2020 Formula Boinc Sprints

[borandi]

I got AVX units straight away fwiw

 

[Endgame124]

I’ve got 2 raspberry pis crunching asteroids right now. I’ll have the 3rd online tonight, but at over a day per wu, it won’t be much help

 

[Ken g6]

I got AVX units straight away fwiw

So did I on the first fetch. Not so much on subsequent fetches.

 

[TennesseeTony]

Keep an eye on your queues, I have seen many very short tasks (4 minutes or less). It probably won't be a problem if you don't plan on bunkering too long after the Sprint starts. (bunkering means to 1.] download as many tasks as you can, 2.] then turn off networking, 3.] letting the tasks run and complete, 4.]only turning networking back on and sending in the results AFTER the contest begins.)

 

[borandi]

Seems to be a substantial time between reporting results and getting the second confirm in. I've had a dual-16 core system doing it for just under a day, and 377 WUs / ~200k points not awarded yet

 

[Endgame124]

Seems to be a substantial time between reporting results and getting the second confirm in. I've had a dual-16 core system doing it for just under a day, and 377 WUs / ~200k points not awarded yet

Good god. It takes a raspberry pi 2 days for 3 work units... I was thinking the points per watt would land in the Pi's favor, but maybe not.

 

[TennesseeTony]

Seems to be a substantial time between reporting results and getting the second confirm in. I've had a dual-16 core system doing it for just under a day, and 377 WUs / ~200k points not awarded yet

I imagine a lot of resources are directed at Rosetta (or other COVID-19) projects at the moment. Therefore 'wingmen' double checking the results are perhaps a bit scarce. At least until the starting bell of the Sprint. I currently have 3189 tasks awaiting validation (3189 x 480 pts = ‭1,530,720‬ points in limbo), it climbs a bit daily, but eventually they will get double checked. If that happens during this Sprint, that would be nice.

This Sprint doesn't officially start for a few more minutes, they announce the project 24 hours early. Those who have temporarily disabled network communications in order to build a bunker, may want to wait an hour after the start before re-enabling comms, making sure that the event host has collected the starting points for each team.

Sprint page is here.

TeAm points are here (updates about 11.30-ish am EST once a day.) I was just about to sneak past Stefan/xii5ku.

 

[StefanR5R]

Sprint page is here.

It is NULL-initialized now, meaning the Formula BOINC site was able to fetch baseline stats, and we have a GO.

TeAm points are here (updates about 11.30-ish am EST once a day.) I was just about to sneak past Stefan/xii5ku.

Well, let's see how that will work out. >:-)

 

[Endgame124]

is there a way to set boinc to use a different number of CPUs at different times? During the day, I would like to use 1 cpu. Over night I would like to use all CPUs. Currently I have to set it to not run at all during the day...

 

[TennesseeTony]

Hiya @Tejas_2 ! Good to see you lurking about, ole friend!

Well we are off to a good start, but we have seen good starts deteriorate in the past, especially with such an easily bunkered project.

 

[TennesseeTony]

is there a way to set boinc to use a different number of CPUs at different times? During the day, I would like to use 1 cpu. Over night I would like to use all CPUs. Currently I have to set it to not run at all during the day...

I am not running the most recent BOINC client, but I don't see an option for changing the percentage of compute power when I select Options/Computing Preferences/Daily Schedules. That is a great idea for the developers to try in incorporate into the next release! (if the newest version doesn't already have it.)

 

[Endgame124]

I am not running the most recent BOINC client, but I don't see an option for changing the percentage of compute power when I select Options/Computing Preferences/Daily Schedules. That is a great idea for the developers to try in incorporate into the next release! (if the newest version doesn't already have it.)

Where does one put suggestions like that? In my case, the home theater pc needs to have minimal usage during the day - mostly to prevent the fans from spinning up and annoying family members. At night, it’s no problem to let it go full tilt.
I was manually changing the number of active CPUs, but I forgot enough before going to bed that it was more productive to let it only compute with all resources at night, than to forget and have it run 25% all the time.

 

[StefanR5R]

is there a way to set boinc to use a different number of CPUs at different times? During the day, I would like to use 1 cpu. Over night I would like to use all CPUs. Currently I have to set it to not run at all during the day...

There is a way, but a convoluted one.

• Set up a second boinc client instance.
  This step has several potential pitfalls; take care.
• Configure one instance to use 1 CPU and to run all day.
• Configure the other instance to use n–1 CPUs and to run only at night, using the Daily schedules tab in boincmgr's computing prefs.

Another way, in other ways convoluted:

• Write a script which periodically edits cc_config.xml and triggers the client to re-read the config.

 

[StefanR5R]

PS,
the number of CPUs which the host has got (from boinc's POV) is set with cc_config.xml, options, n_cpus. (See documentation.)

However, the percentage of CPUs to be used by boinc is set with global_prefs_override.xml, max_ncpus_pct.

boinccmd --read_cc_config and boinccmd --read_global_prefs_override tell the client to re-read these files.

 

[StefanR5R]

Well we are off to a good start, but we have seen good starts deteriorate in the past, especially with such an easily bunkered project.

From the 2nd to 3rd hourly stats update, trends are beginning to look a little more typical:
424 k pph — Gridcoin
300 k pph — The Scottish Boinc Team
214 k pph — XtremeSystems
185 k pph — TeAm AnandTech
112 k pph — LITOMYSL Boinc Team Czech Republic
103 k pph — Czech National Team

 

[Tejas_2]

Hiya @Tejas_2 ! Good to see you lurking about, ole friend!

Hey TT, it has been a while. I see you have been keeping the place lively.

 

[Endgame124]

I now have 3• raspberry pis contributing, watch out team, here I come!

•I may have 3 running, but I can only sustain 1 cpu on one of my pi 3s due to a poor heat sink installation. I used a tiny vga heat sink attached with what appears to be invincible thermal tape. Anyone have a trick to removing thermal tape? Do I need to let it get nice and warm first?

 

[StefanR5R]

Sprint #4 at Asteroids@home — first 12 hours
Top ten teams of league 1:


Median hourly output of the first few:
431 k pph — Gridcoin
267 k pph — XtremeSystems
242 k pph — The Scottish Boinc Team
237 k pph — TeAm AnandTech
120 k pph — LITOMYSL Boinc Team Czech Republic
100 k pph — Czech National Team

A quick check on tasks ready to send, and tasks in progress:

 

[Endgame124]

I now have 3• raspberry pis contributing, watch out team, here I come!

•I may have 3 running, but I can only sustain 1 cpu on one of my pi 3s due to a poor heat sink installation. I used a tiny vga heat sink attached with what appears to be invincible thermal tape. Anyone have a trick to removing thermal tape? Do I need to let it get nice and warm first?

I took the lid off the top of the offending pi 3, and placed it right next to the top exhaust of my host with the 1660 super that is still folding. It gets just enough airflow that it looks like its going to run with all 4 CPUs and stabilize around 62C.

I also found that the thermal tape is much more pliable warm, so if I need to remove the heatsink to upgrade, that won't be an issue.

 

[Orange Kid]

Asteroids@home
Well this time around things seem to be running well, so I will attempt to do some stats.
We are holding on to second at this time after starting out with a bang in the lead.
Maybe we can muster a few more PPD as we go along. Guess I picked a bad time to retire that old X6.



From Free DC

 

[StefanR5R]

XS and TSBT found that our speed is a very good one, and so they adopted it for themselves.


Median hourly output during the day:
442 k pph — Gridcoin
317 k pph — XtremeSystems
251.3 k pph — TeAm AnandTech
250.8 k pph — The Scottish Boinc Team
132 k pph — LITOMYSL Boinc Team Czech Republic
92 k pph — UK BOINC Team

 

[TennesseeTony]

I had hoped my decision pending tasks would decrease, now that extra users are onboard. But it continues to climb. I am quickly approaching 4500 tasks that are waiting for a wingman (2.2M points in limbo).

 

[StefanR5R]

I am quickly approaching 4500 tasks that are waiting for a wingman

I'm already past that. :-P

 

[Endgame124]

Asteroids@home
Well this time around things seem to be running well, so I will attempt to do some stats.
We are holding on to second at this time after starting out with a bang in the lead.
Maybe we can muster a few more PPD as we go along. Guess I picked a bad time to retire that old X6.

View attachment 19747

From Free DC

View attachment 19750

Wahoo! 480 points - go pi go!

 

[pututu]

Go, Johnny-TAAT, Go!

P.S. we are just spectators in this race giving moral support to our friends here