Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
World Community Grid Forums
Category: Active Research Forum: OpenPandemics - COVID-19 Project Thread: OpenPandemics - COVID-19 is Now Live |
No member browsing this thread |
Thread Status: Active Thread Type: Sticky Thread Total posts in this thread: 282
|
Author |
|
Dmit
Cruncher Joined: Jun 28, 2021 Post Count: 2 Status: Offline Project Badges: |
Have some few OPN1 tasks with long hours delays between packs with no any OPN1.
But I replaced my very old 1366 socket PC with 6 core Westmere-E Xeon @3600 MHz to new PC with Ryzen 5600x, which crunch every OPN1 twice faster than 6c/12t Xeon : 3 hours 40 minutes for OPN1 on Xeon and less than 2 hours on 6c/12t Ryzen 5600x. And then decided to look at stats. Every 3:40 hours task from old Xeon PC gets from 130 to 300 points depending on delay before validation/verification, but newer 5600x machine gets only form 20 to 70 points per task while doing them more than twice faster. That shouldn't work this way when 10 years old 32nm hot and now inefficient Xeon machine gets more credits than much faster newer more efficient 5600x machine. I hope points per task should be at least the same between old and new PCs. if no any bonus credits granted for faster execution, because it's just completely illogical to lowering credit points for faster tasks execution. So credit system looks mostly broken as for me, and I hope that someone should fix it at least. |
||
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2089 Status: Offline Project Badges: |
See e.g. this
----------------------------------------Note that we're talking about credit, not points, when it comes to results. Points is what you see in the member and team statistics. Credit is what you see in the results. The credit system has always worked this way. 300 credits is way too much for an OPN1-task in my opinion, but that's the way it works. As a rule of thumb a slow device gets as much credit as a fast device, no matter the time it spent crunching on a task. Let's have a look at a slow device, my Android phone. After WCG's partial restart it started crunching OPN1-tasks in 25 hours and was granted 50 to 60 credit per task. There were also some fluctuations in the following month: 90 credits was quite normal one month ago, but there was also an outlier with 200 credits. At this moment, a credit of 70 seems normal. It fluctuates. And that's normal, too. After the partial restart, my Android device crunched 30 OPN1-tasks and that clocks up as: $ wcglog -xe android -1 -AA [Edit 3 times, last edit by adriverhoef at Aug 22, 2022 1:08:34 PM] |
||
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12146 Status: Offline Project Badges: |
The credits granted are the average of the claims by both machines,
|
||
|
PMH_UK
Veteran Cruncher UK Joined: Apr 26, 2007 Post Count: 766 Status: Offline Project Badges: |
The credits granted are the average of the claims by both machines, Mostly not for my OPNG: https://www.worldcommunitygrid.org/contribution/workunit/153377211 OPNG_0154017_00379_0 Linux Ubuntu Ubuntu 22.04 LTS [5.14.0-1045-oem|libc 2.35] Error 2022-08-19 14:48:19 UTC 2022-08-20 12:46:01 UTC 0.46 / 0.46 17.3 / 0 OPNG_0154017_00379_1 Linux Debian Debian GNU/Linux 11 (bullseye) [5.10.0-14-amd64|libc 2.31 (Debian GLIBC 2.31-13+deb11u3)] Valid 2022-08-19 15:03:02 UTC 2022-08-19 15:57:11 UTC 0.08 / 0.56 58.6 / 915 OPNG_0154017_00379_2 Linux Ubuntu Ubuntu 20.04.4 LTS [5.15.0-46-generic|libc 2.31 (Ubuntu GLIBC 2.31-0ubuntu9.9)] Valid 2022-08-20 12:47:08 UTC 2022-08-20 13:27:07 UTC 0.18 / 0.24 0.3 / 8.4 _2 is mine. https://www.worldcommunitygrid.org/contribution/workunit/153409738 OPNG_0154520_00422_0 Linux Gentoo Gentoo Linux [5.19.2-gentoo-r1|libc 2.35] Valid 2022-08-19 19:41:50 UTC 2022-08-20 19:56:22 UTC 0.6 / 0.6 16.6 / 734.2 OPNG_0154520_00422_1 Linux Ubuntu Ubuntu 18.04.6 LTS [4.15.0-191-generic|libc 2.27 (Ubuntu GLIBC 2.27-3ubuntu1.6)] Valid 2022-08-19 19:42:29 UTC 2022-08-19 20:37:07 UTC 0.06 / 0.79 58.6 / 878.9 _1 is mine. Paul.
Paul.
----------------------------------------[Edit 1 times, last edit by PMH_UK at Aug 22, 2022 10:08:43 AM] |
||
|
Grumpy Swede
Master Cruncher Svíþjóð Joined: Apr 10, 2020 Post Count: 2092 Status: Offline Project Badges: |
How the OPNG_0154017_00379_2 task for WU OPNG_0154017_00379 could be considered "Valid" is beyond me, since it only contains a lot of Beignet: "Exec event errors. Valid with tons Beignet errors, and was only granted 8.4 Credits. There's something very strange with that validation.
----------------------------------------Results log for OPNG_0154017_00379_2 Short examples included <core_client_version>7.16.6</core_client_version> <![CDATA[ <stderr_txt> ntel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842e410 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842fc50 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842f110 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842fc50 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842fe80 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842fc50 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842f110 error, type is 4592, error status is -5" [Edit 2 times, last edit by Grumpy Swede at Aug 22, 2022 12:41:25 PM] |
||
|
alanb1951
Veteran Cruncher Joined: Jan 20, 2006 Post Count: 873 Status: Offline Project Badges: |
The credits granted are the average of the claims by both machines, Mostly not for my OPNG: https://www.worldcommunitygrid.org/contribution/workunit/153377211 OPNG_0154017_00379_0 Linux Ubuntu Ubuntu 22.04 LTS [5.14.0-1045-oem|libc 2.35] Error 2022-08-19 14:48:19 UTC 2022-08-20 12:46:01 UTC 0.46 / 0.46 17.3 / 0 OPNG_0154017_00379_1 Linux Debian Debian GNU/Linux 11 (bullseye) [5.10.0-14-amd64|libc 2.31 (Debian GLIBC 2.31-13+deb11u3)] Valid 2022-08-19 15:03:02 UTC 2022-08-19 15:57:11 UTC 0.08 / 0.56 58.6 / 915 OPNG_0154017_00379_2 Linux Ubuntu Ubuntu 20.04.4 LTS [5.15.0-46-generic|libc 2.31 (Ubuntu GLIBC 2.31-0ubuntu9.9)] Valid 2022-08-20 12:47:08 UTC 2022-08-20 13:27:07 UTC 0.18 / 0.24 0.3 / 8.4 _2 is mine. https://www.worldcommunitygrid.org/contribution/workunit/153409738 OPNG_0154520_00422_0 Linux Gentoo Gentoo Linux [5.19.2-gentoo-r1|libc 2.35] Valid 2022-08-19 19:41:50 UTC 2022-08-20 19:56:22 UTC 0.6 / 0.6 16.6 / 734.2 OPNG_0154520_00422_1 Linux Ubuntu Ubuntu 18.04.6 LTS [4.15.0-191-generic|libc 2.27 (Ubuntu GLIBC 2.27-3ubuntu1.6)] Valid 2022-08-19 19:42:29 UTC 2022-08-19 20:37:07 UTC 0.06 / 0.79 58.6 / 878.9 _1 is mine. Paul. The WCG system works out a credit score based on the "difficulty" of the jobs within the work unit and some inside knowledge of how many iterations the GPU used for each set of dockings performed. It won't do that until the result is validated... The GPU algorithm is not the same as the CPU one -- it can stop iterating early if it appears to have docked. The credit calculation knows about that, and you may not get the same credit as your wingman (if you have one) as their GPU may have arrived at conclusions via a different path... And, if memory serves, that 8.4 credit award seems to happen occasionally but I don't recall the explanation (if there was one!) Hope that helps explain what's going on... Cheers - Al. |
||
|
alanb1951
Veteran Cruncher Joined: Jan 20, 2006 Post Count: 873 Status: Offline Project Badges: |
How the OPNG_0154017_00379_2 task for WU OPNG_0154017_00379 could be considered "Valid" is beyond me, since it only contains a lot of Beignet: "Exec event errors. Valid with tons Beignet errors, and was only granted 8.4 Credits. There's something very strange with that validation. It looks as if it was having problems during wrap-up! All you get to see in the log is the last 64K or so of stderr output -- note that the first line is missing the "drm_i" off the front!Results log for OPNG_0154017_00379_2 Short examples included <core_client_version>7.16.6</core_client_version> <![CDATA[ <stderr_txt> ntel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842e410 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842fc50 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842f110 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842fc50 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842fe80 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842fc50 error, type is 4592, error status is -5" drm_intel_gem_bo_context_exec() failed: Input/output error Beignet: "Exec event 0x1842f110 error, type is 4592, error status is -5" I don't think the validator looks at the log at all; it will have checked the uploaded results (which we can't see) and based its opinion on those. That 8.4 credit thing seems to show up every now and then -- PMH_UK's post above yours also has an example... Cheers - Al. |
||
|
Grumpy Swede
Master Cruncher Svíþjóð Joined: Apr 10, 2020 Post Count: 2092 Status: Offline Project Badges: |
It looks as if it was having problems during wrap-up! All you get to see in the log is the last 64K or so of stderr output -- note that the first line is missing the "drm_i" off the front! Yes, it was PMH_UK's post above, that I took the task example from. I hope your explanation of the Beignet error is correct, otherwise there is some serious validation problems.I don't think the validator looks at the log at all; it will have checked the uploaded results (which we can't see) and based its opinion on those. That 8.4 credit thing seems to show up every now and then -- PMH_UK's post above yours also has an example... Cheers - Al. [Edit 1 times, last edit by Grumpy Swede at Aug 22, 2022 1:32:01 PM] |
||
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 1932 Status: Offline Project Badges: |
It looks as if it was having problems during wrap-up! All you get to see in the log is the last 64K or so of stderr output -- note that the first line is missing the "drm_i" off the front! Yes, it was PMH_UK's post above, that I took the task example from. I hope your explanation of the Beignet error is correct, otherwise there is some serious validation problems.I don't think the validator looks at the log at all; it will have checked the uploaded results (which we can't see) and based its opinion on those. That 8.4 credit thing seems to show up every now and then -- PMH_UK's post above yours also has an example... Cheers - Al. Ralf |
||
|
MarkH
Advanced Cruncher United States of America Joined: May 16, 2020 Post Count: 53 Status: Offline Project Badges: |
I was able to complete some MCM jobs, but then had several MCM jobs stall repeatedly during download, finally complete, start running then crash in seconds and show "computational error". I have 2-3 MCM jobs I gave up on for a day to run a Folding@Home job, since I have a few days to try the MCMs again.
----------------------------------------
"That science of the people, by the people, for the people, shall not perish from the Earth."
|
||
|
|