Any interest in a SiDock challenge at BOINCStats?

Fardringle

Diamond Member
Oct 23, 2000
9,184
753
126
Posted by the SiDock@home project admins:

SiDock@home: SiDock@home September Sailing
Dear participants!

Just now, your computers are hitting three protein targets at once . Due to the low runtimes and (mainly) the friendly contributions of participants and their teams, the data is being processed fast, and we express our gratitude to all of you.
We announce a new competition between teams, SiDock@home September Sailing . Top winners will get the badges:
1st place:
badge_challenge_2021-09_heme_1st_place.png


2nd place:
badge_challenge_2021-09_heme_2nd_place.png


3rd place:
badge_challenge_2021-09_heme_3rd_place.png


Heme B , the well-known ligand in the complex of Hemoglobin , helps to carry oxygen in our bloodstream. As we know today, it has emerged from just two mutations that occurred in genome of one of predecessors of modern vertebrates more than 400 million years ago [1,2].

One of the TAC members would need to sign us up if we want to participate.
 
  • Like
Reactions: crashtech

StefanR5R

Elite Member
Dec 10, 2016
5,459
7,718
136
Posted by the SiDock@home project admins:
Thanks for the heads-up, I haven't noticed it in the boinc client news yet.

Source of the news: https://www.sidock.si/sidock/forum_thread.php?id=146

Challenge specs:

Start time ......... 2021-09-15 00:00 UTC = Tuesday, September 14, 20:00 EDT / 17:00 PDT
End time ........... 2021-09-22 00:00 UTC = Tuesday, September 21, 20:00 EDT / 17:00 PDT
Late entrants allowed? ... Yes

I for one would be interested to participate.

One of the TAC members would need to sign us up if we want to participate.
@Ken g6 <--- poke
 
  • Like
Reactions: Fardringle

Fardringle

Diamond Member
Oct 23, 2000
9,184
753
126
I wasn't planning to run SiDock any more, but I'll put my other projects on hold for a while to contribute what small amounts that I can to the challenge. :)
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
I didn't know about this, and I'm going to have to flush or postpone some pretty long work units in other projects, but I will help.
 

Fardringle

Diamond Member
Oct 23, 2000
9,184
753
126
Isn't SiDock a pain in the rear to bunker? I can't remember.
The last time I ran it, bunkering was only possible by running LOTS of instances, and having each instance download the few tasks that they can get (one per CPU core), then carefully manage the instances to run those tasks either by running the instance and then pausing it, or setting each one to only run one task (or a few) so the work that they received will last longer.
 

cellarnoise

Senior member
Mar 22, 2017
709
394
136
Sign up for this, run that. This team has lots of demands, but I only have one puter running as it is still cooling off here.

I am in, but slowly switching my ding-a-ling over. I miss Chuck Berry. Sorry been reminiscing a lot lately...

SiDock should find a cure for that? :) I should clarify, a cure for "reminiscing" and not my other thing :)
 

Icecold

Golden Member
Nov 15, 2004
1,090
1,008
146
I'm in for the SiDock challenge.

Sign up for this, run that. This team has lots of demands, but I only have one puter running as it is still cooling off here.

I am in, but slowly switching my ding-a-ling over. I miss Chuck Berry. Sorry been reminiscing a lot lately...

SiDock should find a cure for that? :) I should clarify, a cure for "reminiscing" and not my other thing :)
I have to admit, this made me chuckle. I'm going to think every time I run SiDock now that I'm helping to contribute to finding a cure for @cellarnoise 's ding-a-ling.
 
  • Like
Reactions: Fardringle

cellarnoise

Senior member
Mar 22, 2017
709
394
136
I'm in for the SiDock challenge.


I have to admit, this made me chuckle. I'm going to think every time I run SiDock now that I'm helping to contribute to finding a cure for @cellarnoise 's ding-a-ling.
Well this is probably not appropriate for any dirty minds on here, so don't read if you are dirty.. :)

I have had the song "Let my love open the door" stuck in my noggin all day... Love it as I like simple little ditties. "ditties" as in simple but meaning full and supple songs. Sorry :)
 

cellarnoise

Senior member
Mar 22, 2017
709
394
136
On a serious note, how long are these SiDock tasks taking? My first round, may be system tests from Sidock are finishing in around 10 minutes, or else they are crashing and I have a problem....?

All of my tasks on are still pending validation at this point.

Edit?! SiDock server may be borked.... I have ran out of tasks that completed too fast for my liking. Or licking? I am not grate at spelling.
 
  • Like
Reactions: Fardringle

cellarnoise

Senior member
Mar 22, 2017
709
394
136
Server may be toast, like last big event. At least for awhile it seems. Still, no work for me either.

Adding tasks for other projects now. This race will be a race for upgrading the host? :) Please check for a pulse?

I do appreciate their subject matter though!
 

cellarnoise

Senior member
Mar 22, 2017
709
394
136
Server may be toast, like last big event. At least for awhile it seems. Still, no work for me either.

Adding tasks for other projects now. This race will be a race for upgrading the host? :) Please check for a pulse?

I do appreciate their subject matter though!

They may be recalibrating the project to search for my ding-a-ling. Good luck with that ;) Kind of like a where-he project?

Ancient Mysteries:


I should really stop, but with my back issue and family issues, this is my current outlet, so please forgive. Seriously.

Life is hard sometimes.
 

StefanR5R

Elite Member
Dec 10, 2016
5,459
7,718
136
SiDock had a somewhat weak server once, but upgraded it before the Pentathlon.

But during the Pentathlon, they had work which took 8 hours per task or more. Right now their server status shows 0.64 (0 - 12.71) average (min - max) hours runtime of last 100 tasks. That is, if they had the same contributor capacity as during the Pentathlon marathon, the server would have to dish out more than 12 times as many tasks per day now compared to then, which involves a respectively larger database. Result file size would also play a role in server load, but I don't have notes about file sizes back in May compared to now.

SiDock used to have a limit on tasks in progress of 2 per active logical CPU. I haven't checked whether this limit is still in place, but I guess it is. — Edit, from looking around a bit, yes, 2 per active logical CPU is the current limit.

Reporting deadline is 6 days now, quite a bit longer than before and during Pentathlon. But the late announcement of the contest naturally limits the amount of upfront bunkering.
 
Last edited:

StefanR5R

Elite Member
Dec 10, 2016
5,459
7,718
136
Just now I put two computers to SiDock. I see that work requests are frequently answered with "got 0 tasks/ project has no tasks available". Intermittently, other work requests are responded to with some tasks, and server_status.php shows that the work generator keeps up.

That is, the SiDock server most certainly is subjected to so many scheduler requests now that the little in-memory queue between work generator and scheduler is emptied far quicker than the server manages to refill it. AFAIK.

I also see some request timeouts and occasionally stalled downloads.

Clearly, the server setup which was capable enough (from what I remember) during Pentathlon with 8+ hours long tasks, is rather lacking with the current quick 20…30 minutes tasks now.

Edit, the upshot: Work is there for everyone, one should just bring a little patience with the crowded server.
 
Last edited:

Fardringle

Diamond Member
Oct 23, 2000
9,184
753
126
The server itself seems to be responding mostly OK. But there just isn't enough work available to keep up with the demand..
 

cellarnoise

Senior member
Mar 22, 2017
709
394
136
Sorry team for my over the top posts last night. Been a wee bit rough with family, self and friends past week or so... I won't elaborate. Still reminiscing though :)

I think I am giving up on this SiDock challenge. Not much work. Adding other projects back into the mix.

I do enjoy watching the power to heat to amp ratio of this 5950x work. Amazing how you can have 16 to 32 threads loaded up 100% and watch the watts / power / heat go up as the number of cores loaded goes down! I need to read up and experiment more. I really can see how a single chip-let (5800x) or other demands great cooling. AMD really pushes the frequency on these at the expense of power and heat towards turbo / max frequency. Interesting to watch.
 

StefanR5R

Elite Member
Dec 10, 2016
5,459
7,718
136
The server itself seems to be responding mostly OK. But there just isn't enough work available to keep up with the demand..
There is always enough work being generated.

The server's scheduler only has a hard time to pull it out of the work generator. That's why so many client requests for work are unsuccessful.

The unfortunate solution to this is that the client needs to keep asking for work repeatedly. If it does, it will keep getting work.

However, the short runtimes combined with the (how shall I call it politely) unfortunate low limit on tasks in progress of 2 per active logical CPU still may make it difficult for a fast computer to stay busy continuously.

The proper solution to the latter would be the project admins to increase the task quota. I'll go post at their message board later today if it isn't already discussed there.

An alternative solution is to increase the effective task quota yourself locally by the respective client configuration. This is trivial to implement.
 

Kiska

Golden Member
Apr 4, 2012
1,007
290
136
The feeder is probably not fast enough(or rather the disks IO throughput) to keep the shared memory filled with jobs:
1631693135804.png
rts is currently looks like its quite full
 

StefanR5R

Elite Member
Dec 10, 2016
5,459
7,718
136
How does it work? Along the lines of this? ->
  • Work generator creates workunits, puts them in the database.¹
  • Feeder looks up workunits (or tasks) in the database and fills some unassigned tasks into a shared memory buffer.²
  • Scheduler receives a work request, looks up in the database if the requester may receive work and of what kind, looks up tasks in the shared memory buffer, assigns tasks to the host, records this in the database.³

________
¹) One or another server component needs to create tasks out of workunits, based on the workunit parameters 'initial replication', 'minimum quorum', and 'max # of error/total/success tasks'. Perhaps the work generator already performs the initial replication at least. Or perhaps the feeder is doing that? If the latter, perhaps the work generator communicates with the feeder via SHM, not via the database.

²) That is, the mentioned temporary buffer holds far fewer tasks than what we get to see as 'tasks ready to send' in server_status.php.

³) The two operations of responding to the client and of recording it in the database are not transactional. In corner cases, task assignments are recorded in the server's database even though the response to the client doesn't get through.
 
Last edited: