17th Annual WCG Birthday Challenge by SETI.Germany

Gilthanis

Gilthanis

[H]ard|DCer of the Year - 2014
Joined
Jan 29, 2006
Messages
8,759
Nov 16th - Nov 22nd 2021

The only sub project you should run is MCM. Yes you can bunker. These work units have a 7 day deadline so wait until 7 days before the event and watch that your work units aren't resends. Those will have shorter deadlines.
Last year's thread - https://hardforum.com/threads/wcg-16th-birthday-challenge.2002783/
We are signed up.
Rankings can be tracked here https://www.worldcommunitygrid.org/team/challenge/viewTeamChallenge.do?challengeId=10477 and they only update once per day.

MCM is CPU work only. They should have work for Windows, Macs, Linux, and ARM systems.
 
Last edited:
It will be good to see how the new hosts of WCG can handle a hefty challenge.
 
Reminder: You can start bunkering MCM tasks on Nov 9th 00:00 UTC OR Nov 8th (Mon) 4pm./7pm (PT/ET)

To begin bunker, make sure that your hosts are recognized/validated first by returning some valid tasks (usually 10+) otherwise you won't be able to fill up your work cache to the maximum of 1000 tasks or so.

Task deadline: 7 days on the client side.

Added: For those who just want to contribute to this good cause during the challenge period, please check "Mapping Cancer Markers" project in your WCG project selection when you are logged in.

1636143763864.png
 
Last edited:
Reminder: you can start bunkering MCM project at 4pm/7pm PT/ET today.

Side note: This is a worthwhile medical project to contribute due to its work on mapping cancer. There is a recent publication report on this project update here. I had family members and friends who had passed away from cancer. Cancer is a leading cause of death worldwide.
 
I'm wondering if we want to post about this in Genmay. I agree and like keeping the DC help posts in Genmay to a minimum but if we're going to do it it would be a good idea to get it in early. I know at least once or twice we could have pulled an extra rank or two had we said something a little earlier.
 
Unfortunately I don't have genmay account. It will be great to spread this message around, not just solely for this challenge (though it will be a fun thing to do once in a while) but to help in cancer research. It doesn't cost much to run one rig for a week. On my optimized power 3950x rig, it consumes about 142W off the wall. That would be 142W x 24hrs x 7 days = 23.9kWhr. If you are lucky few who have low electricity rates of 10 cents thereabout, it will only $2.40 for this contribution. I'm actually paying close to 32 cents per kWhr :(
 
I happened upon the thread here, just got the ball rolling. I'll move my machines off raptoreum for the challenge. I think a genmay post is always a good thing. I'll be bringing my 5950x, 5600x, and 2600 to the party. Hoping it will start queueing up more work, I barely have more units than I am working on right now.
 
wareyore said:
Same, I have enough to keep a couple machines busy for a couple days but my main is not downloading any additional work.
Click to expand...
Yeah it's disheartening, the 5600x and 2600 have an ocean of work, but my 5950x they are like naw we don't want that doing any work lol.
 
wareyore said:
Same, I have enough to keep a couple machines busy for a couple days but my main is not downloading any additional work.
Click to expand...
If you just added the host, it takes a few completed work units to list your machine as reliable. Once it does, it should download a larger que of work.
 
Gilthanis said:
If you just added the host, it takes a few completed work units to list your machine as reliable. Once it does, it should download a larger que of work.
Click to expand...
It was fine over the weekend. I was limiting the work to 1/2 day pool for testing. Then I switched projects Sunday/Monday as it completed the que. Switched back to WCG last night and now it won't download more WU's than it's actively working - over 12 hours now.

I even tried turning it off and then on, again....

I fired up a 9900k based system on WCG for the first time yesterday and it has several days worth of work.
 
Gilthanis said:
If you just added the host, it takes a few completed work units to list your machine as reliable. Once it does, it should download a larger que of work.
Click to expand...
Just turned in a dozen units on the 5950x, no new units outside of replacing what was turned in. My client/cpu must be blacklisted for whatever reason. My settings for storing work are the same on all my machines, so it's not that. I'll let it go for the rest of the day, but if it doesn't start queueing up work I will stop it.
Edit: After lunch it queued up a bunch, so I have turned off the network and will let it go now.
 
Last edited:
They also only have so many work units in the que for each sub project at once. There is a high likelihood of large donors sucking up a majority of the work units right now. That doesn't explain why some get work and others don't obviously but just one factor to consider. If you have issues with it not fetching more work, copy/paste the event log info so we can look at what the client is reporting. That might help a little.
 
Gilthanis said:
They also only have so many work units in the que for each sub project at once. There is a high likelihood of large donors sucking up a majority of the work units right now. That doesn't explain why some get work and others don't obviously but just one factor to consider. If you have issues with it not fetching more work, copy/paste the event log info so we can look at what the client is reporting. That might help a little.
Click to expand...
Sorry I edited my comment rather than reporting, I came back to find some work queued. I guess what I don't get, is how these events aren't planned better. When contacting projects for these competitions, only those with a hilariously large amount of available work should be considered. If that isn't possible, perhaps split it amongst a few projects that have similar difficulty/reward. I guess after all these years I expected some of this to be fleshed out better, but I'm unaware of the feasibility or difficulty of this.
 
WCG is usually very good with this. However they are under new management and seem incapable of fixing their bungling of their new GPU projects, so there may be some rough patches this time around.
 
EXT64 said:
WCG is usually very good with this. However they are under new management and seem incapable of fixing their bungling of their new GPU projects, so there may be some rough patches this time around.
Click to expand...
The good news is that the GPU application doesn't apply to this challenge. Only the CPU work units for MCM.
 
I thought I had a good queue of work but my 5950x already chewed through its queue. Is there any way to get more work without uploading everything I have bunkered?
 
You have to start multi-clienting (running multiple boinc clients) in that situation.
 
arnemetis said:
I thought I had a good queue of work but my 5950x already chewed through its queue. Is there any way to get more work without uploading everything I have bunkered?
Click to expand...

I'm in a similar boat. Smaller systems will be fine but my TR's and z840 will be finished within another day or so.
 
I found that boinc clients that I installed just in the last couple days would grab some tasks but certainly not more than a day or two, but clients that I have had installed for quite sometime would grab plenty of work. The I noticed that the estimated time on one older i7 was a couple hours per task, but a newer i7-10700 with a new boinc client install estimated 9+ hours per task. It seems that since the estimated time is so high it thinks it has enough work for 7 days.

To resolve I went to Tools\Run CPU Benchmarks in the Advanced view. Then I left the machine alone for a couple hours and when I came back the new tasks that had been downloaded estimated about 90m. The estimate didn't change for me on tasks that were already downloaded until I allowed Network communication. So I let all the work from the past 24 hours upload, knowing that it would run out of work in a day or two in order to download enough work to keep the CPUs busy until the 16th. : ) I guess the client doesnt run a CPU Benchmark right away on a fresh install?

So if you are having the same issue and want to do a bit of bunkering, you might try:
1. Suspend work on the Project tab
2. Run CPU Benchmarks
3. Restart work and go get a beverage of your choice
4. Come back, allow network communication, and allow your client to update with the project
5. Give the client an hour or so to download new work
6. Disallow network communication and watch your box compute

A little patience with the client as it updates in the background goes a long way. Sometimes it takes a few communication cycles with the project to start fetching lots of tasks. Hope this helps. Happy computing.
 
pututu said:
wareyore , arnemetis , our good buddies at TAAT DC forum created a how-to instructions on multi clients setup. Take a look and if you have any questions, please post here.

P.S. We used to have a collection DC guides but it was removed when the [H] website was updated a few years ago. :(
Click to expand...
I tried the guide, but it doesn't work. I followed the instructions exactly, creating a cc_config.xml because I did not have one there. I made the batch file by copy & pasting the code. I tried running normally and as administrator, and shell or command prompt or batch file. A black window always opens and closes. Seems I can't launch boinc this way, even just the normal instance. I will try metallicafan 's method and just dump the existing units I've done.
 
Can you please share your script? The cc_config one.

Here is my script in Windows to launch the second client:
Code: 
"C:\Program Files\BOINC\boinc.exe" --allow_multiple_clients --redirectio --gui_rpc_port 31422 --dir C:\ProgramData\BOINC2
Looks like my batch file is different from the instructions but please try this.
 
Last edited:
pututu said:
Can you please share your script? The cc_config one.

Here is my script in Windows to launch the second client:
Code: 
"C:\Program Files\BOINC\boinc.exe" --allow_multiple_clients --redirectio --gui_rpc_port 31422 --dir C:\ProgramData\BOINC2
Looks like my batch file is different from the instructions but please try this.
Click to expand...
The one I tried is below:
Code: 
"c:\Program Files\BOINC\boinc.exe" --allow_multiple_clients --redirectio --detach_console --gui_rpc_port 31418 --dir C:\ProgramData\BOINC2
There was an exit at the next line, which is probably why it was closing.

Trying your code, the window just sits there doing nothing. By the way, BOINC2 does exist, and it does have some data that was populated there, so it at least tried writing there.

The client has downloaded some more work units, and while it feels bigger than last time, it still doesn't feel like it will hold me for five days. Oh well, I'm headed to bed now.
 
Well I had to reboot and the client forgot all the queued up units and even the ones I was partially complete on. So I decided to start mucking about with the command prompt, and found the issue. The --redirectio flag is the culprit! Without this, it seems to start running, unfortunately however in a command window instead of opening another gui. This of course results in it also having no configuration data so it falls back on defaults. Is it supposed to open another gui instance?

Edit: While I'm still curious about the above, I have attempted to do this in a ham fisted way now. I renamed the folder in ProgramData, and reinstalled boinc. Now I am starting fresh again and getting new units. I can shut down, and rename the two folders, and I have the completed units waiting in queue again. So when it comes time to drop my bunkered units, I can just shut down the client and switch folders. This is messy for sure, but it looks to be a method that works for now.
 
Last edited:
arnemetis said:
Well I had to reboot and the client forgot all the queued up units and even the ones I was partially complete on. So I decided to start mucking about with the command prompt, and found the issue. The --redirectio flag is the culprit! Without this, it seems to start running, unfortunately however in a command window instead of opening another gui. This of course results in it also having no configuration data so it falls back on defaults. Is it supposed to open another gui instance?

Edit: While I'm still curious about the above, I have attempted to do this in a ham fisted way now. I renamed the folder in ProgramData, and reinstalled boinc. Now I am starting fresh again and getting new units. I can shut down, and rename the two folders, and I have the completed units waiting in queue again. So when it comes time to drop my bunkered units, I can just shut down the client and switch folders. This is messy for sure, but it looks to be a method that works for now.
Click to expand...
You can try to create a Boinc Manager to manage and control the projects, tasks and networking of your new client (BOINC2)
Follow the instruction here assuming you do have Boinc Manager icon on your desktop already
1636647871058.png

1636647784657.png
 
You can also fire up VM's if you get desperate enough.
 
Also in our guides section, we used to have a post that listed the total number of work units a host was allowed to download regardless of cpu's. I'm not sure what that limit is anymore off the top of my head but it is possible you are hitting the server side limit.
 
Gilthanis said:
Also in our guides section, we used to have a post that listed the total number of work units a host was allowed to download regardless of cpu's. I'm not sure what that limit is anymore off the top of my head but it is possible you are hitting the server side limit.
Click to expand...
Yeah I am probably running into that.
Edit: Looking into my other clients, the 5600x is going to run out of work today. The 2600 unfortunately had a bunch of non mcm tasks, I realized that I didn't uncheck the box online that allows it to sub in other work when there are no available units. 8300h laptop will be in a similar boat, will run out of mcm units today. Oh well, looks like my bunker will be small. I always forget that this is the frustration that keeps me out of dc most times lol.
 
Last edited:
arnemetis said:
Yeah I am probably running into that.
Edit: Looking into my other clients, the 5600x is going to run out of work today. The 2600 unfortunately had a bunch of non mcm tasks, I realized that I didn't uncheck the box online that allows it to sub in other work when there are no available units. 8300h laptop will be in a similar boat, will run out of mcm units today. Oh well, looks like my bunker will be small. I always forget that this is the frustration that keeps me out of dc most times lol.
Click to expand...
Well, what's really more important is your contribution towards cancer research during the duration of this challenge. The second thing is to have fun. Don't let the frustration overcome the fun ;)
There always be next year, so you have another year to practice and prepare, lol. I'm sure you will learn something here. I learnt a lot from other hordes here too.
 
1000 units is the max you can download at a time, regardless of core count unfortunately. At least that is what I am capped on my high core count/multi socket systems.

Edit: Some of mine got to 1000, however now it looks like some intermittent WU shortages again.
 
pututu said:
Well, what's really more important is your contribution towards cancer research during the duration of this challenge. The second thing is to have fun. Don't let the frustration overcome the fun ;)
There always be next year, so you have another year to practice and prepare, lol. I'm sure you will learn something here. I learnt a lot from other hordes here too.
Click to expand...

This is why I play around with DC stuff. It lets me play around with hardware, recycle/repurpose gear and learn some stuff.

It's to bad we don't have the DC info here like there was in the past.

I found and followed that guide loosely last weekend when testing and couldn't get it going. I figured out my issue trying again yesterday. Used the wrong area to create the .xml file and create the 2nd Boinc instance.

So, I learned something new this challenge, for sure.
 
wareyore said:
They need a subtotal or count function on the columns.
Click to expand...

I select all and hit abort and it says "are you sure you want to abort X number of work units" and i say no. You can shift click to select the entire range
 
I use BoincTasks on my main PC to monitor all my others - it does all the nice GUI stuff the client should do (but doesn't).
 
We are less than 2 days away now. Get those rigs fired up if you haven't already. MCM work units only. If you are willing to bunker, even a few days can make a huge difference.
 
Gilthanis said:
We are less than 2 days away now. Get those rigs fired up if you haven't already. MCM work units only. If you are willing to bunker, even a few days can make a huge difference.
Click to expand...
Since the competition is only a week long, bunkering is huge especially since work units may not be verified for several days. Even if not bunkering you want to get work units running and turning in now because quite a few of them won't register as points until after the competition starts. One of the reasons for this will be because of so many people bunkering and until many of those bunkered work units are turned in a lot of people will have a lot of work units already turned in sitting in limbo.
 
You must log in or register to reply here.
Back
Top