« Architecture » : différence entre les versions

De Assothink Wiki
Aller à la navigation Aller à la recherche
Contenu ajouté Contenu supprimé
(Page créée avec « The 3 possible architectures are: *Java emulation of assothink on 1 machine, 1 process, multiple threads. This is the first tested architecture of assothink. It ran from 20... »)
 
Aucun résumé des modifications
 
(33 versions intermédiaires par le même utilisateur non affichées)
Ligne 1 : Ligne 1 :
The Assothink [[Model|model]] is implemented through various architectures.
The 3 possible architectures are:


The 3 possible architectures for Assothink (active jelly part) are:
*Java emulation of assothink on 1 machine, 1 process, multiple threads. This is the first tested architecture of assothink. It ran from 2010, on the Matscape machines.
*The dream: a dedicated set of millions of microchips, each working autonomously. This might be suggested to some company producing wide scale IC (integrated circuits) products. This would be the most achieved and performing version of Assothink.
*Intermediary, DPDM. Many machines connected in a TCP-IP network, one or many small process per machine. Java emulation.


*The [[Basic Architecture|basic architecture]]
Its is important to note that the passive jelly construction has nothing to do with these 3 architecture. The construction of the passive jelly is a previous step.
*The intermediary architectures (listed below)
*The [[Target architecture|target architecture]]

It is important to note that the passive jelly construction has nothing to do with these 3 architectures. The construction of the passive jelly is a previous step.

This page does not handle the [[UI architecture|UI architecture]] choices.

== Basic architecture ==

Working on equipement available and cheap in 2010:

*standard intel PC
*1 to 4 processors
*1 to 16 Gb ram
*any graphic card
*standard hard disk
*linux ubuntu OS
*apache / tomcat...
*standard browsers (chromium preferred)
*standard JVM (java platforms)

The java emulation of Assothink runs on 1 machine, 1 process, multiple threads.

Available as GWT application working within a browser. Check it at [http://ww.assothink.com www.assothink.com].

This is the first tested architecture of assothink. It ran from 2010, on the Matscape machines.

== Target architecture ==

The hardware suited to adequately run Assothink would be very simple but massively parallel, with millions of basic active components; highly interconnected.

It is named '''IPSE''' ans is described [[Target architecture|here]]. 

== Intermediary architectures ==

This is a pragamatic compromise.

The goal is to maximize computing speed at a reasonale price (this written in 2012).<br>

Various options are considered.

None yet is implemented.

==== Common points ====

In all of these architectures,&nbsp;

*many process and computers contribute to the running app (at least an interactive front-end and one or many computing daemons)&nbsp;
*the control functions and drawing front-end are&nbsp;separated from the computing

==== DPDM architecture ====

Many machines connected in a TCP-IP network, one or many small process per machine.

DPDM stands for "'''D'''istributed '''P'''rocessing/'''D'''ispatching/'''M'''onitoring". <br>

Java emulation, thus java programming.

Describe and discussed at [[DPDM Architecture|DPDM architecture]]&nbsp;.

STATUS&nbsp;: '''<span style="color: rgb(0, 128, 128);">waiting</span>'''.

==== A3 architecture ====

Described and discussed in [[A3 architecture|A3 architecture]] .

STATUS: <span style="color: rgb(0, 128, 128);">'''under development (2012).'''</span>

==== GPGPU architecture ====

Based for instance on Nvidia's CUDA approach, on the computing capacity on graphic processors.

Problem: no java programming, strange programming model, proprietary approach.

STATUS: <span style="color: rgb(0, 128, 128);">'''waiting'''</span>.

==== Parallela architecture ====

Epiphany processors are created by US company Adapteva. There is a project to link 16 or 64 of them in a low cost machine (ubuntu). The project name is parallela.

Constraint: EPM engine&nbsp;programming in C... specific code optimization.

Question: will this project reach its goals&nbsp;?

Here some relevant links:

*[http://www.adapteva.com www.adapteva.com]
*[http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone project summary]
*[http://www.adapteva.com/support/docs/esdk3-manual/ Epiphany SDK]
*[http://www.adapteva.com/support/docs/e3-reference-manual/ Epiphany architecture]

Development step prerequisite: have the active jelly run in a C program.

The parallela architecture should be developed in 2013, when the parallela board will be delivered to Matscape.

STATUS&nbsp;:<span style="color: rgb(0, 128, 128);"> '''promising, but not obvious. Waiting till march 2013.'''</span>

&nbsp;

Dernière version du 15 décembre 2012 à 11:14

The Assothink model is implemented through various architectures.

The 3 possible architectures for Assothink (active jelly part) are:

It is important to note that the passive jelly construction has nothing to do with these 3 architectures. The construction of the passive jelly is a previous step.

This page does not handle the UI architecture choices.

Basic architecture

Working on equipement available and cheap in 2010:

  • standard intel PC
  • 1 to 4 processors
  • 1 to 16 Gb ram
  • any graphic card
  • standard hard disk
  • linux ubuntu OS
  • apache / tomcat...
  • standard browsers (chromium preferred)
  • standard JVM (java platforms)

The java emulation of Assothink runs on 1 machine, 1 process, multiple threads.

Available as GWT application working within a browser. Check it at www.assothink.com.

This is the first tested architecture of assothink. It ran from 2010, on the Matscape machines.

Target architecture

The hardware suited to adequately run Assothink would be very simple but massively parallel, with millions of basic active components; highly interconnected.

It is named IPSE ans is described here

Intermediary architectures

This is a pragamatic compromise.

The goal is to maximize computing speed at a reasonale price (this written in 2012).

Various options are considered.

None yet is implemented.

Common points

In all of these architectures, 

  • many process and computers contribute to the running app (at least an interactive front-end and one or many computing daemons) 
  • the control functions and drawing front-end are separated from the computing

DPDM architecture

Many machines connected in a TCP-IP network, one or many small process per machine.

DPDM stands for "Distributed Processing/Dispatching/Monitoring".

Java emulation, thus java programming.

Describe and discussed at DPDM architecture .

STATUS : waiting.

A3 architecture

Described and discussed in A3 architecture .

STATUS: under development (2012).

GPGPU architecture

Based for instance on Nvidia's CUDA approach, on the computing capacity on graphic processors.

Problem: no java programming, strange programming model, proprietary approach.

STATUS: waiting.

Parallela architecture

Epiphany processors are created by US company Adapteva. There is a project to link 16 or 64 of them in a low cost machine (ubuntu). The project name is parallela.

Constraint: EPM engine programming in C... specific code optimization.

Question: will this project reach its goals ?

Here some relevant links:

Development step prerequisite: have the active jelly run in a C program.

The parallela architecture should be developed in 2013, when the parallela board will be delivered to Matscape.

STATUS : promising, but not obvious. Waiting till march 2013.