Neurona@Home - Invite Code!

[Alyx]

Hey guys!

Neurona@Home is a BOINC-based project with the main aim of simulating the behavior of a large assembly of cellular automata neurons connected in a complex network. (although I'm not completely sure what that means, it does sound useful though...). The project is open for business but requires an invite, and I just so happen to have an invite.

The project has very high RAM requirements, between 2-6 GB per WU. Because of this the admin made it invite only to keep people from coming in and causing issues. If you would like to join please send me a PM and I'll reply with the invite code. Since this is a public forum I'd rather not post it directly here. The project admin did say that I could share the invite with my team though.

Please take a look!

http://falua.cesfelipesegundo.com/Neurona/index.php

 

[bryanW1995]

I'd try it out, but I only have 4gb ram. Even my upgcoming build will on by 8gb.

 

[Alyx]

I've got a machine with 12GB and one with 8GB. I'm just limiting it to 1 task at a time. Hopefully that works.

 

[VirtualLarry]

I'm interested. My two main machines have 8GB of RAM each, and barely use 2GB of that.
How would I limit it to 1 WU at a time? WCG/Clean Energy Project seems to do that automatically.

Edit: Ok, got a WU, says 5 minutes estimated time? They don't run very long? Sounds like they take longer to upload/download than run.

At around 5:00 elapsed, the worker was using up 4GB of RAM, and then suddenly it stopped, and freed all of the RAM.
I think it has to do with the preferences, under disk usage, change max memory usage while computer is running, from 50% (4GB in my case), to 75%. I think, not sure yet, that should stop the WU from stalling out.

This new BOINC project is the *perfect* excuse, to go buy a 2500K/2600K machine, so I can fill it with 16GB of RAM.

Damn, it seems that the Neurona WU re-started itself from the beginning, once I changed my memory prefs to 85/85.

Ok, it seems to be holding at 4GB for this WU. The estimate is wrong, it is going on 8:00 and counting.

Edit: Out of WUs for now?

Edit: If you get "Waiting to Run" on a Neurona task, that means that your memory allocation settings are not high enough. I finally set mine to 95% while working, and 99% while idle. I have 8GB of RAM.

Edit: looks like 8GB of RAM is NOT enough to run some of these WUs. I changed my memory prefs to 100%/100%, and my swapfile pref to 90%, and it's still not running this WU that is just sitting there, "Waiting to Run".

 

[Alyx]

Ehh, actually now that I look through their forums, the server limits it to one task per machine. So if you have the extra memory for one task, I don't think you have to worry.

 

[Rudy Toody]

Ehh, actually now that I look through their forums, the server limits it to one task per machine. So if you have the extra memory for one task, I don't think you have to worry.

I have 16GB on my new rig, so I will try this when I'm done with Correlizer.

Thanks for the info:thumbsup:

 

[VirtualLarry]

I think I'm going to stop Neurona@home. Can't get the WUs to run on my PC. Seems like you really need 16GB to run Neurona?

Hmm, maybe not. I had to manually suspend all of my other tasks, and then Neurona started running again.

 

[Sunny129]

I think I'm going to stop Neurona@home. Can't get the WUs to run on my PC. Seems like you really need 16GB to run Neurona?

Hmm, maybe not. I had to manually suspend all of my other tasks, and then Neurona started running again.

given that you've already maxed out your memory and disk usage preferences, it sounds like maybe you have a resource sharing issue between projects...

i run Test4Theory@Home, which is also a project that only allows 1 task per machine at a time. so on my X6 1090T machine, a T4T@H task only consumes ~16.7% (1 core) of the CPU. at first i thought T4T@H had to be allocated a majority of all project resources in order to make sure the project would run all the time. but b/c of BOINC's quirky resource sharing and task priority algorithms (or possibly my poor understanding of them), T4T@H seemed to run only sometimes, and tasks of other projects would often run in high priority mode. so i set T4T@H's resource allocation proportional to its max possible CPU consumption, which is again ~16.7% (or 1 core) of my 6-core CPU. i then allocated the remaining ~83.3% of resources among the other projects accordingly. sure enough, ever since made that change, T4T@H has been running pretty much all the time, and tasks from other projects don't typically run in high priority mode anymore, which is nice...

 

[Alyx]

Thanks a million Sunny129, I went through and changed the priority for everything like you described and it seems to be working now. No more "High Priority"!! Before I always gave up on stuff like this because I'm attached to so many projects, so instead a went through and lowered the rest down to 1. I like having them so i know if they get more work.

 

[Sunny129]

Thanks a million Sunny129, I went through and changed the priority for everything like you described and it seems to be working now. No more "High Priority"!! Before I always gave up on stuff like this because I'm attached to so many projects, so instead a went through and lowered the rest down to 1. I like having them so i know if they get more work.

glad i was able to help. keep in mind that you may still see some projects run in high priority mode from time to time. while the changes i made to project resource allocation minimized the number of tasks running in high priority mode and allowed Test4Theory@Home to run almost all the time, occasionally Einstein@Home and/or LHC@Home tasks will run in high priority mode b/c BOINC thinks that my host previously "bit off more than it could chew" with the most recent chunk of tasks that it downloaded. but it never has actually DLed more than it can handle, and they only run in high priority mode for a short while until the queue has been reduced accordingly. i think as the DCF (duration correction factor) adjusts over time, it will converge on a specific value, and BOINC will then calculate the amount of work a host actually needs much more accurately, thus further reducing the time tasks spend in high priority mode due to miscalculations of required work loads.

 

[Rudy Toody]

I was able to get this running this morning. I have it on my 1100T w/16GB and one 940 w/4GB.

I just started the second one to see what will happen. (Just checked, it's using the swap file.)

It kept suspending and starting over so I suspended Correlizer until Neurona was done. I won't do any more on this box unless I double the memory.

 

[VirtualLarry]

has Neurona@home stopped sending WUs? I don't think I've gotten one in a few days.

Edit: Thanks RT. It must just be me. I have Neurona@home set to 25, and PrimeGrid, Correlizer, and WCG set to 100. For a long time, I would just run a single Neurona WU, along with one WCG (Clean Energy Project), and various PG and Correlizer WUs.
Now I don't seem to be getting Neurona@home WUs.

AFAIK, I didn't error on any of them. I have 8GB of RAM, and with a N WU running, total RAM usage was around 6GB.

Edit: I manually clicked on Projects, Neurona, and Update, and I got a WU.

 

[Rudy Toody]

has Neurona@home stopped sending WUs? I don't think I've gotten one in a few days.

I'm still getting them. One at a time.

 

[Rudy Toody]

I'm User of the Day at Neurona@Home! Yippee! Skippee!

 

[VirtualLarry]

Congrats, RT!

I changed my Neurona@Home priority to 100, like my other projects, and now I'm getting a steady stream of WUs.

Current one is using up 6.1GB of RAM. Previous WUs only used like 4GB. If this keeps up, I may not be able to run Neurona@home any more, until I upgrade to SB, and go with either 16 or 32GB of RAM. (32GB of RAM would require moving to Win7 Pro.)

 

[VirtualLarry]

Wow, recent Neurona@Home WU took up 6.7GB of RAM by itself. With my other programs and Windows 7, I was right at my 8GB limit.

If this progression continues, I might have to upgrade to an SB, and 32GB of RAM.

Although, I'm a little bit worried. Apparently, there are still bugs in Z68 chipsets when dealing with 32GB of RAM.

http://forums.anandtech.com/showthread.php?t=2220655

 

[Rudy Toody]

Check your swap file to see if it is growing. When I run Neurona on a 4GB rig, I get a lot of swapping.

On my 16GB rig, I have seen 9.5GB as my largest work unit so far. (No swapping.)

 

[VirtualLarry]

Wow, I think I'm swapping. This last Neurona@home WU is taking up 7.2GB of RAM! My free ram is down to zero.

I think I'm going to stop running Neurona@home WUs, just so that my SSD doesn't wear out. Don't want to go swap-happy for a half an hour at a time.

Will look into Sandy Bridge upgrade, and 32GB of RAM.

Edit: Opening Resource Monitor, it has a commit charge of 7.8GB.

 

[Rudy Toody]

If you only have 4GB - 8GB of memory, it's best for you to get back up on the porch and play with the little dogs!

 

[VirtualLarry]

If you only have 4GB - 8GB of memory, it's best for you to get back up on the porch and play with the little dogs!

LOL. I have my new X6 1045T @ 3.51Ghz rig up and running, with a single GTX460 for now, and 16GB of RAM.

Now I laught at Neurona@home WUs that take 7.8GB of RAM for themselves. Thank goodness for 64-bit OSes!

 

[Rudy Toody]

I seems that they are letting the work run out. No message as to why. So, no work for now.

 

[Rudy Toody]

Work is being issued again.

 

[Assimilator1]

Hey guys!

Neurona@Home is a BOINC-based project with the main aim of simulating the behavior of a large assembly of cellular automata neurons connected in a complex network. (although I'm not completely sure what that means, it does sound useful though...).

Well I know neurons are brain cells so it sounds like they are trying to simulate a small part of the brain! Cool!

Sadly my rig has 'only' 4GB so it doesn't cut it.
Anyone know what the CPU requirements are for Neurona?

 

[VirtualLarry]

Gee, I have 8GB of RAM, and then the WUs start to take all of that, leading to swapping (got to have room for Windows too), so I upgrade to a new rig with 16GB of RAM, and now the WUs are taking... 1.2GB? (Edit: 2.2GB. Still, not nearly 8GB like before.)

 

[blckgrffn]

Gee, I have 8GB of RAM, and then the WUs start to take all of that, leading to swapping (got to have room for Windows too), so I upgrade to a new rig with 16GB of RAM, and now the WUs are taking... 1.2GB? (Edit: 2.2GB. Still, not nearly 8GB like before.)

You are being rewarded! Now you can run multiple WU's?