Formula Boinc Sprints 2018

[StefanR5R]

"(won't finish in time) BOINC runs 32.7% of time, computation enabled 99.4% of that"

Dear client. The fact that you ran 32.7 % of the time in the past does not necessarily mean that you are going to run 32.7 % of the time in the future. Sincerely, the user.

 

[TennesseeTony]

Simon and Garfu*%nkle! I remembered the Sprint yesterday but was busy and forgot to bunker! Oh, wait, it's yoyo.....errrr uhmmm, ...awesome.

 

[crashtech]

Hmm, UK BOINC Team in 1st?

 

[ao_ika_red]

Below is my yoyo results after about 24 hours. I think OGR is not as bad as I imagine and ECM is not worth the risk for machine with limited resource. From now on, I will work only with OGR and PerfectCuboid.

 

[crashtech]

I guess I am a little confused as to which app is the memory hungry one. I have some machines that are a little short on memory.

 

[Howdy]

I guess I am a little confused as to which app is the memory hungry one. I have some machines that are a little short on memory.

Seems like maybe the operator might be a "little short on memory" too.

Appears it is ECM your thinking of

 

[crashtech]

Seems like maybe the operator might be a "little short on memory" too.

Appears it is ECM your thinking of

 

[zzuupp]

Well, I just dropped my small cache.

Good luck all. and go TeAm!

 

[StefanR5R]

Below is my yoyo results after about 24 hours.

Run times are not overly relevant (if they are all moderate). Take a look at credits/hour. If you can average only over a few tasks, then keep in mind that c/h vary quite a bit at yoyo@home even within a single subproject. And towards the end of the sprint, take validation lag into account in case of quorum 2 projects.

(edit) ECM's PPD per host is unpredictable though. If a host receives a bad batch of tasks, CPUs may fall idle because a number of tasks are suspended to wait for memory.

 

[ao_ika_red]

Run times are not overly relevant (if they are all moderate). Take a look at credits/hour. If you can average only over a few tasks, then keep in mind that c/h vary quite a bit at yoyo@home even within a single subproject. And towards the end of the sprint, take validation lag into account in case of quorum 2 projects.

(edit) ECM's PPD per host is unpredictable though. If a host receives a bad batch of tasks, CPUs may fall idle because a number of tasks are suspended to wait for memory.

The unpredictable nature of yoyo sub-projects put me in dilemma. Of course ECM has the best c/h but if left unattended when some problem occurs, we will lose precious running time. Risk vs reward really comes into play here.

 

[StefanR5R]

Looks like yet another project with bogus crediting.
https://www.rechenkraft.net/forum/viewtopic.php?f=56&t=11559&start=192

It doesn't only look like that, it is.

My first few Perfect Cuboid validations are in. The claimed credit of my wingmen is all over the place. Some claim half of what my client claims. Others claim 8 times of what my client claims.

This is pure nonsense; worse than YAFU.

 

[Howdy]

It doesn't only look like that, it is.

My first few Perfect Cuboid validations are in. The claimed credit of my wingmen is all over the place. Some claim half of what my client claims. Others claim 8 times of what my client claims.

This is pure nonsense; worse than YAFU.

What are the "given" credits? All of the completed (I can find) and I have checked (through their confusing layout) are given the same amount between myself and the wing men. Supposedly by the last statement in the thread they limited the amount of credit. My average after completing and verified (77) is now 187/ project.
Time Claimed Awarded
Wingman> Over / Success / Done / 9,210.00 / 73.62 / 147.25
Me> Over / Success / Done / 13,134.59 / 99.18 / 147.25

 

[StefanR5R]

Re Perfect Cuboid:

It is correct that both get the same credit. However, I don't understand either how the server computes the awarded credit. I have only few Cuboid results completed yet, and fewer validated, but looking at other hosts I see many WUs for which more credit is awarded than the clients claim.

Maybe the server also looks at the FLOPS estimation of all participating hosts across all WUs, not just at the two (or sometimes three) hosts of the one WU. The global FLOPS estimation would be inflated by those hosts which send bogus claims.​

Re ECM:

More unpredictable factors which influence host PPD: How many tasks will crash? Will they crash early or after hours of running? Will Windows helpfully present you with a crash dialogue which you must acknowledge before the client carries on and starts the next task?​

 

[TennesseeTony]

Check out the scores in League 2. Wow, [H] is virtually tied with Gridcoin, but XtremeSys is way on up there.

 

[Howdy]

[H] is split like us though aren't they (well apparently bunkering SETI- so it seems) whereas Xtreme is not running SETI?

 

[pututu]

I guess I am a little confused as to which app is the memory hungry one. I have some machines that are a little short on memory.

I had problems with task that has "_np_" in it. So I aborted these tasks. You can check how much memory this task is consuming by clicking the task name and "properties" in bionic manager. I saw something like 7GB per task.
ecm_uc_1535074782_np_161_2900e6_01_795_P2

 

[biodoc]

I had problems with task that has "_np_" in it. So I aborted these tasks. You can check how much memory this task is consuming by clicking the task name and "properties" in bionic manager. I saw something like 7GB per task.
ecm_uc_1535074782_np_161_2900e6_01_795_P2

I've got one of those running now.

Spoiler

Application
ecm 705.01
Name
ecm_uc_1535071884_np_161_2900e6_01_11_P2
State
Running
Received
Thu 23 Aug 2018 09:01:03 PM EDT
Report deadline
Tue 28 Aug 2018 09:01:01 PM EDT
Estimated computation size
22,334 GFLOPs
CPU time
00:53:24
CPU time since checkpoint
00:53:24
Elapsed time
00:54:39
Estimated time remaining
01:56:15
Fraction done
40.108%
Virtual memory size
8.06 GB
Working set size
8.00 GB
Directory
slots/37
Process ID
37367
Progress rate
43.920% per hour
Executable
ecmwrapper_705.01_x86_64-pc-linux-gnu

 

[StefanR5R]

The yoyo@home web site is garbage.

Not only is it impossible to filter results lists by subproject or by state. The server also removes completed tasks from these tables so quickly that it is practically impossible to track progress and throughput.

Edit,
I just thought I should be a good citizen and post complaints like the above at the yoyo@home forum. But I now see that the forum requires an extra registration; the boinc user credentials are not accepted.

 

[biodoc]

I've got 9 of those ecm "np" work units running on my 2P (40 threads and 96 GB of RAM). RAM usage is up to 81 GB since I took this screenshot.

 

[Howdy]

I've got 9 of those ecm "np" work units running on my 2P (40 threads and 96 GB of RAM). RAM usage is up to 81 GB since I took this screenshot.

Looks like you're trying to hack the Matrix to me

Just hope you do not have me pointed towards the "F9" button.

@biodoc : I kid you not and it was freaky to me........Not soon after I posted that maybe 5 mins my DD turned off, no warning just OFF!! Tried to restart- no dice- The lights on the MOBO were even on!! Check PSU- OK not tripped checked even though the MOBO looked to be on, everything on the outlet was working, Tried again and it booted............NEO WE FOUND YOU!!!!!

 

[StefanR5R]

The first 18 hours... Now With 22% Extra Hours!

 

[Orange Kid]

Well we have settled in to our comfortable spot at #3. Looks like a good place to maintain as we split our fight on this and the SetiWOW races.
Tuff project with the best paying app a memory hog. Now we just need to wary of any bunkers/


1 Gridcoin 25 3,089,241
2 SETI.USA 18 1,532,828
3 TeAm AnandTech 15 1,327,778
4 Czech National Team 12 514,190
5 UK BOINC Team 10 497,660
6 Planet 3DNow! 8 343,094
7 Rechenkraft.net 6 243,258
8 Russia Team 4 235,823
9 L'Alliance Francophone 2 213,425
10 SETI.Germany 1 99,849

From FreeDC
crashtech 388,299
xii5ku 196,593
biodoc 190,980
Tejas 107,011
[TA]OrangeKid 99,743
10esseeTony 56,097
Ken_g6 53,713
petrusbroder 50,027
Howdy2u2 34,931
iwajabitw 26,880
zzuupp 18,364
GLeeM 14,881
MarcelliusIkaP 8,533
Kiska 1,939

 

[crashtech]

Some of you guys are sandbagging for sure, but for now i can revel in the #1 spot!

 

[Markfw]

Should I take my 2990WX off of seti and put it on yoyo ?

 

[biodoc]

Should I take my 2990WX off of seti and put it on yoyo ?

That's up to you. I've got my GPUs on Seti and CPUs on yoyo.