPC cluster computing.

'''We plan on releasing our first book at in the near future ''' -- ''[[Essential HPC]]: Practical high performance concepts, techniques, and procedures''

'''VENDORS DON'T MISS A UNIQUE AND EFFECTIVE SPONSORSHIP OPPORTUNITY: [[Funding The Cluster Documentation Project]]'''

[[File:E-HPC-cover2.png|thumb|alt=Book Cover.|Contents [[Essential HPC]]]]

===Project Scope===

The word [http://en.wikipedia.org/wiki/Computer_cluster cluster] has many meanings. The Cluster Documentation Project is designed to document and professionally publish High Performance Computing (HPC) Cluster information. The focus will be on tutorial and how-to information rather than theoretical aspects of HPC. Currently there are [[Cluster Topics]] which is a compendium of current benchmarks, products, technologies, methods, best practices, etc. that will help those interested in clustering. When funded the project will enhance this resource.

See an [http://www.admin-magazine.com/HPC/Articles/Cluster-Documentation-Project/%28language%29/eng-US article] in HPC Admin magazine for more background.

Background: [[Beowulf Cluster History Project]]

===Writers and Contributors===

Contributions are accepted from the community -- both users and vendors. All content is subject to approval by the editors. If you are interested in contributing please contact [https://www.clustermonkey.net/contact-mainmenu-3/53-head-monkey/1-douglas-eadline.html Douglas Eadline]. All submitted content will be made available under a [http://creativecommons.org/licenses/by-nc-sa/2.5/ Creative Commons License].

===Questions===
If you have specific questions about clusters please visit our [[Cluster Questions]] section and enter your question.

If you have questions about this site, please contact [http://www.clustermonkey.net//component/option,com_contact/task,view/contact_id,1/Itemid,3/ Douglas Eadline] with any questions.

===History===
ClusterWorld magazine was the initial sponsor of a site called The Cluster Agenda. When ClusterWorld Magazine ceased publication the Agenda effort was stopped. Douglas Eadline, editor of ClusterWorld magazine, was the principle organizer of the Agenda effort and has decided with the help of Jeff Layton to preserve some of the content from the Agenda and morph the site into Cluster Tweaks. The original Agenda Working Committee was Jeff Layton, Glen Otero, Dan Stanzione, and Douglas Eadline.

Limulus Benchmarks

2026-09-29T16:07:13Z

Deadline: Created page with "===Various Limulus Benchmarks (historical)=== ==HPL Performance== '''i5-2400S Sandy Bridge''' * 200.3 GFLOPS N=40220 (Raw HPL Results) * 58% of Peak (3.3GHz * 4 cores * 8 DP FLOPS/cycle) + (2.5Ghz * 12 cores * 8 FLOPS/cycle) = 345.6 GFLOPS Peak * Three i5-2400S and one i5-2500K each with 4 MB DDR3 RAM, GbE, Intel MKL and compilers '''i5-3470S Ivybridge''' * 256.4 GFLOPS N=58800 (Raw HPL Results) * 69% of Peak (2.9GHz * 16 cores * 8 DP FLOPS/cycle) = 371.2 GFLOPS..."

===Various Limulus Benchmarks (historical)===
==HPL Performance==
'''i5-2400S Sandy Bridge'''

* 200.3 GFLOPS N=40220 (Raw HPL Results)
* 58% of Peak (3.3GHz * 4 cores * 8 DP FLOPS/cycle) + (2.5Ghz * 12 cores * 8 FLOPS/cycle) = 345.6 GFLOPS Peak
* Three i5-2400S and one i5-2500K each with 4 MB DDR3 RAM, GbE, Intel MKL and compilers

'''i5-3470S Ivybridge'''

* 256.4 GFLOPS N=58800 (Raw HPL Results)
* 69% of Peak (2.9GHz * 16 cores * 8 DP FLOPS/cycle) = 371.2 GFLOPS Peak
* Four i5-3470S each with 8 MB DDR3 RAM, GbE, Intel MKL and compilers

'''i5-4570S Haswell'''

* 385.5 GFLOPS N=60000 (Raw HPL Results)
* 52% of Peak (2.9GHz * 16 cores * 16 DP FLOPS/cycle) = 742.4 GFLOPS Peak (Note: Haswell is now 16 FLOPS/cycle)
* Four i5-4570S each with 8 MB DDR3 RAM, GbE, Intel MKL and compilers
 
* 567.4 GFLOPS N=60000 (Raw HPL Results)
* 76% of Peak (2.9GHz * 16 cores * 16 DP FLOPS/cycle) = 742.4 GFLOPS Peak (Note: Haswell is now 16 FLOPS/cycle)
* Four i5-4570S each with 8 MB DDR3 RAM, 10-GbE, Intel MKL and compilers

'''i7-4770S Haswell'''

* 444.8 GFLOPS N=86000 (Raw HPL Results)
* 56% of Peak (3.1GHz * 16 cores * 16 DP FLOPS/cycle) = 793.6 GFLOPS Peak (Note: Haswell is now 16 FLOPS/cycle)
* Four i7-4770S each with 16 MB DDR3 RAM, GbE, HT disabled, Intel MKL and compilers
 
* 498.3 GFLOPS N=126000 (Raw HPL Results)
* 63% of Peak (3.1 GHz * 16 cores * 16 DP FLOPS/cycle) = 793.6 GFLOPS Peak (Note: Haswell is now 16 FLOPS/cycle)
* Four i7-4770S each with 32 MB DDR3 RAM, GbE, HT disabled, Intel MKL and compilers

'''i5-6500 Skylake'''

* 480.2 GFLOPS N=86000 (Raw HPL Results)
* 59% of Peak (3.2GHz * 16 cores * 16 DP FLOPS/cycle) = 819.2 GFLOPS Peak (Note: Skylake is 16 FLOPS/cycle)
* Four i5-6500 each with 16 MB DDR4 RAM, GbE, Intel MKL and compilers
 
* 658.3 GFLOPS N=86000 (Raw HPL Results)
* 80% of Peak (3.2GHz * 16 cores * 16 DP FLOPS/cycle) = 819.2 GFLOPS Peak (Note: Skylake is 16 FLOPS/cycle)
* Four i5-6500 each with 16 MB DDR4 RAM, 10-GbE, Intel MKL and compilers

'''i7-6700 Skylake'''

* 592.5 GFLOPS N=126000 (Raw HPL Results)
* 68% of Peak (3.4GHz * 16 cores * 16 DP FLOPS/cycle) = 870.4 GFLOPS Peak (Note: Skylake is 16 FLOPS/cycle)
* Four i7-6700 each with 32 MB DDR4 RAM, GbE, HT disabled, Intel MKL and compilers
 
* 640.4 GFLOPS N=180000 (Raw HPL Results)
* 74% of Peak (3.4GHz * 16 cores * 16 DP FLOPS/cycle) = 870.4 GFLOPS Peak (Note: Skylake is 16 FLOPS/cycle)
* Four i7-6700 each with 64 MB DDR4 RAM, GbE, HT disabled, Intel MKL and compilers

===Hadoop===

'''Historical Results'''

The cluster consists of four nodes with the following hardware specifications.

* Processor : Intel Core i5-3470S CPU @ 2.90GHz (65W Quad-core Ivy Bridge)
* Memory: 16 GB (DDR3-1600)
* Interconnect: Intel GbE (MTU=4500)
* HDFS Storage: 192 GB SSD (Plexstor M5S 128 GB, Plexstor M5S 64 GB)
* Hadoop Version: 2.2.0 (Hortonworks HDP 2.0 )
* HDFS size: 696.23 GB
* HDFS replication: 2
* OS: Scientific Linux 6.4
* Number of Data nodes: 4
* Number of worker nodes: 4

TestDFSIO (average of 10 runs):

* write: 13.86 MB/sec Single File; 221.80 MB/sec Total Throughput
* read: 53.97 MB/sec Single File; 863.49 MB/sec Total Throughput

Terasort (100 GB):

* 886 seconds, 112.87 MB/sec

'''Previous Results'''

Presented a tutorial on Hadoop recently and used my original Limulus to demonstrate a four node Hadoop cluster. Some of the hardware is actually quite old. The nodes have a single dual core E6550, with 4 GB RAM and a 64MB SSD. The head node has a quad-core Sandy Bridge (i5-2400S) with 4 GB RAM, 64MB SSD, and .5 TB of RAID1. Connections were using the mediocre consumer based Realtek GigE Ethernet controllers on the nodes. Overall it worked well. I plan on making a Hadoop VNFS so that it is possible to boot the nodes into Hadoop. Used the Hortonworks HDP for the Hadoop install. The following are results of the TestDFSIO benchmark (tests the Hadoop File System)

<pre>
----- TestDFSIO ----- : write
Date & time: Wed Feb 06 22:23:24 EST 2013
Number of files: 10
Total MBytes processed: 10000
Throughput mb/sec: 9.684751649071087
Average IO rate mb/sec: 9.751626968383789
IO rate std deviation: 0.8502143053358981
Test exec time sec: 141.977

----- TestDFSIO ----- : read
Date & time: Wed Feb 06 22:26:34 EST 2013
Number of files: 10
Total MBytes processed: 10000
Throughput mb/sec: 44.49229838314988
Average IO rate mb/sec: 51.67889404296875
IO rate std deviation: 23.5285090607248
Test exec time sec: 58.388
</pre>

The Commercial Limulus Cluster

2026-09-29T15:52:49Z

Deadline: new

===Limulus Computing for Commercial Limulus Systems===

Please see [https://www.limulus-computing.com/ Limulus Computing] for full details on the commercial Limulus systems. These system are fully tested, supported, and professional constructed.

Full HPL Results 2008

2026-09-29T15:49:12Z

Deadline: new

<pre>
============================================================================
HPLinpack 1.0a -- High-Performance Linpack benchmark -- January 20, 2004
Written by A. Petitet and R. Clint Whaley, Innovative Computing Labs., UTK
============================================================================

An explanation of the input/output parameters follows:
T/V : Wall time / encoded variant.
N : The order of the coefficient matrix A.
NB : The partitioning blocking factor.
P : The number of process rows.
Q : The number of process columns.
Time : Time in seconds to solve the linear system.
Gflops : Rate of execution for solving the linear system.

The following parameter values will be used:

N : 28600
NB : 220
PMAP : Row-major process mapping
P : 2
Q : 2
PFACT : Left
NBMIN : 4
NDIV : 2
RFACT : Right
BCAST : 1ringM
DEPTH : 0
SWAP : Mix (threshold = 64)
L1 : transposed form
U : transposed form
EQUIL : yes
ALIGN : 8 double precision words

----------------------------------------------------------------------------

- The matrix A is randomly generated for each test.
- The following scaled residual checks will be computed:
1) ||Ax-b||_oo / ( eps * ||A||_1 * N )
2) ||Ax-b||_oo / ( eps * ||A||_1 * ||x||_1 )
3) ||Ax-b||_oo / ( eps * ||A||_oo * ||x||_oo )
- The relative machine precision (eps) is taken to be 1.110223e-16
- Computational tests pass if scaled residuals are less than 16.0

============================================================================
T/V N NB P Q Time Gflops
----------------------------------------------------------------------------
WR01R2L4 28600 220 2 2 291.85 5.344e+01
----------------------------------------------------------------------------
||Ax-b||_oo / ( eps * ||A||_1 * N ) = 0.0038030 ...... PASSED
||Ax-b||_oo / ( eps * ||A||_1 * ||x||_1 ) = 0.0054760 ...... PASSED
||Ax-b||_oo / ( eps * ||A||_oo * ||x||_oo ) = 0.0011149 ...... PASSED
============================================================================

Finished 1 tests with the following results:
1 tests completed and passed residual checks,
0 tests completed and failed residual checks,
0 tests skipped because of illegal input values.
----------------------------------------------------------------------------

End of Tests.
============================================================================

</pre>

Full HPL Results 2007

2026-09-29T15:47:49Z

Deadline: new

<pre>
============================================================================
HPLinpack 1.0a -- High-Performance Linpack benchmark -- January 20, 2004
Written by A. Petitet and R. Clint Whaley, Innovative Computing Labs., UTK
============================================================================

An explanation of the input/output parameters follows:
T/V : Wall time / encoded variant.
N : The order of the coefficient matrix A.
NB : The partitioning blocking factor.
P : The number of process rows.
Q : The number of process columns.
Time : Time in seconds to solve the linear system.
Gflops : Rate of execution for solving the linear system.

The following parameter values will be used:

N : 19360
NB : 220
PMAP : Row-major process mapping
P : 2
Q : 2
PFACT : Left
NBMIN : 4
NDIV : 2
RFACT : Right
BCAST : 1ringM
DEPTH : 0
SWAP : Mix (threshold = 64)
L1 : transposed form
U : transposed form
EQUIL : yes
ALIGN : 8 double precision words

----------------------------------------------------------------------------

- The matrix A is randomly generated for each test.
- The following scaled residual checks will be computed:
1) ||Ax-b||_oo / ( eps * ||A||_1 * N )
2) ||Ax-b||_oo / ( eps * ||A||_1 * ||x||_1 )
3) ||Ax-b||_oo / ( eps * ||A||_oo * ||x||_oo )
- The relative machine precision (eps) is taken to be 1.110223e-16
- Computational tests pass if scaled residuals are less than 16.0

============================================================================
T/V N NB P Q Time Gflops
----------------------------------------------------------------------------
WR01R2L4 19360 220 2 2 101.40 4.771e+01
----------------------------------------------------------------------------
||Ax-b||_oo / ( eps * ||A||_1 * N ) = 0.0041180 ...... PASSED
||Ax-b||_oo / ( eps * ||A||_1 * ||x||_1 ) = 0.0062675 ...... PASSED
||Ax-b||_oo / ( eps * ||A||_oo * ||x||_oo ) = 0.0011940 ...... PASSED
============================================================================

Finished 1 tests with the following results:
1 tests completed and passed residual checks,
0 tests completed and failed residual checks,
0 tests skipped because of illegal input values.
----------------------------------------------------------------------------

End of Tests.
============================================================================
</pre>

The Norbert Limulus Cluster

2026-09-29T15:36:52Z

Deadline:

==Norbert Design==

Norbert was designed to use totally off-the-shelf parts. There is no custom case work required. Of course the hardware could be placed in a smaller volume with a custom case design, but that would require extra work and our goal is to create an easy to assemble system.

Norbert basically consists of a head node and three diskless worker nodes. All nodes have an extra PCIe GigE card for compute traffic. The head node has a DVD, a hard drive, and an extra network card to connect to the LAN. Each node has a single socket Gigbyte GA-G33M-DS2R motherboard with a 2.33 GHz Core2 Duo (E6550) processor, and one GByte of RAM (upgraded to 2GB in 2008). There are a total of eight cores in Norbert.
Norbert Pictures

Norbert measures 26 inches (66 cm) across, 16 (41 cm) inches deep, and 25 (64 cm) inches high. I placed the cases on a inexpensive wire shelf cart so it can be moved easily.

The following is a front view. The two GigE switches are on the top. Units can be removed without too much trouble, however, there is little need to take anything off the shelves. The cases have build in temperature sensors (blue lights on the front). The head node is located on the top left (it has a DVD drive).

[[File:limulus-front-500x414_AwwV.jpg|center]]

The following a picture of the back. The two switches can be seen on the top. Of course there is a need for some cabling clean up (I plan on making custom length cables). There is a powerstrip on bottom middle of the shelf. The single powerstrip cord is the only power connection needed.

[[File:limulus-back-500x375_AwwV.jpg|center]]

===Hardware Manifest===

[[File:HW_Manifest.png]]

===Performance===

*HPL 2007: 47.71 GFLOPS before RAM upgrade (1 GB per node) [[Full HPL Results 2007]]
*HPL 2008: 53.44 GFLOPS after RAM upgrade (2 GB per node) [[Full HPL Results 2008]]

===Price to Performance===

September 2007: $48.25/GFLOP
September 2008: $39.15/GFLOP

===Power===

Note: all measurements were made "at the wall" using a Kill-O-Watt meter.

'''Node Idle and Load Power'''

The following table summarizes the idle and loaded power draw (in Watts) measured for each node and processor (E6550=Core2 Dual 2.33GHz, Q6600=Core2 Quad 2.4GHz). All cores were busy running the single copies of the NAS parallel benchmark. The highest Watts valued was used.

[[File:Power-Use.png]]

Of course, if the nodes are not being used, they are manually powered down to stand-by mode (5Watts) and then restarted with WOL (wake on lan). I would like to integrate this into the scheduler.

===Power Load Running HPL===

When it is fully loaded running the HPL benchmark Norbert (four E6550's) draws between 371-483 Watts. The number varied as the benchmark progressed which I assumed is an indication of how modern processors are managing power consumption based on load. Even with the maximum power draw, the power performance ratio is a comfortable 10-11 Watts per GFLOPS.

File:Power-Use.png

2026-09-29T15:31:10Z

Deadline:

The Norbert Limulus Cluster

2026-09-29T15:30:51Z

Deadline:

==Norbert Design==

Norbert was designed to use totally off-the-shelf parts. There is no custom case work required. Of course the hardware could be placed in a smaller volume with a custom case design, but that would require extra work and our goal is to create an easy to assemble system.

Norbert basically consists of a head node and three diskless worker nodes. All nodes have an extra PCIe GigE card for compute traffic. The head node has a DVD, a hard drive, and an extra network card to connect to the LAN. Each node has a single socket Gigbyte GA-G33M-DS2R motherboard with a 2.33 GHz Core2 Duo (E6550) processor, and one GByte of RAM (upgraded to 2GB in 2008). There are a total of eight cores in Norbert.
Norbert Pictures

Norbert measures 26 inches (66 cm) across, 16 (41 cm) inches deep, and 25 (64 cm) inches high. I placed the cases on a inexpensive wire shelf cart so it can be moved easily.

The following is a front view. The two GigE switches are on the top. Units can be removed without too much trouble, however, there is little need to take anything off the shelves. The cases have build in temperature sensors (blue lights on the front). The head node is located on the top left (it has a DVD drive).

[[File:limulus-front-500x414_AwwV.jpg|center]]

The following a picture of the back. The two switches can be seen on the top. Of course there is a need for some cabling clean up (I plan on making custom length cables). There is a powerstrip on bottom middle of the shelf. The single powerstrip cord is the only power connection needed.

[[File:limulus-back-500x375_AwwV.jpg|center]]

===Hardware Manifest===

[[File:HW_Manifest.png]]

===Performance===

HPL 2007: 47.71 GFLOPS before RAM upgrade (1 GB per node) Full HPL Results
HPL 2008: 53.44 GFLOPS after RAM upgrade (2 GB per node) Full HPL Results
Price to Performance

September 2007: $48.25/GFLOP
September 2008: $39.15/GFLOP

===Power===

Note: all measurements were made "at the wall" using a Kill-O-Watt meter.

'''Node Idle and Load Power'''

The following table summarizes the idle and loaded power draw (in Watts) measured for each node and processor (E6550=Core2 Dual 2.33GHz, Q6600=Core2 Quad 2.4GHz). All cores were busy running the single copies of the NAS parallel benchmark. The highest Watts valued was used.

Of course, if the nodes are not being used, they are manually powered down to stand-by mode (5Watts) and then restarted with WOL (wake on lan). I would like to integrate this into the scheduler.

===Power Load Running HPL===

When it is fully loaded running the HPL benchmark Norbert (four E6550's) draws between 371-483 Watts. The number varied as the benchmark progressed which I assumed is an indication of how modern processors are managing power consumption based on load. Even with the maximum power draw, the power performance ratio is a comfortable 10-11 Watts per GFLOPS.

File:HW Manifest.png

2026-09-29T15:29:38Z

Deadline:

The Norbert Limulus Cluster

2026-09-29T15:29:09Z

Deadline:

==Norbert Design==

Norbert was designed to use totally off-the-shelf parts. There is no custom case work required. Of course the hardware could be placed in a smaller volume with a custom case design, but that would require extra work and our goal is to create an easy to assemble system.

Norbert basically consists of a head node and three diskless worker nodes. All nodes have an extra PCIe GigE card for compute traffic. The head node has a DVD, a hard drive, and an extra network card to connect to the LAN. Each node has a single socket Gigbyte GA-G33M-DS2R motherboard with a 2.33 GHz Core2 Duo (E6550) processor, and one GByte of RAM (upgraded to 2GB in 2008). There are a total of eight cores in Norbert.
Norbert Pictures

Norbert measures 26 inches (66 cm) across, 16 (41 cm) inches deep, and 25 (64 cm) inches high. I placed the cases on a inexpensive wire shelf cart so it can be moved easily.

The following is a front view. The two GigE switches are on the top. Units can be removed without too much trouble, however, there is little need to take anything off the shelves. The cases have build in temperature sensors (blue lights on the front). The head node is located on the top left (it has a DVD drive).

[[File:limulus-front-500x414_AwwV.jpg|center]]

The following a picture of the back. The two switches can be seen on the top. Of course there is a need for some cabling clean up (I plan on making custom length cables). There is a powerstrip on bottom middle of the shelf. The single powerstrip cord is the only power connection needed.

[[File:limulus-back-500x375_AwwV.jpg|center]]

===Hardware Manifest===

===Performance===

HPL 2007: 47.71 GFLOPS before RAM upgrade (1 GB per node) Full HPL Results
HPL 2008: 53.44 GFLOPS after RAM upgrade (2 GB per node) Full HPL Results
Price to Performance

September 2007: $48.25/GFLOP
September 2008: $39.15/GFLOP

===Power===

Note: all measurements were made "at the wall" using a Kill-O-Watt meter.

'''Node Idle and Load Power'''

The following table summarizes the idle and loaded power draw (in Watts) measured for each node and processor (E6550=Core2 Dual 2.33GHz, Q6600=Core2 Quad 2.4GHz). All cores were busy running the single copies of the NAS parallel benchmark. The highest Watts valued was used.

Of course, if the nodes are not being used, they are manually powered down to stand-by mode (5Watts) and then restarted with WOL (wake on lan). I would like to integrate this into the scheduler.

===Power Load Running HPL===

When it is fully loaded running the HPL benchmark Norbert (four E6550's) draws between 371-483 Watts. The number varied as the benchmark progressed which I assumed is an indication of how modern processors are managing power consumption based on load. Even with the maximum power draw, the power performance ratio is a comfortable 10-11 Watts per GFLOPS.

File:Limulus-back-500x375 AwwV.jpg

2026-09-29T15:16:01Z

Deadline:

File:Limulus-front-500x414 AwwV.jpg

2026-09-29T15:15:43Z

Deadline:

The Norbert Limulus Cluster

2026-09-29T15:15:00Z

Deadline: new

'''Norbert Design'''

Norbert was designed to use totally off-the-shelf parts. There is no custom case work required. Of course the hardware could be placed in a smaller volume with a custom case design, but that would require extra work and our goal is to create an easy to assemble system.

Norbert basically consists of a head node and three diskless worker nodes. All nodes have an extra PCIe GigE card for compute traffic. The head node has a DVD, a hard drive, and an extra network card to connect to the LAN. Each node has a single socket Gigbyte GA-G33M-DS2R motherboard with a 2.33 GHz Core2 Duo (E6550) processor, and one GByte of RAM (upgraded to 2GB in 2008). There are a total of eight cores in Norbert.
Norbert Pictures

Norbert measures 26 inches (66 cm) across, 16 (41 cm) inches deep, and 25 (64 cm) inches high. I placed the cases on a inexpensive wire shelf cart so it can be moved easily.

The following is a front view. The two GigE switches are on the top. Units can be removed without too much trouble, however, there is little need to take anything off the shelves. The cases have build in temperature sensors (blue lights on the front). The head node is located on the top left (it has a DVD drive).

limulus-front-500x414.jpg

The following a picture of the back. The two switches can be seen on the top. Of course there is a need for some cabling clean up (I plan on making custom length cables). There is a powerstrip on bottom middle of the shelf. The single powerstrip cord is the only power connection needed.

Limulus Software

2026-09-29T15:12:02Z

Deadline: added image

'''Limulus Software'''

The following is a list of basic cluster RPMS that will be included in the software stack. The base distribution will be Scientific Linux. An open [https://www.limulus-computing.com/software/software up-to-date software stack] is used for modern Limulus machines.

# Scientific Linux V6.3 - RHEL work alike distribution
# Warewulf Cluster Toolkit - Cluster provisioning and administration (V3.1)
# PDSH - Parallel Distributed Shell for collective administration
# Whatsup - A cluster node up/down detection utility
# Open Grid Scheduler - previously Sun Grid Engine Resource Scheduler
# Ganglia - Cluster Monitoring System
# GNU Compilers (gcc, g++, g77, gdb) - Standard GNU compiler suite
# Modules - Manages User Environments
# PVM - Parallel Virtual Machine (message passing middleware)
# MPICH2 - MPI Library (message passing middleware)
# OPEN-MPI - MPI Library (message passing middleware)
# Open-MX - Myrinet Express over Ethernet
# ATLAS - host tuned BLAS library
# OpenBLAS - optimized BLAS library (previously Goto library)
# FFTW - Optimized FFT library
# FFTPACK - FFT library
# LAPACK and BLAS - Linear Algebra library
# ScaLAPACK - Scalable Linear Algebra Package
# PetSc Scalable PDE solvers
# GNU GSL - GNU Scientific Library (over 1000 functions)
# PADB - Parallel Application Debugger Inspection Tool
# Julia - Easy To Use High Performance Parallel Scientific Language
# Userstat - a "top" like job queue/node monitoring application
# Beowulf Performance Suite - benchmark and testing suite
# relayset - power relay control utility
# ssmtp - mail forwarder for nodes

'''Automatic Power Control'''

One key design component of the new Limulus Case is software controlled power to the nodes. This feature will allow nodes to be powered-on only when needed. As an experiment, a simple script was written that monitors the Grid Engine queue. If there are jobs waiting in the queue, nodes are powered-on. Once the work is done (i.e. nothing more in the queue) the nodes are powered off). '''Note:''' Newer Limulus systems use the SLURM Scheduler which provides hooks to power nodes off and on similar to this Grid Engine example.

As an example, an 8 core job was run on the Norbert cluster (in the Limulus case). The head node has 4 cores and each worker node has 2 cores for a total of 10 cores. An 8 node job was submitted via Grid Engine with only the head node powered-on. The script noticed the job waiting in the queue and turned on a single worker node to give 6 cores total, which were still not enough. Another node was powered-on and the total cores reached 8 and the job started to run. After completion, the script noticed that there was nothing in the queue and shutdown the nodes.

To see how the number of nodes changes with the work in the queue, consider the Ganglia trace below. Note that the number of nodes in the cluster load graph (green line) changes from 1 to 2 then 3 then back down to 1. Similarly the number of CPUs (red line) raises to 8 then back to down to 4 (the original 4 cores in the head node). Similar changes can be seen in the system memory.

[[File:Ganglia-sge-control3-600x258_fMvP.jpg|center]]

File:Ganglia-sge-control3-600x258 fMvP.jpg

2026-09-29T15:10:14Z

Deadline:

Limulus Software

2026-09-29T15:09:20Z

Deadline: new

Limulus Hardware

2026-09-29T14:58:03Z

Deadline: new

The following is the Limulus Reference design. This design was used for the latest Norbert cluster. Notice that there are two separate Ethernet switches. Ostensibly, one is for compute traffic and one for administration and file system traffic, however, there is no restriction on how applications may use these two channels. Also note that one port on the switch is used for expansion within the private cluster network.

[[File:Limulus-Design.png|center]]

File:Limulus-Design.png

2026-09-29T14:52:44Z

Deadline:

Limulus Design

2026-09-29T14:13:33Z

Deadline: new

The definition of a Limulus is as follows:

# Uses standard hardware off the shelf hardware and allows for both file system and compute node expansion;
# Uses an Open Source cluster software stack that contains an integrated set of software packages to manage the cluster and create and run software. The software stack must run on top of existing Linux distributions;
# Consumes less than 150 watts when idle and less then 1200 watts when fully loaded;
# Has a power supply efficiency above 80% to reduce heat
# Has a noise signature that is acceptable in a modern office environment (less than 45dB).
# Achieves a price to performance of less then $100(US) per GFLOPS

As you can see some of the design parameters are rather loose allowing design flexibility, while others are very tight providing desktop/office functionality.

A good explanation of the [https://www.limulus-computing.com/Limulus-Manual/doku.php?id=limulus_design_philosophy Limulus Design Philosophy] is available from Limulus Computing.

Frequently Asked Questions

2026-09-29T13:57:52Z

Deadline: new

Rather than managing two copies, the [https://www.limulus-computing.com/help/frequenly-asked-questions Limulus FAQ] has been transferred to Limulus Computing.

Limulus Project

2026-09-29T13:54:49Z

Deadline:

Information on the Limulus project is mostly for historical value. Current open development is performed by [https://www.limulus-computing.com Limulus Computing] (consult the [https://www.limulus-computing.com/Limulus-Manual/doku.php Systems Manual] for details of these systems).

'''What is the Limulus Project?'''

The Limulus project goal is to create and maintain an open specification and software stack for a personal workstation cluster. As "edge of cloud" computing becomes more important, these systems are designed to offer HPC, Big Data (Hadoop-Spark), and Deep Learning in almost any location. Ideally, a user should be able to build or purchase a small personal workstation cluster using the Limulus reference design and low cost hardware. In addition, a freely available turn-key Linux based software stack will be created and maintained for use on the Limulus design. A Limulus is intended to be a local workstation cluster platform where users can develop software, test ideas, run edge applications, and teach HPC methods. Consult the following draft of [https://www.clustermonkey.net/download/whitepapers/Limulus-Nov-2007-draft.pdf The Personal Linux Cluster: Putting GFLOPS on Desktops (pdf)] from Linux Magazine for more background information. The Limulus idea actually came from the [https://www.clustermonkey.net/learning-about-clusters-topmenu-44/projects/41-the-value-cluster-cheaper-better-faster.html Clusterworld Magazine 2005 AMD Value Cluster Project].

'''Topics:'''

[[Frequently Asked Questions]]
[[Limulus Design]]
[[Limulus Hardware]]
[[Limulus Software]]
[[The Norbert Limulus Cluster]]
[[The Commercial Limulus Cluster]]
[[Limulus Benchmarks]]

'''Project Status'''

The research project has transferred to an open commercial endeavor (Limulus Computing).

The project is was hosted by Basement Supercomputing (now [https://www.limulus-computing.com Limulus Computing]).

'''Limulus Trademarks and Licensing'''

The Limulus Logo and the Limulus Project are trademarks of [https://www.limulus-computing.com Limulus Computing]. For commercial use of project trademarks or sponsorships, contact Limulus Computing.

Software used by the project is subject to license terms and copyrights as per the authors. Please consult individual packages for more information.

Cluster Documentation Project:About

2026-09-29T13:52:12Z

Deadline:

The Cluster Documentation Project is designed to document and professionally publish High Performance Computing (HPC) Cluster information. The focus will be on tutorial and how-to information rather than theoretical aspects of HPC. Our goal is to produce a freely available wiki called The Cluster Documentation Project. The site provides a central location for cluster information, trick, tips, how-to's, benchmarks, products, etc. that may be of interest to HPC cluster users. It a document for the community by the community.

Join and contribute !

'''The Cluster Documentation Project''' is maintained by the HPC community and [http://www.clustermonkey.net ClusterMonkey]

[[Image:CM_logo-440x150-tm.jpg|center]]

For banner advertising please contact [https://www.clustermonkey.net/contact-mainmenu-3/53-head-monkey/1-douglas-eadline.html The Head Monkey] as ClusterMonkey.

Limulus Project

2026-09-29T13:46:26Z

Deadline: updated links

Information on the Limulus project is mostly for historical value. Current open development is performed by [https://www.limulus-computing.com Limulus Computing] (consult the [https://www.limulus-computing.com/Limulus-Manual/doku.php Systems Manual] for details of these systems).

'''What is the Limulus Project?'''

The Limulus project goal is to create and maintain an open specification and software stack for a personal workstation cluster. As "edge of cloud" computing becomes more important, these systems are designed to offer HPC, Big Data (Hadoop-Spark), and Deep Learning in almost any location. Ideally, a user should be able to build or purchase a small personal workstation cluster using the Limulus reference design and low cost hardware. In addition, a freely available turn-key Linux based software stack will be created and maintained for use on the Limulus design. A Limulus is intended to be a local workstation cluster platform where users can develop software, test ideas, run edge applications, and teach HPC methods. Consult the following draft of [https://www.clustermonkey.net/download/whitepapers/Limulus-Nov-2007-draft.pdf The Personal Linux Cluster: Putting GFLOPS on Desktops (pdf)] from Linux Magazine for more background information. The Limulus idea actually came from the [https://www.clustermonkey.net/learning-about-clusters-topmenu-44/projects/41-the-value-cluster-cheaper-better-faster.html Clusterworld Magazine 2005 AMD Value Cluster Project].

'''Topics:'''

[[Frequently Asked Questions]]
[[Limulus Design]]
[[Limulus Hardware]]
[[Limulus Software]]
[[The Norbert Limulus Cluster]]
[[The Commercial Limulus Cluster]]
[[Limulus Benchmarks]]

'''Project Status'''

The research project has transferred to an open commercial endeavor (Limulus Computing).

The project is was hosted by Basement Supercomputing (now [https://www.limulus-computing.com Limulus Computing]).
'''Limulus Trademarks and Licensing'''

The Limulus Logo and the Limulus Project are trademarks of Limulus Computing. For commercial use of project trademarks or sponsorships, contact Limulus Computing.

Software used by the project is subject to license terms and copyrights as per the authors. Please consult individual packages for more information.

Local Desk-side Clusters

2026-09-28T21:30:16Z

Deadline:

The Limulus project is focused on desk-side computational clusters for both HPC and Data Analytics. '''Limulus''' is an acronym for LInux MULti-core Unified Supercomputer.

* [[Limulus Project]]
* [https://www.limulus-computing.com/ Limulus Computing]
* [https://www.limulus-computing.com/Limulus-Manual/doku.php Limulus Computing Systems Manual]

Limulus Project

2026-09-28T21:28:22Z

Deadline: new

Information on the Limulus project is mostly for historical value. Current open development is by [https://www.limulus-computing.com Limulus Computing] (consult the [https://www.limulus-computing.com/Limulus-Manual/doku.php Systems Manual]

'''What is the Limulus Project?'''

The Limulus project goal is to create and maintain an open specification and software stack for a personal workstation cluster. As "edge of cloud" computing becomes more important, these systems are designed to offer HPC, Big Data (Hadoop-Spark), and Deep Learning in almost any location. Ideally, a user should be able to build or purchase a small personal workstation cluster using the Limulus reference design and low cost hardware. In addition, a freely available turn-key Linux based software stack will be created and maintained for use on the Limulus design. A Limulus is intended to be a local workstation cluster platform where users can develop software, test ideas, run edge applications, and teach HPC methods. Consult the following draft of a Linux Magazine Limulus Article (pdf) for more information. The Limulus idea actually came from the 2005 AMD Value Cluster Project.

'''Topics:'''

[[Frequently Asked Questions]]
[[Limulus Design]]
[[Limulus Hardware]]
[[Limulus Software]]
[[The Norbert Limulus Cluster]]
[[The Commercial Limulus Cluster]]
[[Limulus Benchmarks]]

'''Project Status'''

The research project has transferred to an open commercial endevour (Limulus Computing).

The project is hosted by Basement Supercomputing.
'''Limulus Trademarks and Licensing'''

The Limulus Logo and the Limulus Project are trademarks of Limulus Computing. For commercial use of project trademarks or sponsorships, contact Limulus Computing.

Software used by the project is subject to license terms and copyrights as per the authors. Please consult individual packages for more information.

Local Desk-side Clusters

2026-09-28T21:17:55Z

Deadline: added links

The Limulus project is focused on desk-side computational clusters for both HPC and Data Analytics
* [[Limulus Project]]
* [https://www.limulus-computing.com/ Limulus Computing]
* [https://www.limulus-computing.com/Limulus-Manual/doku.php Limulus Computing Systems Manual]

Local Desk-side Clusters

2026-09-28T21:11:33Z

Deadline: new

Limulus Project
Limulus Systems

Cluster Topics

2026-09-28T21:10:32Z

Deadline: added Local Desk-side Clusters

==Main Categories==

'''Background'''
* [[Cluster Concepts]]
* [[Cluster Questions]]
* [[Cluster Design]]
* [[Cluster How-To's]]

'''Applications/Programming'''
* [[Cluster Application Areas/Markets]]
* [[Open/Freely Available Cluster Applications]]
* [[Commercial Cluster Applications]]
* [[Programming Tools]]

'''Benchmarking'''
* [[Cluster Benchmarking Packages]]
* [[Benchmarking Results]]

'''Cluster Plumbing'''
* [[Hardware]]
* [[System Software]]
* [[File Systems]]

'''Other Topics'''
* [[Local Desk-side Clusters]]
* [[Surveys Market Data]]
* [[Grid Computing]]
* [[Services/Consulting/Assistance]]
* [[Resources]]

Cluster How-To's

2026-09-28T21:07:51Z

Deadline:

* [[Passwordless SSH (and RSH) Logins]]
* [[Tuning Intel e1000 NICs]]
* [[Limiting MPI (1.2x) traffic to one NIC]]

Cluster Design

2026-09-28T21:06:23Z

Deadline:

Clusters are all about design. Indeed, clusters are about designing machines
around problems. When designing a cluster the following '''basic''' questions
may be helpful. If you need help with answers, try asking on the [http://www.beowulf.org/mailman/listinfo/beowulf Beowulf Mailing List] or submit a question to the [[Cluster Questions]] section.

'''Consider your budget'''
* Do you need help with software and support? If so, reserve some funds for this part of the project. While clusters can be easy to set up initially, continued operation can be time consuming.
* Do you plan on assembling the cluster yourself? If yes, at what level? Starting from motherboards, cases, memory, processors, etc. can be very time consuming. Buy quality parts.
* If you plan on using a vendor, provide them with a detailed specification and don't forget to run benchmarks.

'''How many and what type of processors do I need?'''
* This depends on your application. How well does you application scale. Have you tested it on a real cluster? If it is a common application, have other run it on a cluster?
*It also depends on how you want to arrange your processors. The introduction dual-core CPUs make this a more interesting question.

'''What Type of Interconnect?'''
*Again, this depends on your application. If you can use Gigabit Ethernet, then you can probably buy more nodes. If you need a faster interconnect, then you may have to sacrifice nodes for interconnects to stay in your budget. Benchmarking your code(s) is important.
* Remember Gigabit Ethernet performance will vary by chip-set, some are better than others.

'''What type of software?'''
* This depends on your needs. There are several preconfigured distributions, but which one is best depends upon your needs.

'''Do I need disk drives on my nodes?'''
* This decision also depends on on your needs. In some cases nodes can be run without hard drives, in others local swap or scratch space is needed.

'''What type of storage?'''
* do you need a parallel file system?
* Will a basic NFS setup be enough? Will it scale to the nodes you need?
* How do I construct a diskless cluster?
** Can I circumvent the local swap dilemma?

There are many other questions that need to be answered, but these should give you a start. The book below (freely available) is also a good place to start.

'''Books'''

* [http://www.phy.duke.edu/~rgb/Beowulf/beowulf_book.php Engineering a Beowulf-Style Compute Cluster by Robert G. Brown ] - A freely available book on building/designing Beowulf style clusters.

* See the Cluster Monkey [https://www.clustermonkey.net/links-mainmenu-23/80-books.html Cluster Books Page ]

'''Tools'''

* [http://aggregate.org/CDR/ The Cluster Design Rules ] - A tool to help design a cluster tuned for an application

Surveys Market Data

2026-09-28T21:04:34Z

Deadline:

Professional survey organizations (often offer free reports):

* [https://hyperionresearch.com/ Hyperion Research]

* [http://www.intersect360.com/ Intersect360 Research]

There are also links to informal surveys and free marketing reports on ClusterMonkey.

* [https://www.clustermonkey.net/links-mainmenu-23/84-cluster-hpc-polls-and-surveys.html Cluster/HPC Polls and Surveys ]

Surveys Market Data

2026-09-28T21:01:58Z

Deadline:

Professioanl survey organizations:

* [http://www.idc.com/research/viewfactsheet.jsp?containerId=IDC_P7&sectionId=null&elementId=null&pageType=SYNOPSIS IDC]

* [http://www.intersect360.com/ Intersect360]

There are also links to informal surveys and free marketing reports on ClusterMonkey.

* [https://www.clustermonkey.net/links-mainmenu-23/84-cluster-hpc-polls-and-surveys.html Cluster/HPC Polls and Surveys ]

Surveys Market Data

2026-09-28T21:01:27Z

Deadline:

Professioanl survey organizations:

* [http://www.idc.com/research/viewfactsheet.jsp?containerId=IDC_P7&sectionId=null&elementId=null&pageType=SYNOPSIS IDC]

* [http://www.intersect360.com/ InterSect360]

There are also links to informal surveys and free marketing reports on ClusterMonkey.

* [https://www.clustermonkey.net/links-mainmenu-23/84-cluster-hpc-polls-and-surveys.html Cluster/HPC Polls and Surveys ]

Essential HPC

2026-09-28T20:32:14Z

Deadline:

__NOTOC__
''Practical High Performance Concepts, Techniques, and Procedures''

This book will be based on CDP wiki content and is intended to help answer questions and get your HPC projects moving forward quickly. The book is expected to be available in the near future. A soft version will be available at no charge and a special limited edition hard copy will be printed "on demand." We are looking for sponsors. (more details soon)

==Tentative Table of Contents==
[[File:E-HPC-cover2.png|right|alt=Book Cover|400x618px]]
=== Fundamental Concepts ===
*What is HPC and Why Does It Matter
*Introduction To Building Blocks
*Setting Expectations
*Parallel and Distributed Computing 101
*What Do I Need to Know As An Administrator?
*What Do I need to Know As A User?
===Taking The First Steps===
*It all Depends
*Understanding Costs
*Success Metrics
*Choosing Hardware
*Choosing Software
*Using A Cloud
===Techniques===
*Using an Open Approach To HPC
*Designing Your Solution
*Cluster Provisioning and Control
*Sharing The Resource
*Parallel Programming
===Procedures===
*Software Management
*File Systems
*Testing and Tuning
*Monitoring
===Next Steps===
*HPC Applications
*Other Resources

Main Page

2026-09-28T20:31:24Z

Deadline: fixed links

__NOTOC__
== Cluster Documentation Project™ ==
[[File:cdp_logo_1.png|left|250x250px]]
Looking for Cluster Documentation? Jump right to [[Cluster Topics]].

Computational Clusters have grown from a back room curiosity to a driving force in the HPC community. The roll your own nature of cluster computing has also created a wide community of users, vendors, and applications. The Cluster Documentation Project (CDP) is an effort to help document best practices, tutorials, projects, benchmarks, products, and anything else that helps facilitate HPC cluster computing.

'''We plan on releasing our first book at in the near future ''' -- ''[[Essential HPC]]: Practical high performance concepts, techniques, and procedures''

'''VENDORS DON'T MISS A UNIQUE AND EFFECTIVE SPONSORSHIP OPPORTUNITY: [[Funding The Cluster Documentation Project]]'''

[[File:E-HPC-cover2.png|thumb|alt=Book Cover.|Contents [[Essential HPC]]]]

===Project Scope===

The word [http://en.wikipedia.org/wiki/Computer_cluster cluster] has many meanings. The Cluster Documentation Project is designed to document and professionally publish High Performance Computing (HPC) Cluster information. The focus will be on tutorial and how-to information rather than theoretical aspects of HPC. Currently there are [[Cluster Topics]] which is a compendium of current benchmarks, products, technologies, methods, best practices, etc. that will help those interested in clustering. When funded the project will enhance this resource.

See an [http://www.admin-magazine.com/HPC/Articles/Cluster-Documentation-Project/%28language%29/eng-US article] in HPC Admin magazine for more background.

Background: [[Beowulf Cluster History Project]]

===Writers and Contributors===

Contributions are accepted from the community -- both users and vendors. All content is subject to approval by the editors. If you are interested in contributing please contact [https://www.clustermonkey.net/contact-mainmenu-3/53-head-monkey/1-douglas-eadline.html Douglas Eadline]. All submitted content will be made available under a [http://creativecommons.org/licenses/by-nc-sa/2.5/ Creative Commons License].

===Questions===
If you have specific questions about clusters please visit our [[Cluster Questions]] section and enter your question.

If you have questions about this site, please contact [http://www.clustermonkey.net//component/option,com_contact/task,view/contact_id,1/Itemid,3/ Douglas Eadline] with any questions.

===History===
ClusterWorld magazine was the initial sponsor of a site called The Cluster Agenda. When ClusterWorld Magazine ceased publication the Agenda effort was stopped. Douglas Eadline, editor of ClusterWorld magazine, was the principle organizer of the Agenda effort and has decided with the help of Jeff Layton to preserve some of the content from the Agenda and morph the site into Cluster Tweaks. The original Agenda Working Committee was Jeff Layton, Glen Otero, Dan Stanzione, and Douglas Eadline.

Beowulf Cluster History Project

2026-09-28T20:20:00Z

Deadline: /* The Beowulf Cluster History Project */ added link to video

==The Beowulf Cluster History Project ==

For a history of the Beowulf Project (with interviews of the original participants) see the [https://www.youtube.com/watch?v=P-epcSlAFvI The Beowulf Disruption: Changing the Course of HPC] video.

User:Deadline

2026-09-05T14:36:10Z

Deadline: Blanked the page

CAE/CAD/EDA Modeling

2018-09-11T12:41:48Z

Deadline: added FreeCAD and OpenSCAD

* [http://sourceforge.net/projects/brlcad/ BRL-CAD] is a powerful cross-platform constructive solid geometry solid modeling system that includes an interactive geometry editor, ray-tracing for rendering & geometric analyses, network distributed framebuffer support, image & signal-processing tools.

* [http://www.salome-platform.org/ SALOME] is an open-source software that provides a generic platform for Pre- and Post-Processing for numerical simulation. It is based on an open and flexible architecture made of reusable components. SALOME can be used as standalone application for generation of CAD models, their preparation for numerical calculations and post-processing of the calculation results. SALOME can also be used as a platform for integration of the external third-party numerical codes to produce a new application for the full life-cycle management of CAD models.

* [http://www.code-aster.org/V2/spip.php?rubrique2 Code_Aster] is an Open Source software package for Civil and Structural Engineering finite element analysis and numeric simulation in structural mechanics originally developed as an in-house application by the French company EDF. This tool covers a large range of applications : 3D thermal analyses and mechanical analyses in linear and non-linear statics and dynamics, for machines, pressure vessels and civil engineering structures. Beyond the standard functionalities of a FEM software for solid mechanics, Code_Aster compiles specific research in various fields : fatigue, damage, fracture, contact, geomaterials, porous media, multi-physics coupling. It is widely used at EDF for the expertise and the maintenance of power plants and electrical networks.

* [http://kicad-pcb.org KiCad EDA] is a free software suite for electronic design automation (EDA). It facilitates the design of schematics for electronic circuits and their conversion to PCB designs. KiCad was originally developed by Jean-Pierre Charras. It features an integrated environment for schematic capture and PCB layout design. Tools exist within the package to create a bill of materials, artwork, Gerber files, and 3D views of the PCB and its components.

* [http://www.wings3d.com/ Wings 3D] is an advanced subdivision modeler that is both powerful and easy to use. Wings 3D offers a wide range of modeling tools, a customizable interface, support for lights and materials, and a built-in AutoUV mapping facility. There is no support in Wings for animation. Originally inspired by Nendo and Mirai from Izware, Wings 3D has been developed since 2001, when Björn Gustavsson (bjorng) and Dan Gudmundsson (dgud) first started the project. Richard Jones (optigon) maintained Wings and coded many new features between 2006 and 2011. Wings 3D is currently maintained by Dan and Richard with the help of the great community.

* [https://www.freecadweb.org FreeCAD] is a free and open-source (under the LGPLv2+ license) general-purpose parametric 3D CAD modeler and a building information modeling (BIM) software with finite-element-method (FEM) support. FreeCAD is aimed directly at mechanical engineering product design but also expands to a wider range of uses around engineering, such as architecture or electrical engineering. FreeCAD can be used interactively, or its functionality can be accessed and extended using the Python programming language and is currently in a beta stage of development.

* [http://www.openscad.org OpenSCAD] is a free software application for creating solid 3D CAD (computer-aided design) objects. It is a script-only based modeller that uses its own description language; parts can be previewed, but it cannot be interactively selected or modified by mouse in the 3D view. An OpenSCAD script specifies geometric primitives (such as spheres, boxes, cylinders, etc.) and defines how they are modified and combined (for instance by intersection, difference, envelope combination and Minkowski sums) to render a 3D model. As such, the program does constructive solid geometry (CSG). OpenSCAD is available for Windows, Linux and OS X.

CAE/CAD/EDA Modeling

2018-09-11T12:34:04Z

Deadline: added wings 3d

CAE/CAD Modeling

2018-09-11T12:30:08Z

Deadline: Deadline moved page CAE/CAD Modeling to CAE/CAD/EDA Modeling

#REDIRECT [[CAE/CAD/EDA Modeling]]

CAE/CAD/EDA Modeling

2018-09-11T12:30:07Z

Deadline: Deadline moved page CAE/CAD Modeling to CAE/CAD/EDA Modeling

Open/Freely Available Cluster Applications

2018-09-11T12:28:41Z

Deadline: added EDA

Below are the various categories of Cluster HPC applications. Please note, most of these applications are parallel (i.e. MPI applications).

You may also find other applications from [http://en.wikipedia.org/wiki/List_of_free_and_open_source_software_packages List of Free and Open Software Packages ]

* [[Aerospace]]
* [[Bioinformatics]]
* [[CAE/CAD/EDA Modeling]]
* [[Cosmology]]
* [[Computational Fluid Dynamics]]
* [[Database]]
* [[Electronic Structure/Quantum Chemistry]]
* [[Finite Element Analysis]]
* [[Molecular Modeling]]
* [[Oil and Gas]]
* [[Visualization/Rendering]]
* [[Weather Modeling]]

* [[Other]]

CAE/CAD/EDA Modeling

2018-09-11T12:27:53Z

Deadline: added kicad

Statcounter-project-id

2018-08-17T16:26:20Z

Deadline: Created page with "1303510"

1303510

Other

2017-08-05T16:24:30Z

Deadline: updated caos link

'''Genetic Algorithms'''

* [http://ftp.mcs.anl.gov/pub/pgapack/ PGAPack] (Parallel Genetic Algorithm Package) a general-purpose, data-structure-neutral, parallel genetic algorithm library developed at Argonne National Laboratory.
* [http://www.cisl.ucar.edu/css/staff/travis/mpikaia/ MPIKAIA] is a slightly modified version of the PIKAIA 1.2 code developed by Paul Charbonneau and Barry Knapp at the High Altitude Observatory. It uses MPI code to parallelize the "Full Generational Replacement" evolutionary strategy. This package comes with a sample fitness function, which you should replace with your own function/model after you successfully compile and run the sample code.
* [http://gpocl.sourceforge.net/ GPOLC] (Genetic Programming in OpenCL) is a parallel implementation of genetic programming targeted at heterogeneous devices, such as CPU and GPU. It is written in OpenCL, an open standard for portable parallel programming across many computing platforms.

'''Cellular Automata'''
* [https://staff.fnwi.uva.nl/c.u.grelck/caos.html CAOS] (Cells, Agents and Observers for Simulation) is a high-level domain-specific programming language for the definition of extended cellular automata and their parallel simulation on shared and distributed memory computing architectures.
* [http://code.google.com/p/reaction-diffusion/ Ready] is a program for exploring continuous cellular automata, including reaction-diffusion systems, on grids and arbitrary meshes. OpenCL is used as the computation engine, to take advantage of the many-core architectures on graphics cards and modern CPUs. OpenCL also allows rules to be written in a text format and compiled on the fly. Ready supports a compact XML-based file format so that images and rules can be shared easily.

'''Scientific Data Mining'''
* [http://www.nutonian.com/ Eureqa II] is a scientific data mining software package that searches for mathematical patterns hidden in your data. Free download for a single computer.

Programming Tools

2017-03-17T18:28:07Z

Deadline: added Linux performance page

==Higher Level==
*[http://www.sagemath.org Sage] is a free open-source mathematics software system licensed under the GPL. It combines the power of many existing open-source packages into a common Python-based interface. The Sage Mission is to create a viable free open source alternative to Magma, Maple, Mathematica and Matlab.
*[http://numpy.scipy.org/ NumPy] is the fundamental package for scientific computing with Python. It contains among other things:
**a powerful N-dimensional array object
**sophisticated (broadcasting) functions
**tools for integrating C/C++ and Fortran code
**useful linear algebra, Fourier transform, and random number capabilities<br>Besides its obvious scientific uses, NumPy can also be used as an efficient multi-dimensional container of generic data. Arbitrary data-types can be defined. This allows NumPy to seamlessly and speedily integrate with a wide variety of databases. Numpy is licensed under the BSD license, enabling reuse with few restrictions.
*[http://www.r-project.org/ R] is a language and environment for statistical computing and graphics. R provides a wide variety of statistical (linear and nonlinear modeling, classical statistical tests, time-series analysis, classification, clustering, ...) and graphical techniques, and is highly extensible. One of R's strengths is the ease with which well-designed publication-quality plots can be produced, including mathematical symbols and formulae where needed. Great care has been taken over the defaults for the minor design choices in graphics, but the user retains full control.
*[http://julialang.org/ Julia] is a high-level, high-performance dynamic programming language for technical computing, with syntax that is familiar to users of other technical computing environments. It provides a sophisticated compiler, distributed parallel execution, numerical accuracy, and an extensive mathematical function library. The library, mostly written in Julia itself, also integrates mature, best-of-breed C and Fortran libraries for linear algebra, random number generation, FFTs, and string processing. More libraries continue to be added over time. Julia programs are organized around defining functions, and overloading them for different combinations of argument types (which can also be user-defined).
*[http://www.erlang.org/ Erlang] Erlang is a programming language used to build massively scalable soft real-time systems with requirements on high availability. Some of its uses are in telecoms, banking, e-commerce, computer telephony and instant messaging. Erlang's runtime system has built-in support for concurrency, distribution and fault tolerance.
*[http://www.haskell.org/haskellwiki/Haskell Haskell] is an advanced purely-functional programming language. An open-source product of more than twenty years of cutting-edge research, it allows rapid development of robust, concise, correct software. With strong support for integration with other languages, built-in concurrency and parallelism, debuggers, profilers, rich libraries and an active community, Haskell makes it easier to produce flexible, maintainable, high-quality software.

== Compiler Enhancements/Code Conversion==

These enhancements are used with Fortran and C/C++ compilers.

*[http://openmp.org/wp/ OpenMP] is a standard for parallel programming on shared memory systems, continues to extend its reach beyond pure HPC to include embedded systems, multicore and real time systems. A new version is being developed that will include support for accelerators, error handling, thread affinity, tasking extensions and Fortran 2003. Note: OpenMP is not a cluster programming tool. It works for multi-core cluster nodes and is supported by virtually all compilers.
*[http://www.openacc-standard.org/ OpenACC] is an Application Program Interface (API) that describes a collection of compiler directives to specify loops and regions of code in standard C, C++ and Fortran to be offloaded from a host CPU to an attached accelerator (e.g. GPUs), providing portability across operating systems, host CPUs and accelerators.
*[http://people.nas.nasa.gov/~hjin/CAPO/ CAPO ] (Computer-Aided Parallelizer and Optimizer) automates the insertion of compiler directives to facilitate parallel processing on shared memory parallel (SMP) machines. While CAPO is currently integrated seamlessly into CAPTools (developed at the University of Greenwich, now marketed as ParaWise) CAPO is independently developed at NASA Ames Research Center as one of the components for the Legacy Code Modernization (LCM) project. The current version takes serial FORTRAN programs, performs interprocedual data dependence analysis, and generates OpenMP directives. Due to the widely support of the OpenMP standard, the generated OpenMP codes can potentially run on a wide range of SMP machines.

== Lower Level Parallel Programming Libraries ==

These are programming libraries that can be used with Fortran, C/C++, and Java.

*[http://www.mcs.anl.gov/research/projects/mpich2/ MPICH2] is a freely available, portable implementation of MPI, the Standard for message-passing libraries.
*[http://mvapich.cse.ohio-state.edu/ MVAPICH2] enhanced MPICH2 version that delivers best performance, scalability and fault tolerance for high-end computing systems and servers using InfiniBand, 10GigE/iWARP and RoCE networking technologies.
*[http://www.open-mpi.org/ Open MPI] is a project combining technologies and resources from several other projects (FT-MPI, LA-MPI, LAM/MPI, and PACX-MPI) in order to build the best MPI library available. Support runtime selection of interconnect.
*[http://www.jppf.org/ The Java Parallel Processing Framework] is a suite of software libraries and tools providing convenient ways to parallelize CPU-intensive processing. It is written in the Java programming language and is platform independent.
*[http://www.csm.ornl.gov/pvm/pvm_home.html PVM] (Parallel Virtual Machine) is a software package that permits a heterogeneous collection of Unix and/or Windows computers hooked together by a network to be used as a single large parallel computer.

==Profilers Analyzers==
*[http://www.mcs.anl.gov/research/projects/perfvis/software/viewers/index.htm Jumpshot] is a Java-based visualization tool for doing postmortem performance MPICH2 analysis.
*[http://mpip.sourceforge.net/ mpiP]is a lightweight profiling library for MPI applications. Because it only collects statistical information about MPI functions, mpiP generates considerably less overhead and much less data than tracing tools. All the information captured by mpiP is task-local. It only uses communication during report generation, typically at the end of the experiment, to merge results from all of the tasks into one output file.
*[http://www.openspeedshop.org/wp/ Open|SpeedShop] is explicitly designed with usability in mind and is for application developers and computer scientists. The base functionality include, Sampling Experiments, Support for Callstack Analysis, Hardware Performance Counters, MPI Profiling and Tracing,I/O Profiling and Tracing, and Floating Point Exception Analysis. In addition, Open|SpeedShop is designed to be modular and extensible. It supports several levels of plug-ins which allow users to add their own performance experiments.
*[http://developer.amd.com/tools/codeanalyst/pages/default.aspx AMD CodeAnalyst] Performance Analyzer helps software developers to improve the performance of applications, drivers and system software. Well-tuned software delivers a better end-user experience through shorter response time, increased throughput and better resource utilization.
*[http://ipm-hpc.sourceforge.net/ IPM] is a portable profiling infrastructure for parallel codes. It provides a low-overhead performance profile of the performance aspects and resource utilization in a parallel program. Communication, computation, and IO are the primary focus. While the design scope targets production computing in HPC centers, IPM has found use in application development, performance debugging and parallel computing education.
*[http://hpctoolkit.org/index.html HPCToolkit] is an integrated suite of tools for measurement and analysis of program performance on computers ranging from multicore desktop systems to the nation's largest supercomputers. By using statistical sampling of timers and hardware performance counters, HPCToolkit collects accurate measurements of a program's work, resource consumption, and inefficiency and attributes them to the full calling context in which they occur. HPCToolkit works with multilingual, fully optimized applications that are statically or dynamically linked. Since HPCToolkit uses sampling, measurement has low overhead (1-5%) and scales to large parallel systems. HPCToolkit's presentation tools enable rapid analysis of a program's execution costs, inefficiency, and scaling characteristics both within and across nodes of a parallel system. HPCToolkit supports measurement and analysis of serial codes, threaded codes (e.g. pthreads, OpenMP), MPI, and hybrid (MPI+threads) parallel codes.
*[http://www.brendangregg.com/linuxperf.html This Page ] provides a nice map of Linux performance "zones" in the kernel and the tools that are used to analyze them. Includes slide decks on Linux performance.

==Parallel Debuggers==
*[http://padb.pittman.org.uk/ Padb] is a Job Inspection Tool for examining and debugging parallel programs, primarily it simplifies the process of gathering stack traces on compute clusters however it also supports a wide range of other functions. Padb supports a number of parallel environments and it works out-of-the-box on the majority of clusters. It's an open source, non-interactive, command line, script-able tool intended for use by programmers and system administrators alike.

Essential HPC

2017-02-14T19:48:15Z

Deadline:

__NOTOC__
''Practical High Performance Concepts, Techniques, and Procedures''

This book will be based on CDP wiki content and is intended to help answer questions and get your HPC projects moving forward quickly. The book is expected to be available in the near future. A soft version will be available at no charge and a special limited edition hard copy will be printed "on demand." We are looking for sponsors for a SC 2017 version. (more details soon)

==Tentative Table of Contents==
[[File:E-HPC-cover2.png|right|alt=Book Cover|400x618px]]
=== Fundamental Concepts ===
*What is HPC and Why Does It Matter
*Introduction To Building Blocks
*Setting Expectations
*Parallel and Distributed Computing 101
*What Do I Need to Know As An Administrator?
*What Do I need to Know As A User?
===Taking The First Steps===
*It all Depends
*Understanding Costs
*Success Metrics
*Choosing Hardware
*Choosing Software
*Using A Cloud
===Techniques===
*Using an Open Approach To HPC
*Designing Your Solution
*Cluster Provisioning and Control
*Sharing The Resource
*Parallel Programming
===Procedures===
*Software Management
*File Systems
*Testing and Tuning
*Monitoring
===Next Steps===
*HPC Applications
*Other Resources

Funding The Cluster Documentation Project

2017-02-14T19:47:25Z

Deadline: /* Credible and Professional */

==A Unique and Effective Sponsorship Opportunity==

The Cluster Documentation Project (CDP) is seeking corporate and organizational sponsors to help fund additional content. The content areas will cover a wide range of both hardware and software issues and is planned to include basic HPC fundamentals, parallel programming, HPC Cloud, GPU computing, Hadoop and Big Data, software tools and applications, system design and procurement, and more. The project is expected to have a high impact on both users and vendors.

The project will focus on "how to" and essential background information. The resource will be created on-line as a wiki similar to Wikipedia and will be open and shareable to everyone. It will also be possible to create books directly from the wiki content. The entire wiki and any derived books will be available under a Creative Commons license. That is, readers are free to re-use, improve, and republish the information provided they keep the original author(s) attribution and do not charge for the content.

===Wrap a Book Around Your Technology or Service===

Getting your message out to the HPC community and market is important. New and existing HPC users are constantly searching the Internet for up-to-date and accurate information, and the CDP is designed to be the first place users land when they are seeking cluster and HPC insights. You need to be there. In addition to contributing your technical value proposition, the CDP offers both banner ad and logo branding. We can also help you use the CDP to '''create professionally branded manuals, books, and documentation for your customers'''.

Understanding the full impact of an HPC product or service can require deep technical conversations. Companies often
produce high-level white papers to help explain new products and services. The CDP can be used to create short booklets or small books
that help potential customer fully understand your '''technological value proposition''' and give weight to your sales approach.

In order to kick-start the project, we will be releasing our first free book, entitled ''[[Essential HPC]]: Practical high performance concepts, techniques, and procedures.'' We are seeking sponsors for the 2017 edition Please contact [http://www.xandmarketing.com/contact.php Xand Marketing] for more details.

===Credible and Professional===

The principal editor of the CDP will be Douglas Eadline, a practitioner and chronicler of the Linux Cluster HPC revolution. Doug has worked with HPC systems since 1988 and is a co-author of the original Beowulf How-To document. Prior to starting and editing the popular Cluster Monkey web site in 2005, he served as Editor-in-chief for ClusterWorld Magazine. Doug has written numerous articles in both print and web-based publications and was selected by AMD to write the short, but popular, book High Performance Computing for Dummies. He has also co-authored books on Big Data and Hadoop. A team of capable and experienced writers, editors, and artists has been assembled to work with Doug to create a professional resource for the entire community.

The project will be managed by [http://www.xandmarketing.com/ Xand Marketing] and Seagrove LLC, the publisher of [http://www.clustermonkey.net Cluster Monkey]. At this point in time, some basic information is already in place As mentioned, the project will also create books directly from the wiki content (print-on-demand and ebooks). At a minimum one new book will be professional produced each year. Most books will be freely available in Open Document Format (ODF - OpenOffice format), PDF, and as paper books using print on demand. The inaugural book will be entitled ''[[Essential HPC]]: Practical high performance concepts, techniques, and procedures'' and serve as an introduction to HPC.

==The Big Picture==

An open reservoir of documentation will '''help grow the community and market'''. HPC users and vendors are free to use any CDP content for their own projects and products as long as the license terms are maintained. A professional up-to-date documentation resource will bring HPC to new users and markets. Continued support beyond the initial start-up period is expected to come from the production of CDP books and vendor sponsorships.

Beowulf Cluster History Project

2016-11-28T18:35:35Z

Deadline: Protected "Beowulf Cluster History Project" ([Edit=Allow only autoconfirmed users] (indefinite) [Move=Allow only autoconfirmed users] (indefinite))

==The Beowulf Cluster History Project ==

History links, pictures

Beowulf Cluster History Project

2016-11-28T18:32:13Z

Deadline: Initial creation

==The Beowulf Cluster History Project ==

History links, pictures