Formula Boinc 2019 Strategy

Yeah... they focus on Einstein anyways so it is win-win for them when he does this as he is going to hit the project heavy anyways.
 
Here is the app_config.xml I am using for SRBase that utilizes 4 threads per WU (supposedly found to be the best for LLR work - I have not verified this, but am running with it). This covers every type of WU that SRBase can send. Adjust the first number (project max concurrent) based on the number of CPU threads you want to use for SRBase. When I run these Floating Point intensive projects like SRBase and PrimeGrid, I use the app_config to effectively disable Hyper-Threading by using only 50% of the total threads available. In this app_config I have it set to 2, because this is the one I use in my 9900K rig. In my 4790K rigs, I use 1. In my TR 1950X/2950X rigs I use 4, in my 2970WX I use 6, and in my 2990WX I use 8.

<app_config>

<project_max_concurrent>2</project_max_concurrent>
<report_results_immediately/>

<app_version>

<app_name>srbase</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase2</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase3</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase4</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase5</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase6</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase7</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase8</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase9</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase10</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase11</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

<app_version>

<app_name>srbase12</app_name>
<cmdline>-t4</cmdline>
<avg_ncpus>4</avg_ncpus>

</app_version>

</app_config>
 
Last edited:
So after running this for 3 days my opinion is it was worse. I am looking at this from a factor of 8 because it runs 8 less work units than stock with HT on. by day 3 i was seeing scores continue to drop and sometimes i would see work units sitting for 6 hours or longer. that would be 48 hours with the factor of 8 and i never saw any units take near that long when running stock.
 
How many threads were you giving each work unit and were these Intel or AMD chips?
 
Were you running the same work unit type or did you have multiple selected? I'm just curious if there is a connection is all. More info is good for everyone for future endeavors. RFGuy_KCCO , have you seen any number drops with your AMD's?

Shamefully, I don't pay attention to mine close enough to notice anything right now even on my Intels
 
Just the usual thought. If you decide to run some projects not yet in the sprint list, maybe consider bunkering and release them when the sprint project is announced. VietOZ keeps hammering on Einstein and he is persistent.

I might test Einstein on RX480 card and see what this card can do. Hopefully to get it on Mon or Tues. Collatz is a good candidate since it has 14 days deadline and you guys were hitting this project hard before the Moo sprint.
 
Well... VietOz does it for multiple reasons. Since XS already hits Einstein hard year round, he has the luxury of potential bunker working for a project he will it anyways. It is actually pretty smart to do if someone has the time on their hands to do it. I wish I did. Between work, kids, and mowing... I've not had more than 10 minutes to play on a pc at any given time. We are still doing conversion work. Got home at midnight Friday night. I have to drive 2.5 hours tomorrow to start converting another location. 4 laptops and 3 desktops, docking stations, and tear down of the old equipment while moving it all over to our network. None of the wall plates are labeled and the lines are so shitty, you have to pull the plug in the patch panel in order for the toner to pick up signal.....
 
We would actually have chickens and maybe a goat if the city didn't have an ordinance against farm animals in town...
 
Back
Top