Formula Boinc 2019 Strategy


[H]ard|DCer of the Year 2018
Oct 3, 2010
Thought I'd better get off my arse and kick off our strategy thread for this year. We're currently running 6th in FB but it's only to be expected prior to the start of the events in which we normally make up ground, the sprints.

I am currently just focusing on personal targets right now and engaging in the ongoing game of cat and mouse with RFGuy_KCCO (dude, please stop buying new kit, we're going to cost each other a fortune :p).

To hopefully assist with our combined project knowledge, below is the latest version project data spreadsheet I first shared early last year. Work demands continue to get in the way of life which means that I haven't been able to update much in the last 3 months so please take the info with an appropriate pinch of salt.

Boinc project data.png

The parts that are probably most out of date are ppd and tasks run times. I'll try and find the time to start bringing them up to date and to start to reflect TR2 and RTX2080 performance.

Edit: To enable the team to download and copy/paste IPs/scripts, I've created a Google spreadsheet link for the above data sheet here. Also, an assortment of Linux update scripts have been added. I've not set a password so please don't share this link outside of this strategy sub-forum.
Last edited:
I am working to convert this to a google sheet so you can copy from it. Also would be nice to add the configs for the projects it suggests custom config
So there was a discussion about the work for amicable running out soon on the current project and we had discussed the strategy of hitting it hard. I am pausing my other goals to switch to that project.
I think we need to look at the current progress on certain project and decide if they are worth it. Myself and a few other have been hitting Einstein for a few weeks and if you look at Hard-DC we are not gaining at all on the Top 8 spots. I think projects like this might be ones to avoid and instead focus on the other projects that the big fish don't run all year such as Seti / WCG etc. If we run our strategy right we can do alot of damage even with a power deficit in certain areas.
Amicable isn't going anywhere. He's going to extend it to the next range of nimbers.
Yes i am aware of that its not going away but might their not be some delay potentially before he is ready to extend to the next range? might make getting what points you can get now important.
That all depends on how long it would take him and when it occurs. If it is merely him spending an hour loading up new work, then the only concern would be if the new work units earn the same PPD that you were getting with old work units. Since he hasn't done this yet in the past, there is no telling without getting him to chime in. Has he gone over any of this in the forums?
FB Strategy? Focus on Sprint! Sprint! Sprint!. Next sprint is Mar 14-17 UTC. XS is pretty good at this.

Outside of sprint, my personal opinion is to focus on personal goals (if you have one), hard to get tasks, projects ending soon (if any) and enjoy the marathon runs which will last another 50 weeks.

In May, prepare for the pentathlon event. Hmm sounds like we are participating in Olympics year round :)

Currently, I'm running Einstein, Gerasim, Rakesearch and playing with monkeys.
On hard to get tasks, try xansons. The first set of 2019 tasks will be available on Jan 19th.

In last year FB, team LAF, TAAT and SG beat us in overall points in this project.

Please post here if there are rare tasks available or connect and set the hard-to-get projects at high resource priority in BM.
1. Set those projects to high priorities
2. Use Phoenicis' scrypt to hammer server for requests
3. possibly have a VM whose sole purpose is to only pull from the hard to get projects (would sit idle most of the year)
4. Low end systems could be set up to catch those as you will probably not keep them busy long term anyways
I am working on personal goals that also help the team, as well as playing cat and mouse with phoenicis. Currently hammering Collatz, Numberfields, Rosetta, ODLK, Latin Squares, Rakesearch, DHEP, and Climate. Will go back to hammering SRBase once the current PG challenge is finished on the 22nd.
Need code for Xansons. Let me know if there are any other projects on this list that need a code, too.

Oh, and reminder for what the update script is; although think I have it somewhere for Windows. Did someone whip one up for Linux, too?

My lovely army of 2p-5118 boxes took a pretty big hit. It was expected, but still sucks. Lost the 4P E7-8867, too, which wasn't expected. With any luck I should be keeping two 4P-8160's in my possession... but nothing is for certain.
I am working to convert this to a google sheet so you can copy from it. Also would be nice to add the configs for the projects it suggests custom config

I've imported into a google spreadsheet and linked to it in the opening post. My comment about using custom configs was just to remind me to think whether an existing app_config was still valid e.g. the number of threads for Amicable changes all the time depending on search stage and card generation

Did someone whip one up for Linux, too?

My lovely army of 2p-5118 boxes took a pretty big hit. It was expected, but still sucks. Lost the 4P E7-8867, too, which wasn't expected. With any luck I should be keeping two 4P-8160's in my possession... but nothing is for certain.

I've added a number of Linux update scripts to Google sheet linked in the opening post.

Sounds like you're still going to make a serious dent in whatever project you decide to victimise next!
Today, I got 333 Xansons tasks processed. Hopefully others got more. I think we can take the top spot in FB for this project. It only takes about 2-3 hours of crunching time every two weeks but since you can only download limited tasks (~20?) per rig, the more rigs attached to this project, the better are our chances. Setting up multiple clients is a PITA since the tasks finished quickly with high-end GPU.
Looks like I got 80 Xansons total. Mostly CPU work from my VM's that are set up to catch work from hard to get projects.
We landed top spot in Xansons FB marathon and moved us to 5th behind XS.. Great participation [H]!

The next available work is scheduled on Feb 2nd.
Remember to run the Xansons script tomorrow, Saturday to get some tasks. All the tasks will be gone in about 1-2 hours or so. I normally run the script around 10am PST time. Project runs on both CPU and GPU but GPU has huge advantage. The project limits like 20+ tasks per rig so quick turnaround is desirable to fill up the work cache to the max. We need to stay on top of this in FB marathon. Good luck!
Nevermind, had to use BAM! and the invitation code x4c-updates
Guys, take a look at the spreadsheet. I added some Windows Batch script download links to projects that I've made/copied batch scripts for on my hosts. If you have one that isn't listed post it here or PM it to me so I can upload the batch script and add it.

I choose to just add download links to download the batch scripts since if I paste them into a single cell they'll make the cells 3 times taller and IMHO, disturb the nice layout.
Remember to run MN Scout script this Sat morning (Feb 16) to defend Xanson and Delilah's position. It only takes a few hours of crunching depending on your setup.

I tested on one rig and the script works fine with 5 clients. Make sure you create multiple BOINC folders first per the instructions in GoofyxGrid, then run the batch file below.

This is my slow starting script for a Windows .bat file:
set _boincproject=""
set _boincmdexelocation="c:\Program Files\BOINC\boinccmd.exe"

set _lowinstance=101
set _highinstance=120

set _timerwait=10

title Multiboinc Slow start loop for %_boincproject%

echo Sending start command to instances %_lowinstance% to %_highinstance% every %_timerwait% seconds
for /l %%a IN (%_lowinstance%,1,%_highinstance%) DO (
start "Boinc_%%a" /MIN "c:\Program Files\BOINC\boinc.exe" -allow_multiple_clients -dir "c:\Multiboinc\BOINC_%%a" -gui_rpc_port 50%%a
timeout /t 20 /nobreak
%_boincmdexelocation% --host --passwd multi --set_gpu_mode always
timeout /t %_timerwait% /nobreak
) (edited)

This is my updating loop script for a Windows .bat file:

set _boincproject=""
set _boincmdexelocation="c:\Program Files\BOINC\boinccmd.exe"

set _lowinstance=101
set _highinstance=120

set _timerwait=30

title Multiboinc Update loop for %_boincproject%

for /l %%x in (1, 1, 10000) do (
echo Update round %%x. Sending update command to instances %_lowinstance% to %_highinstance%
for /l %%a IN (%_lowinstance%,1,%_highinstance%) DO (
%_boincmdexelocation% --host --passwd multi --project %_boincproject% update
timeout /t %_timerwait% /nobreak

Last edited:
Do you recall off the top of your head what the deadlines on their work units are?
Probably 3 days based on someone else system taken last year. I didn't notice it since all the WUs finished in seconds or minutes on GPU. On CPU, I'm not sure how long it will take but I would imagine if you have many clients, there is a good chance that some won't meet the dateline.

Last edited:
OK..just wondering if it would be easier to run a shit ton of clients but have the setting to use like 10% of cpu time so that your host actually runs more but takes like 10 times as long allowing for you to bunker more work...
I only run the Xansons GPU tasks. So I should create 5 folders and run both scripts? This will work ok on my slower 280X cards?
On each rig, create 5 folders or whatever the number of clients that you want to run as long as the tasks are returned in time (3 days I think) and then run MN Scout script once where the batch file is located . Repeat on other rigs.
On a 1080TI there is room for more than 5 instance. 15 to 20 maybe, depends on your computer and what else is running. There is a point where the computer will just crash because too many are running. The run times will elongate to 10+ minutes each depending on how many instances you have going.

Improvements that could be made to this script, would be for it to check how many tasks the instance has, shutdown the instance if the max quantity, then start up another instance. Keep starting more up, keep a list of the ones that are filled up, then go back to them. All doable, but not by me, since I'll be away from the computer for a while hiking. This can be done manually if you are around the computer to issue the commands to the instances.
Xansons result from yesterday. TAAT, Rkraft and XS are ahead of us. We still hold the top spot but we need to continue to allocate ~2hrs of GPU every two weeks to defend our position.

Next one is scheduled on Sat, Mar 30th.

Not sure how much a single older GPU will help but I did recently swap out the ancient one in my main rig for a slightly less outdated 1070 FE card. If it will be useful in upcoming stuff let me know. Also upgraded the CPU's in this box to dual 2667 V3 so I have a little more grunt on the CPU tasks now as well.
don't like WCG as sprint RIGHT now ... more people stealing HSTB on my way to 10yr ... hopefully with those runtimes of HSTB people opt out ... not very sprint-friendly
I see we are in First Place for MOO now...,
I'll keep my 290x's on it until the Summer heatup anyways... (Not a wide selection for ATI cards in boinc...)
Keep in mind that Pentathlon is May. We may need those cards around then.
VietOz bunkered almost 38M in Einstein for this sprint. He keeps trying to guess the next sprint.

Didn't RakeSearch have the same BS server problems last time they were the sprint? They need to be blacklisted from sprints.
Yes they did. If they are given warning in time, they allegedly can roll with the punches. However, they aren't given that courtesy at all. Nice to see your name changed
~46M Einstein bunker by VietOZ. Eventually he will get this right.