Calculating task completion estimates

Everything about the project RNA World
Nachricht
Autor
Benutzeravatar
gemini8
Vereinsvorstand
Vereinsvorstand
Beiträge: 5878
Registriert: 31.05.2011 10:30
Wohnort: Hannover

Re: Calculating task completion estimates

#25 Ungelesener Beitrag von gemini8 » 02.10.2016 06:49

No need to tell you to 'have fun', is there? :-D
Gruß, Jens
- - - - - -
Lowend-User und Teilzeit-Cruncher

Bild Bild Bild
Bild

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#26 Ungelesener Beitrag von Jacob Klein » 02.10.2016 07:34

:) You can bet I will 'have fun'!

Actually, while spending all those hours nailing down that Dell Alienware software instability on my main rig Speed ...
...
My aging 7+-year-old Dell XPS 730X, RacerX, is having problems where he was randomly rebooting. I found a way to repro it easily at the time, just run PrimeGrid on each of the 3 GPUs, then start Prime95, and watch it reboot. I blew out a little dust, and reseated the 3 video cards, and unplugged/replugged the power supply harness... and now he doesn't have that insta-reboot problem anymore... But Prime95 is giving me "Application has stopped working" after a while, even when the GPUs are idle, and even when CPU is running stock non-overclocked!
...
My bet is power supply, especially after I read a lot of info about them, and how they become less efficient over time and can become more susceptible to voltage droop. When RacerX is running optimally, I try to push 860 watts at the wall, out of a 1000 watt power supply, which is quite a challenge for it.
...
So, to fix it, I now have a $50 OEM XPS 730X "new OEM/Dell" power supply on order (hell of a deal for an old part!), and RacerX gets to take a break from BOINC for 10 days (since I refuse to crunch on a known unstable system, for fear of damaging the tasks, or worse yet, silently passing them with invalid results).

I really hope it is the power supply and not the motherboard. Looking forward to installing the new power supply, and retesting everything :)

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#27 Ungelesener Beitrag von Jacob Klein » 02.10.2016 17:41

Speed is ready to take on new work. The body (CPU) is willing, and now the flesh (RAM) is no-longer weak ;) Technically, I could probably do (64GB / 3GB_per_task = 21) tasks, but doing more than 16 on an 8-core-PC would be stupid, and ... I'm actually going for 15, since Oracle has them run at normal priority, and I'd like at least 1 thread to run a "regular BOINC task" (at idle priority), to allow OS/UI user interaction responsiveness, since I use this machine too, hehe! So, looking for 6 more!

Work continues on diagnosing RacerX. I reseated all the RAM, and am possibly seeing positive results in testing, but I'm going to keep him sidelined from BOINC until he's rock solid stable.

And Nitro surpassed 400 days :good: on a task! I certainly celebrated :lol: , even if my best-guess-estimate for some of these tasks is in the 500-700 day range :o

Code: Alles auswählen

Speed Tasks: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz
                     122.8d -- 10/2/2016 -- UI 98.765%, progress.txt 0.98765
                     96.1d  -- 10/2/2016 -- UI 96.696%, progress.txt 0.966965
                     61.3d  -- 10/2/2016 -- UI 77.436%, progress.txt 0.774370
                     59.7d  -- 10/2/2016 -- UI 50.935%, progress.txt 0.509359
                     58.0d  -- 10/2/2016 -- UI 98.765%, progress.txt 0.98765
                     52.0d  -- 10/2/2016 -- UI 81.898%, progress.txt 0.818992
                     34.3d  -- 10/2/2016 -- UI 34.189%, progress.txt 0.341894
                     5.4d   -- 10/2/2016 -- UI  5.240%, progress.txt 0.052407
                     2.1d   -- 10/2/2016 -- UI  1.759%, progress.txt 0.017599
                     37.8d  -- 9/27/2016 -- ***COMPLETED!***
                     28.7d  -- 9/9/2016  -- ***COMPLETED!***
                     9.0d   -- 7/13/2016 -- ***FAILED!***
RacerX Tasks: Intel(R) Core(TM) i7 CPU 965 @ 3.20GHz
                     66.7d  -- 10/2/2016 -- UI 42.544%, progress.txt 0.425577
                     33.2d  -- 10/2/2016 -- UI 19.293%, progress.txt 0.193060
                     302.7d -- 9/13/2016 -- ***COMPLETED!***
                     174.1d -- 8/7/2016  -- ***COMPLETED!***
                     182.3d -- 7/13/2016 -- ***COMPLETED!***
                     16.5d  -- 7/12/2015 -- ***COMPLETED!***
Nitro Tasks: Intel(R) Core(TM) i7 CPU Q 740 @ 1.73GHz
                     406.1d -- 10/2/2016 -- UI 82.219%, progress.txt 0.98765
                     331.4d -- 10/2/2016 -- UI 80.003%, progress.txt 0.98765
                     185.4d -- 10/2/2016 -- UI 96.425%, progress.txt 0.964253
                     181.4d -- 10/2/2016 -- UI 98.765%, progress.txt 0.98765
Zathras Tasks: Intel(R) Core(TM) i5 CPU M 520 @ 2.40GHz
                     374.7d -- 10/2/2016 -- UI 90.119%, progress.txt 0.98765
                     60.6d  -- 10/2/2016 -- UI 26.883%, progress.txt 0.268833
                     303.3d -- 6/26/2016 -- ***COMPLETED!***

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#28 Ungelesener Beitrag von Jacob Klein » 09.10.2016 23:38

Regarding RacerX: I have spent the past week diagnosing RacerX, but could not determine why he fails Prime95. I've replaced the power supply, relaxed memory timings, removed all non-beta software, reinstalled the original memory, underclocked almost everything.... and it still fails. My only 2 remaining hunches are something bad on the motherboard (which I'm not going to replace), or maybe the Northbridge is overheating during a Prime95 run (I left it without a heatsink, when I redid the water cooling several months ago). Either way, I'm going to restore all my normal overclocks, and push forward with crunching, as I've not yet seen any "real" instability miscalculations yet. RacerX's week-long break is OVER - BACK TO WORK!

Regarding Windows: Microsoft's latest Insider Build of Windows 10... is awesome, but Oracle's shoddy hardening code doesn't play nice. VirtualBox is broken on Windows 10 Insider Build 14942, so I've spend the past 2 days updating my "Release Partitions" on all 4 of my PCs. I'm glad Oracle keeps me on my toes, but it's a lot of work. Anyway, I've got 3 of the PCs crunching RNA tasks again in Release partitions, but Speed is giving me a bit of trouble that I'll still have to work through. So he's taking a small break.

Did I mention that I ***love*** living on the bleeding edge, while crunching long-running RNA World VMs? :)

Code: Alles auswählen

Speed Tasks: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz
                     127.7d -- 10/9/2016 -- UI 98.765%, progress.txt 0.98765
                     100.9d -- 10/9/2016 -- UI 98.765%, progress.txt 0.98765
                     66.1d  -- 10/9/2016 -- UI 83.517%, progress.txt 0.835333
                     64.5d  -- 10/9/2016 -- UI 55.056%, progress.txt 0.550647
                     62.8d  -- 10/9/2016 -- UI 98.765%, progress.txt 0.98765
                     56.8d  -- 10/9/2016 -- UI 89.507%, progress.txt 0.895193
                     39.2d  -- 10/9/2016 -- UI 38.979%, progress.txt 0.389899
                     10.2d  -- 10/9/2016 -- UI  9.889%, progress.txt 0.098974
                     6.9d   -- 10/9/2016 -- UI  5.710%, progress.txt 0.057164
                     37.8d  -- 9/27/2016 -- ***COMPLETED!***
                     28.7d  -- 9/9/2016  -- ***COMPLETED!***
                     9.0d   -- 7/13/2016 -- ***FAILED!***
RacerX Tasks: Intel(R) Core(TM) i7 CPU 965 @ 3.20GHz
                     67.1d  -- 10/9/2016 -- UI 42.827%, progress.txt 0.428282
                     33.6d  -- 10/7/2016 -- UI 19.555%, progress.txt 0.195550
                     302.7d -- 9/13/2016 -- ***COMPLETED!***
                     174.1d -- 8/7/2016  -- ***COMPLETED!***
                     182.3d -- 7/13/2016 -- ***COMPLETED!***
                     16.5d  -- 7/12/2015 -- ***COMPLETED!***
Nitro Tasks: Intel(R) Core(TM) i7 CPU Q 740 @ 1.73GHz
                     412.5d -- 10/9/2016 -- UI 82.698%, progress.txt 0.98765
                     337.9d -- 10/9/2016 -- UI 80.616%, progress.txt 0.98765
                     191.8d -- 10/9/2016 -- UI 99.756%, progress.txt 0.997566
                     187.8d -- 10/9/2016 -- UI 98.765%, progress.txt 0.98765
Zathras Tasks: Intel(R) Core(TM) i5 CPU M 520 @ 2.40GHz
                     381.1d -- 10/7/2016 -- UI 90.500%, progress.txt 0.98765
                     67.0d  -- 10/9/2016 -- UI 29.709%, progress.txt 0.297094
                     303.3d -- 6/26/2016 -- ***COMPLETED!***

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#29 Ungelesener Beitrag von Jacob Klein » 10.10.2016 14:07

Woot - Speed's back in business crunching again, AND he picked up a 10th concurrent task! So, he's still looking for 5 more, and RacerX is looking for 1 more. :)

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#30 Ungelesener Beitrag von Jacob Klein » 12.10.2016 02:55

RacerX got his task - what fun!

And Oracle has Test Builds that fix VirtualBox to work correctly on Windows 10 Insider Build 14942 ... for 5.0.x and 5.1.x :) But they're not going to fix 4.3.x, meaning that the "Anniversary Update" version 1607, will be the last Windows 10 Release build to run VirtualBox 4.3.x correctly. I'll be required to use my "Release partitions" on any PCs that require 4.3.x, and also I must set "Defer Upgrades" to prolong the automatic upgrading to new Windows 10 Release builds. Time is ticking for me to get these done hehe.

Summary:
Speed has 10 tasks, wanting 5 more. 7x 1.17, 3x 1.18, so he can run VirtualBox 5.0.x, on the Insider partition.
RacerX has 3 tasks, wanting 0 more. 2x 1.17, 1x 1.18, so he can run VirtualBox 5.0.x, on the Insider partition.
Nitro has 4 tasks, wanting 0 more. 2x 1.15, 2x 1.17, so he must run VirtualBox 4.3.x, on the Release partition with "Defer Upgrades".
Zathras has 2 tasks, wanting 0 more. 1x 1.15, 1x 1.17, so he must run VirtualBox 4.3.x, on the Release partition with "Defer Upgrades".

Good thing I have a text file full of notes, for me to keep track of this stuff :)

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#31 Ungelesener Beitrag von Jacob Klein » 14.10.2016 23:01

Speed picked up an 11th concurrent task. Wow. He's starting to bog down a little bit, but still looking for 4 more :)

Code: Alles auswählen

Speed Tasks: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz
                     132.1d -- 10/14/2016 -- UI 98.765%, progress.txt 0.98765
                     105.3d -- 10/14/2016 -- UI 98.765%, progress.txt 0.98765
                     70.5d  -- 10/14/2016 -- UI 89.090%, progress.txt 0.890911
                     68.9d  -- 10/14/2016 -- UI 58.828%, progress.txt 0.588288
                     67.3d  -- 10/14/2016 -- UI 98.765%, progress.txt 0.98765
                     61.2d  -- 10/24/2016 -- UI 96.478%, progress.txt 0.964791
                     43.6d  -- 10/14/2016 -- UI 43.389%, progress.txt 0.433896
                     14.6d  -- 10/14/2016 -- UI 14.174%, progress.txt 0.141753
                     11.4d  -- 10/14/2016 -- UI  9.342%, progress.txt 0.093422
                     4.2d   -- 10/14/2016 -- UI  3.669%, progress.txt 0.036700
                     1.0d   -- 10/14/2016 -- UI  1.016%, progress.txt 0.010167
                     37.8d  -- 9/27/2016  -- ***COMPLETED!***
                     28.7d  -- 9/9/2016   -- ***COMPLETED!***
                     9.0d   -- 7/13/2016  -- ***FAILED!***
RacerX Tasks: Intel(R) Core(TM) i7 CPU 965 @ 3.20GHz
                     71.7d  -- 10/14/2016 -- UI 45.747%, progress.txt 0.457474
                     38.2d  -- 10/14/2016 -- UI 22.226%, progress.txt 0.222263
                     3.2d   -- 10/14/2016 -- UI  3.069%, progress.txt 0.030697
                     302.7d -- 9/13/2016  -- ***COMPLETED!***
                     174.1d -- 8/7/2016   -- ***COMPLETED!***
                     182.3d -- 7/13/2016  -- ***COMPLETED!***
                     16.5d  -- 7/12/2015  -- ***COMPLETED!***
Nitro Tasks: Intel(R) Core(TM) i7 CPU Q 740 @ 1.73GHz
                     417.2d -- 10/14/2016 -- UI 83.039%, progress.txt 0.98765
                     342.5d -- 10/14/2016 -- UI 81.053%, progress.txt 0.98765
                     196.5d -- 10/14/2016 -- UI 98.765%, progress.txt 0.98765
                     192.5d -- 10/14/2016 -- UI 98.765%, progress.txt 0.98765
Zathras Tasks: Intel(R) Core(TM) i5 CPU M 520 @ 2.40GHz
                     385.8d -- 10/14/2016 -- UI 90.775%, progress.txt 0.98765
                     71.7d  -- 10/14/2016 -- UI 31.813%, progress.txt 0.318141
                     303.3d -- 6/26/2016  -- ***COMPLETED!***

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#32 Ungelesener Beitrag von Jacob Klein » 31.10.2016 19:38

Status update!

- A couple days ago, Nitro completed his **first ever** RNA World task! Happy dance! Since he's now working on 3, and he can support 4, he's now "on the search" for a new task!

- Zathras's long task, has now exceeded 400 days. He's happy too!

- Microsoft has made a change in the Windows 10 Insider Fast Ring, and Oracle VirtualBox's windows process hardening logic broke with it. Oracle fixed the 5.1.x and 5.0.x lines, but 4.3.x will not be fixed. Nitro and Zathras, both have a "1.15" task each, so they are required to use the 4.3.x version of VirtualBox. So... I switched both those PCs over to use their "Windows Release" partitions, and I changed their Windows Update settings to use "Defer Upgrades". But, I estimate that the upcoming "1703" version of Windows 10, will be forced on them, around July 2017. To get those "1.15" tasks done quicker, I've set them for "Use at most 50% CPUs" (half-threads). I call this "Get 'er DONE" mode, hehe. If they're not done, when "1703" is forced upon the "Defer Upgrades" PCs, then I'll take further action.

- Speed continues to search for 4 more tasks, to get 15 total.

Code: Alles auswählen

Speed Tasks: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz
                     146.5d -- 10/31/2016 -- UI 98.765%, progress.txt 0.98765
                     119.8d -- 10/31/2016 -- UI 98.765%, progress.txt 0.98765
                     85.0d  -- 10/31/2016 -- UI 98.765%, progress.txt 0.98765
                     83.3d  -- 10/31/2016 -- UI 71.113%, progress.txt 0.711137
                     81.7d  -- 10/31/2016 -- UI 98.765%, progress.txt 0.98765
                     75.6d  -- 10/31/2016 -- UI 98.765%, progress.txt 0.98765
                     58.0d  -- 10/31/2016 -- UI 57.723%, progress.txt 0.577235
                     29.1d  -- 10/31/2016 -- UI 28.116%, progress.txt 0.281165
                     25.8d  -- 10/31/2016 -- UI 21.187%, progress.txt 0.211876
                     18.6d  -- 10/31/2016 -- UI 16.230%, progress.txt 0.162307
                     15.5d  -- 10/31/2016 -- UI 15.221%, progress.txt 0.152219
                     37.8d  -- 9/27/2016  -- ***COMPLETED!***
                     28.7d  -- 9/9/2016   -- ***COMPLETED!***
                     9.0d   -- 7/13/2016  -- ***FAILED!***
RacerX Tasks: Intel(R) Core(TM) i7 CPU 965 @ 3.20GHz
                     88.3d  -- 10/31/2016 -- UI 56.334%, progress.txt 0.563343
                     54.8d  -- 10/31/2016 -- UI 31.885%, progress.txt 0.318857
                     19.8d  -- 10/31/2016 -- UI 18.952%, progress.txt 0.189526
                     302.7d -- 9/13/2016  -- ***COMPLETED!***
                     174.1d -- 8/7/2016   -- ***COMPLETED!***
                     182.3d -- 7/13/2016  -- ***COMPLETED!***
                     16.5d  -- 7/12/2015  -- ***COMPLETED!***
Nitro Tasks: Intel(R) Core(TM) i7 CPU Q 740 @ 1.73GHz
                     433.9d -- 10/31/2016 -- UI 84.207%, progress.txt 0.98765  -- switched to running HALF THREADS
                     213.3d -- 10/31/2016 -- UI 98.765%, progress.txt 0.98765  -- switched to running HALF THREADS
                     209.3d -- 10/31/2016 -- UI 98.765%, progress.txt 0.98765  -- switched to running HALF THREADS
                     353.3d -- 10/27/2016 -- ***COMPLETED!***
Zathras Tasks: Intel(R) Core(TM) i5 CPU M 520 @ 2.40GHz
                     402.7d -- 10/31/2016 -- UI 91.689%, progress.txt 0.98765  -- switched to running HALF THREADS
                     88.6d  -- 10/31/2016 -- UI 39.301%, progress.txt 0.393019 -- switched to running HALF THREADS
                     303.3d -- 6/26/2016  -- ***COMPLETED!***

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#33 Ungelesener Beitrag von Jacob Klein » 21.11.2016 03:52

Status update!

- Nitro completed his SECOND TASK, clocking in at 226.9 days! It wasn't his v1.15 task like I was hoping, but it still counts as a HUGE WIN :roll2: :) He's now looking for 2 new tasks, to accompany his current 2. Speed still looks for 4 more.

- Windows 10 Build 14971, is currently incompatible with Oracle VirtualBox 5.0.x and 5.1.x, with no workaround expected for weeks at least. So, in addition to Nitro and Zathras reverting to "Release Partitions" due to requiring 4.3.x for their v1.15 tasks... both Speed and RacerX have also now reverted to "Release Partitions", in order to continue working on RNA World tasks. This meant several hours of updating those partitions, on my part. :)

- Nitro's current long-running task, has now reached 450+ days! Several other tasks have reached their 100-day milestones. Keep crunching!

Code: Alles auswählen

Speed Tasks: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz
                     163.2d -- 11/20/2016 -- UI 98.765%, progress.txt 0.98765
                     136.4d -- 11/20/2016 -- UI 98.765%, progress.txt 0.98765
                     101.6d -- 11/20/2016 -- UI 98.765%, progress.txt 0.98765
                     100.0d -- 11/20/2016 -- UI 85.325%, progress.txt 0.853256
                     98.4d  -- 11/20/2016 -- UI 98.765%, progress.txt 0.98765
                     92.3d  -- 11/20/2016 -- UI 98.765%, progress.txt 0.98765
                     74.7d  -- 11/20/2016 -- UI 74.313%, progress.txt 0.743133
                     45.7d  -- 11/20/2016 -- UI 44.211%, progress.txt 0.442120
                     42.4d  -- 11/20/2016 -- UI 34.891%, progress.txt 0.348911
                     35.3d  -- 11/20/2016 -- UI 30.763%, progress.txt 0.307641
                     32.1d  -- 11/20/2016 -- UI 31.603%, progress.txt 0.316040
                     37.8d  -- 9/27/2016  -- ***COMPLETED!***
                     28.7d  -- 9/9/2016   -- ***COMPLETED!***
                     9.0d   -- 7/13/2016  -- ***FAILED!***
RacerX Tasks: Intel(R) Core(TM) i7 CPU 965 @ 3.20GHz
                     105.5d -- 11/20/2016 -- UI 67.293%, progress.txt 0.672932
                     71.9d  -- 11/20/2016 -- UI 41.894%, progress.txt 0.418943
                     36.9d  -- 11/20/2016 -- UI 35.387%, progress.txt 0.353878
                     302.7d -- 9/13/2016  -- ***COMPLETED!***
                     174.1d -- 8/7/2016   -- ***COMPLETED!***
                     182.3d -- 7/13/2016  -- ***COMPLETED!***
                     16.5d  -- 7/12/2015  -- ***COMPLETED!***
Nitro Tasks: Intel(R) Core(TM) i7 CPU Q 740 @ 1.73GHz
                     454.1d -- 11/20/2016 -- UI 85.504%, progress.txt 0.98765
                     233.5d -- 11/20/2016 -- UI 98.765%, progress.txt 0.98765
                     226.9d -- 11/19/2016 -- ***COMPLETED!***
                     353.3d -- 10/27/2016 -- ***COMPLETED!***
Zathras Tasks: Intel(R) Core(TM) i5 CPU M 520 @ 2.40GHz
                     422.9d -- 11/20/2016 -- UI 92.662%, progress.txt 0.98765
                     108.8d -- 11/20/2016 -- UI 48.239%, progress.txt 0.482393
                     303.3d -- 6/26/2016  -- ***COMPLETED!***

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#34 Ungelesener Beitrag von Jacob Klein » 22.11.2016 17:31

The most recent Oracle VirtualBox problem, affecting Windows 10 Insider builds, has been fixed, even on the 4.3.x line!

The fixed builds are:
- Oracle VirtualBox v5.1.10 Build 112026
- Oracle VirtualBox v5.0.29 Test Build 112021
- Oracle VirtualBox v4.3.41 Test Build 112025 --> https://www.virtualbox.org/download/tes ... 25-Win.exe

We were tracking the issue here:
https://forums.virtualbox.org/viewtopic ... 38&t=80701

---------------------------------------------

This means that Nitro and Zathras can run Windows 10 Insider partitions again, and might work correctly even when the Windows 10 Creators Update (v1703) is forced upon them around July 2017.

Also, Speed picked up a record *12th concurrent task*! There was some UI contention issues running 11, and they have worsened a bit with 12, but still manageable. Looking for 2 more to get 14 :)

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#35 Ungelesener Beitrag von Jacob Klein » 28.11.2016 02:04

Speed has picked up his record *13th concurrent* RNA World VM task. WOOT! The UI gets quite bogged down, so... my plan is to only allow 12 of them to run when I am using the PC, and allow all 13+ to run when I'm away. Lots of manual babysitting, but it's worth it. He's still looking for 1 more, to do 14 concurrently. And Nitro is still looking for 2 more, to do 4 concurrently.

Code: Alles auswählen

Speed Tasks: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz
                     169.9d -- 11/27/2016 -- UI 98.765%, progress.txt 0.98765
                     143.1d -- 11/27/2016 -- UI 98.765%, progress.txt 0.98765
                     108.3d -- 11/27/2016 -- UI 98.765%, progress.txt 0.98765
                     106.7d -- 11/27/2016 -- UI 91.080%, progress.txt 0.910804
                     105.1d -- 11/27/2016 -- UI 98.765%, progress.txt 0.98765
                     99.1d  -- 11/27/2016 -- UI 98.765%, progress.txt 0.98765
                     81.4d  -- 11/27/2016 -- UI 81.019%, progress.txt 0.810194
                     52.5d  -- 11/27/2016 -- UI 50.736%, progress.txt 0.507369
                     49.2d  -- 11/27/2016 -- UI 40.429%, progress.txt 0.404295
                     42.0d  -- 11/27/2016 -- UI 36.641%, progress.txt 0.366416
                     38.9d  -- 11/27/2016 -- UI 38.228%, progress.txt 0.382286
                     5.9d   -- 11/27/2016 -- UI  5.442%, progress.txt 0.054420
                     0.2d   -- 11/27/2016 -- UI  0.147%, progress.txt 0.001479
                     37.8d  -- 9/27/2016  -- ***COMPLETED!***
                     28.7d  -- 9/9/2016   -- ***COMPLETED!***
                     9.0d   -- 7/13/2016  -- ***FAILED!***
RacerX Tasks: Intel(R) Core(TM) i7 CPU 965 @ 3.20GHz
                     112.3d -- 11/27/2016 -- UI 71.680%, progress.txt 0.716801
                     78.8d  -- 11/27/2016 -- UI 45.897%, progress.txt 0.458974
                     43.8d  -- 11/27/2016 -- UI 41.971%, progress.txt 0.419711
                     302.7d -- 9/13/2016  -- ***COMPLETED!***
                     174.1d -- 8/7/2016   -- ***COMPLETED!***
                     182.3d -- 7/13/2016  -- ***COMPLETED!***
                     16.5d  -- 7/12/2015  -- ***COMPLETED!***
Nitro Tasks: Intel(R) Core(TM) i7 CPU Q 740 @ 1.73GHz
                     460.1d -- 11/27/2016 -- UI 85.872%, progress.txt 0.98765
                     239.5d -- 11/27/2016 -- UI 98.765%, progress.txt 0.98765
                     226.9d -- 11/19/2016 -- ***COMPLETED!***
                     353.3d -- 10/27/2016 -- ***COMPLETED!***
Zathras Tasks: Intel(R) Core(TM) i5 CPU M 520 @ 2.40GHz
                     428.9d -- 11/27/2016 -- UI 92.929%, progress.txt 0.98765
                     114.8d -- 11/27/2016 -- UI 50.902%, progress.txt 0.509022
                     303.3d -- 6/26/2016  -- ***COMPLETED!***

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Calculating task completion estimates

#36 Ungelesener Beitrag von Jacob Klein » 18.12.2016 04:31

Well, I was HOPING that one of my tasks would have completed by now, but they're all still happily chugging along. Speed and Nitro are still looking for more tasks, and it looks like some users are doing a great job of keeping them running, which is GOOD!

I also double-checked the <rsc_fpops_bound> values of the tasks.
They all ship with a default of:
<rsc_fpops_bound>200000000000000000.000000</rsc_fpops_bound>
and when I divide by the <p_fpops> values for my PCs
<p_fpops>4218471840.499818</p_fpops>
<p_fpops>3791600660.506046</p_fpops>
<p_fpops>2168560082.995922</p_fpops>
<p_fpops>2766771727.416799</p_fpops>
I get fpops limits of:
548.7d (1.5yr)
610.5d (1.7yr)
1067.4d (2.9yr)
836.6d (2.3yr)
... which may not be enough!

A long time ago, I changed these values (in the client_state.xml file, while BOINC was NOT running), to be:
<rsc_fpops_bound>1200000000000000000.000000</rsc_fpops_bound>
... basically multiplying it by 6, such that my limits are now:
3292.4d (9.0yr)
3663.1d (10.0yr)
6404.7d (17.5yr)
5019.9d (13.7yr)
... That should be a better buffer! It would REALLY suck to have a task work 1.5-2.3 years and then fail due to <rsc_fpops_bound>! I think the admins should multiply it by a factor of 1.5 or 2.0, for safety!

Code: Alles auswählen

Speed Tasks: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz
                     187.9d -- 12/17/2016 -- UI 98.765%, progress.txt 0.98765
                     161.2d -- 12/17/2016 -- UI 98.765%, progress.txt 0.98765
                     126.4d -- 12/17/2016 -- UI 98.765%, progress.txt 0.98765
                     124.8d -- 12/17/2016 -- UI 98.765%, progress.txt 0.98765
                     123.1d -- 12/17/2016 -- UI 98.765%, progress.txt 0.98765
                     117.1d -- 12/17/2016 -- UI 98.765%, progress.txt 0.98765
                     99.4d  -- 12/17/2016 -- UI 98.967%, progress.txt 0.989671
                     70.5d  -- 12/17/2016 -- UI 68.182%, progress.txt 0.681833
                     67.2d  -- 12/17/2016 -- UI 55.224%, progress.txt 0.552250
                     60.0d  -- 12/17/2016 -- UI 52.363%, progress.txt 0.523637
                     56.9d  -- 12/17/2016 -- UI 55.955%, progress.txt 0.559553
                     23.9d  -- 12/17/2016 -- UI 22.012%, progress.txt 0.220128
                     10.9d  -- 12/17/2016 -- UI  9.895%, progress.txt 0.098957
                     37.8d  -- 9/27/2016  -- ***COMPLETED!***
                     28.7d  -- 9/9/2016   -- ***COMPLETED!***
                     9.0d   -- 7/13/2016  -- ***FAILED!***
RacerX Tasks: Intel(R) Core(TM) i7 CPU 965 @ 3.20GHz
                     131.7d -- 12/17/2016 -- UI 84.066%, progress.txt 0.840663
                     98.2d  -- 12/17/2016 -- UI 57.195%, progress.txt 0.571957
                     63.2d  -- 12/17/2016 -- UI 60.557%, progress.txt 0.605580
                     302.7d -- 9/13/2016  -- ***COMPLETED!***
                     174.1d -- 8/7/2016   -- ***COMPLETED!***
                     182.3d -- 7/13/2016  -- ***COMPLETED!***
                     16.5d  -- 7/12/2015  -- ***COMPLETED!***
Nitro Tasks: Intel(R) Core(TM) i7 CPU Q 740 @ 1.73GHz
                     478.5d -- 12/17/2016 -- UI 86.934%, progress.txt 0.98765
                     257.4d -- 12/17/2016 -- UI 98.765%, progress.txt 0.98765
                     226.9d -- 11/19/2016 -- ***COMPLETED!***
                     353.3d -- 10/27/2016 -- ***COMPLETED!***
Zathras Tasks: Intel(R) Core(TM) i5 CPU M 520 @ 2.40GHz
                     448.0d -- 12/17/2016 -- UI 93.716%, progress.txt 0.98765
                     133.9d -- 12/17/2016 -- UI 59.366%, progress.txt 0.593668
                     303.3d -- 6/26/2016  -- ***COMPLETED!***

Antworten

Zurück zu „RNA World Discussions (english)“