Distributed Hardware Evolution Project

Associate
Joined
23 May 2018
Posts
7
Hello we'd like to announce to the World's Finest Team that the World's Finest Distributed Computing project is (back) online! ;) ;)

Host an island with a population of circuits struggling for survival in a hostile online world. During your PCs idle time individuals from this population will evolve through artificial evolution in a process of survival of the meekest into circuits with Built-In Self-Test (BIST) and will compete with those hosted on other PCs by migrating to and from them. These circuits will not be constrained by conventional design rules since evolution finds efficient solutions without worrying about how complex they are to understand - just as it did with our own bodies and brains. You can join into this cluster in one minute by installing the client found here. Check up on how your population is doing compared to others here and name your best creations if they enter the "better than human" hall of fame

Read more about us in the New Scientist Magazine https://www.newscientist.com/article/dn4012-circuit-design-evolving-in-distributed-digital-world/

And download our client and get cracking early in the stats here http://dhep.ga/download.php

Cheers,
The DHEP Team
 

PDW

PDW

Associate
Joined
21 Feb 2014
Posts
1,867
This is now trying to get itself set up as a Boinc project at http://dhep.ga/boinc/

Definitely still a work in progress, no work has completed and validated yet.
It downloads and runs Java but it is an old v7 Java and @MichaelDHEP really needs to get it to run the latest Java version.
He is using it to connect to the university website/database and it would be much better to utilise a user's installed Java and just download the parameters needed to run the application.
 
Associate
OP
Joined
23 May 2018
Posts
7
Hello PDW, thanks for your comments. We are indeed trying to get onto BOINC. We are focusing on POSIX at the moment and are struggling with an issue where the (standard BOINC build) wrapper is crashing on some Linux boxes but not others.

As to the JDK, according to Yoyo, shipping the JDK is the standard BOINC way of distributing a Java application so I will be following his advice on this as it is the official BOINC line on how this is done. If you'd like a client that uses the Java installed on your computer you can use this: http://dhep.ga/download.php

As to using JDK7. Well depending on which benchmark you focus on (eg. https://stackoverflow.com/questions...lderappendint-faster-in-java-7-than-in-java-8, https://bugs.openjdk.java.net/browse/JDK-8132651) some will say 7 is faster, some will say 8 is faster - by <1% margins. However the really crucial thing we require is for the server VM to be available, this can make a 30% speed difference. Fortunately on 64 bit systems this will always be available, however on 32-bit machines most will have the client VM JRE installed which is poorly suited for high performance computing; so being able to ship the VM there brings a definite advantage. At the moment - given how things have panned out while testing and setting up BOINC - the POSIX platforms are shipping JRE 8 and Windows JRE 7, but as I say this is all work in progress and we'll be evaluating which version to ship out once it's better cooked and closer to being in prod state.

We will focus on POSIX platforms first (Mac OS is pretty stable we think so if anyone would like to give some feedback on that), and we will focus on Windows last.
 

PDW

PDW

Associate
Joined
21 Feb 2014
Posts
1,867
Hi, thanks for the reply Michael.

I will bow to Yoyo's greater knowledge on Boinc dev stuff but the recently closed dDM project (http://www.distributeddatamining.org/) ran Java locally rather than download and copy its own version in.

My main concerns are more with security (using an older out of date product) and disk IO (copying the java folders in for each work unit).

I know you are just starting to get everything to work under Boinc and understanding how Boinc works.
I see that the jre.zip file is loaded into each work unit's Boinc slot, it is then unzipped into the jre7 folder structure under that slot folder so you can reference the java.exe specifically from there. I'm not a java programmer but I don't really see why if you just want to have java available for the executable to run you can't use what is installed on the machine. Admittedly if java isn't installed then the work units fail, they did for the dDM project but the project managed that so clients didn't just churn through work units endlessly.

I don't understand what the VMs you are referring to are either, my initial comments were just based on the 3 seconds of run time I managed to get from the few work units I got :)
 
Associate
OP
Joined
23 May 2018
Posts
7
Hi PDW, as is clearly stated at www.dhep.ga/boinc BOINC support for DHEP is non-operational. We have not announced it as being ready and I have no idea how you got to connect to it. However I do appreciate your feedback. Is there any security issue in particular you are concerned about? As far as I'm aware running a VM even version 7 with the Java SecurityManager and policy file used is as secure as anything gets. As to the file system efficiency yes I believe there are ways to improve on this. However as I've mentioned before we will be following BOINC official advice on how to distribute a Java application whilst using their platform.

The DHEP Linux BOINC app should be fairly stable now so we would appreciate feedback as to how that is working on various systems.
 

PDW

PDW

Associate
Joined
21 Feb 2014
Posts
1,867
You would be surprised how quickly word gets round about a new project on the horizon :D

I've never owned a Mac, had a stolen one for a few days but that's another story :rolleyes:
I've connected a Ubuntu machine to the project but no work at the moment for it.

I would suggest that you open up the forum at the Boinc project in order to centralise helpful responses from around the world and save you visiting the numerous forums. However be aware that you will get spammers so some way of limiting/stopping this will be needed, the usual way of only allowing posts when you have credit will only work when users can get it. As well as the spam you'll also get some negative feedback so don't take things to heart. You can plaster the main page with as much warning that it is in very early development as you like but you'll still get people complaining that they lost time/credit because something went wrong that wasn't their fault.
 
Associate
OP
Joined
23 May 2018
Posts
7
Stolen PC's are against our general policy ;) (http://www.dhep.ga/download.php)

Thanks for the wise words.

Try your Ubuntu again? Seems like there's so many people connected the workunits get gobbled up instantly. (We've been tweaking the estimated time per task as this seems to resolve an issue with the standard build wrapper on some Linux versions. However this will likely be raised. Just so you get a background of how we intend to run this on BOINC. We will be relying on the trickle system for credit allocation. Tasks sent as file are dummys and don't require validation as progress is sent direclty to the server (see http://dhep.ga/statsgoal.php) ).
 

PDW

PDW

Associate
Joined
21 Feb 2014
Posts
1,867
It had been stolen from a university and dumped, someone was probably coming back to get it when safe. Had to wait for the police to come and collect it, the thief was caught and found guilty so all ended well.

I had come back to post that I now had work, they are running at the moment, guessing at 30 minutes according to the script in the job I looked at.
I've suspended one for a while to see how it likes that, I'll report back when I have anything of interest :)
 

PDW

PDW

Associate
Joined
21 Feb 2014
Posts
1,867
They get gobbled up quickly because their expected run time is 2 seconds, and they had run for 35+ minutes.
I did have 1,018 of them but they have all now gone from the machine and from the account at the Boinc website :confused:
 
Associate
OP
Joined
23 May 2018
Posts
7
Yes as you can see it's all in construction at the moment. We've just confirmed the estimated runtime of the tasks causes a crash on some Linux boxes and are now trying to figure out the optimal task length.
 

PDW

PDW

Associate
Joined
21 Feb 2014
Posts
1,867
The event log shows...
Code:
Thu 14 Jun 2018 15:44:23 BST | Distributed Hardware Evolution Project | Computation for task wkwhzF76xLxLWaqNz_0 finished
Thu 14 Jun 2018 15:44:24 BST | Distributed Hardware Evolution Project | Computation for task woDCRcO1wnzvziWm7_0 finished
Thu 14 Jun 2018 15:44:25 BST | Distributed Hardware Evolution Project | Computation for task wudEUskjc3U5Oq5Oc_0 finished
Thu 14 Jun 2018 15:44:26 BST | Distributed Hardware Evolution Project | Computation for task wFp2ooHdjbGYc5l4a_0 finished
Thu 14 Jun 2018 15:44:28 BST | Distributed Hardware Evolution Project | Computation for task wxDeNqUVRoYyPfEjY_0 finished
Thu 14 Jun 2018 15:44:29 BST | Distributed Hardware Evolution Project | Computation for task wvTuFIAqjuAqrMbyE_0 finished
Thu 14 Jun 2018 15:44:30 BST | Distributed Hardware Evolution Project | Computation for task wQ7hUxNSu35ZHE1oO_0 finished
Thu 14 Jun 2018 15:44:31 BST | Distributed Hardware Evolution Project | Computation for task wXvjZIJI0ynveheUI_0 finished
Thu 14 Jun 2018 15:44:32 BST | Distributed Hardware Evolution Project | Sending scheduler request: To report completed tasks.
Thu 14 Jun 2018 15:44:32 BST | Distributed Hardware Evolution Project | Reporting 1018 completed tasks
Thu 14 Jun 2018 15:44:32 BST | Distributed Hardware Evolution Project | Not requesting tasks: some task is suspended via Manager
Thu 14 Jun 2018 15:44:40 BST | Distributed Hardware Evolution Project | Scheduler request completed
... so everything all got reported even the suspended task and all the work not started yet.
Couldn't get more work because there was a suspended task which had already been reported :confused:
 
Back
Top Bottom