EDGES@home

Gilthanis

[H]ard|DCer of the Year - 2014
Joined
Jan 29, 2006
Messages
8,731
I can confirm that the previous thought of the limitations on work units either no longer applies or was only affecting the original tester. I have been able to run up to 32 at a time on a couple 32 thread systems. I did not test under Linux, so there may still be a limitation there.

teamdaily.php
 
Last edited:
Your experience seems to be different than my own. Running only this project, settings set for 1 internet connection/day plus 1 additional day of cache I cannot keep 8 cores occupied without continually mashing the update button.
 
KMac, my experience isn't saying you get a full cache on a single update. My experience is saying that the previous statements of not being able to get more than a specific number regardless of number of cores is debunked. There has been reports that they wont let you download beyond your core count, but I haven't tested that just yet. So, running with a large cache may not be capable.

Are you seeing a limitation to the total number you can pull even after mashing the update multiple times?
 
Multiple mashing results in a max of 2*Core count, but I have only seen that when all current tasks were to complete withing the next 2 minutes. Left on its own, it struggles to keep 8 cores running and I have seen a max of 3 tasks ready to start even with large cache settings.
 
This is the issues currently being reported by others:

1. (known issue) Zeta work does not adhere to any of the suspend options in BOINC.
2. The project seems to be ran at a very low priority. So low that other projects will butt in front of it to run work.
3. Not sure if check pointing works as the work units appear to start over when the system is restarted.
4. Several are having trouble getting work. I don't know if there is just a limited amount of work to go around or if you have set limits server side. Many can not keep their systems loaded with work without manually updating their clients repeatedly.

I would say you are experiencing where the project just isn't keeping up with demand as I have watch a VM run dry that had 32 cores assigned to it for testing the feeding behavior. It sometimes reports back that work isn't available. I am guessing they just aren't capable of handling the demand and work units are being created as they run out.
 
Got it. Seems not to be ready for Vault status if it can't keep itself busy as the only running project. I'm out.
 
That is what I'm thinking too. That is the whole point of this challenge is to test if they can supply such demand. Glad you helped. We still have a few days of the challenge left and I have posted in their forums just in case someone actually takes the time to address any of it.
 
On a single core VM it appears to limit me to 2 tasks. So there may be a limit like that set server side. The event log actually says: (reached limit of 2 tasks)
 
On a single core VM it appears to limit me to 2 tasks. So there may be a limit like that set server side...

I have no issue with that limit, but the project does not keep all available cores supplied with work regardless of settings. I have no desire for a large cache, it was just a final attempt to keep the cores supplied with tasks.

The wind chills expected here tonight are again -25 to -35 degrees below zero F. I want my cores busy, and this project alone cannot seem to do that. WCG is keeping the office warm and without supply issues, so I'm good now.
 
If you want the systems to run a bit warmer, you might want to check out the PrimeGrid challenge currently going. There are only two of us showing up right now. This challenge is for 3 of the sub projects. SOB-LLR, PSP-LLR, ESP-LLR are the three that count for it. Typically PG projects run systems harder and hotter. And they seem to have a ton of work. Warning: If you haven't ran them before, SOB can take a long time to run . ESP are the shorter tasks in this challenge.
 
The Edges homepage is now occupied by a cyber-squatter. It appears the Edges project may be dead.
 
BOINCStats has put this project on the retired list. Their website domain also was not renewed.
 
Back
Top