Here is our situation in the Pentathlon challenge on Day 4
- we are currently #6 overall from 34 BOINC teams (see stats)
As you can see, we had at begin (May 5) a bad start :-(
- (or: the others teams just bunkered too much?)
WCG: in the last 24h we seem in a constant approach to the team 2 places above us (128%)
- will the remaining 43h be enough to catapult us on position 4 in the "City Run" discipline ?
While for Cosmology we seem to lose ground since the last hour (67%)
The problem lately was:
- that the Cosmology@home server was not validating WUs (which seems to work now, btw). But due to that some of our crunchers moved away from Cosmology.
What do you think our strategy should be now?
- "All in": one strategy could be to focus all on WCG to get there 70 points (see below), but: the switching also costs time, so we might lose more in the end?
- Also, we have to consider that due to some WUs being in validation queue the table might actually be different?
- Current standing of points are: 123 (63 Cosmology + 60 WCG)
The points for the positions are like these (see here):
- #1 100 points
- #2 90
- #3 80
- #4 70
- #5 63
- #6 60 points
I've switched to crunching wcg openzika. I found my rig littered with computation errors due to the cosmology downtime, so I lost out on credit and contribution. I might switch to one of the other projects later on though.
Do you think that users within team gridcoin use alternative accounts to crunch for their old teams during competitions like this or that other teams bunker?
Yes, I crunch for my old team and bunker.
Got some problem to add all host from BAM day before start then i need to setup app_config to make Docker and Planck to limit amount of cpus to each task.
When all was running fine i notice that there was only legacy task left so i started to fetch more work of these. Some legacy have been running more then 24 hours and credits to these task different compare to cputime/runtime.
I will keep Docker and Planck running until LHC start and Einstein should be up and running with gpus.
here are some hints by TG13088:
For Cosmology, the "paying" tasks in descending order are Classic, Planck and finally Docker.
Classic is long (11 hrs on i7's) but "pays 125+ credits per.
Planck is about 15 minutes and pays 50 credits while Docker pays the least ~28 credits but is ~10 minutes per
Also, both Docker and Planck are VBox jobs while Classic is straight CPU
In Cosmology, I'm doing both Planck and docker, but mostly Planck as I feel it allows me to do as many wus as quickly as possible. Also I don't have the most powerful machines, so this suits my capabilities better.
The problems with the Cosmology site and validation has I'm sure put off some users, but I suggest we hang in there.
I certainly have quite a number of wus still in validation so I'm expecting a boost when they catch up.
WCG (although I'm not crunching them) work units seem, as I've seen from discussions in IRC, to be suffering slow validation and acquisition as well so I again think that we also need to hang in there.
I think switching at this juncture would be a mistake.
Am I right that both Einstein@home and LHC@home are up soon?
If I'm correct we need to play the other teams game and start bunkering NOW to ensure we have a good start.
It's also possible that some of our new users don't know how to bunker and may need advice on how to approach it.
Thanks for the input, Mercosity!
and yes, Einstein@home starts in 15h, and LHC@Home on May 12
Cranking away on Cosmology here. I7-6700k through grcpool... does that still count?