Message boards :
News :
30 Workunit Limit Per Request - Fix Implemented
Message board moderation
Previous · 1 . . . 3 · 4 · 5 · 6
Author | Message |
---|---|
Send message Joined: 8 May 09 Posts: 3339 Credit: 524,010,781 RAC: 0 |
I don't know if this is the right thread or not, but since no one seems to know what the problem is, I will try it. Setup a zero resource share project as a backup, PrimeGrid would work well for you if you pick the GFN-15 units, then after the 10 minute wait MilkyWay will refill your cache and you will be crunching here again. |
Send message Joined: 18 Nov 08 Posts: 291 Credit: 2,461,693,501 RAC: 0 |
PrimeGrid would work well for you if you pick the GFN-15 units, then after the 10 minute wait MilkyWay will refill your cache and you will be crunching here again. Primegrid is not on the Gridcoin whitelist, discussion here Clearly, the principals involved have a fear of miners faking prime number to get credits which is rightly called the "gridcoin derangement syndrome". However, that does not stop them from accepting gridcoin contributions in addition to paypal. Reminds me of the "A-listers" who took private jets to lecture us on climate change. The typical celebrity uses more electricity in one month than first Worlders do in 2 or 3 years. |
Send message Joined: 5 Jul 11 Posts: 990 Credit: 376,143,149 RAC: 0 |
PrimeGrid would work well for you if you pick the GFN-15 units, then after the 10 minute wait MilkyWay will refill your cache and you will be crunching here again. I agree with this post in its entirety. |
Send message Joined: 8 May 09 Posts: 3339 Credit: 524,010,781 RAC: 0 |
PrimeGrid would work well for you if you pick the GFN-15 units, then after the 10 minute wait MilkyWay will refill your cache and you will be crunching here again. Umm not to prolong this thought process but that's NOT what PrimeGrid said when they declined to be whitelisted. |
Send message Joined: 5 Jul 11 Posts: 990 Credit: 376,143,149 RAC: 0 |
Umm not to prolong this thought process but that's NOT what PrimeGrid said when they declined to be whitelisted. What did they say? I would have thought a project should welcome gridcoin users, they are after all contributing computing power (albeit for the wrong reason). |
Send message Joined: 29 Apr 17 Posts: 33 Credit: 7,041,502,264 RAC: 0 |
Hey Guys, I've read thru this thread trying to find a solution to an issue I've recently experienced... I have been running 2 Titan Vs dedicated to MW which have been able to post 3.5-4M as a RAC over the past few months. I "repurposed" a 3rd TV to this rig expecting to increase the productivity, but that;s not what I see. Each TV is set to run 6 tasks and each task completes in about 55-60sec (so ~ 18 tasks per minute, each task has 0.5 cpu which has been sufficient for 2 TVs). Adding a 3rd I expected to see at least 4.5-5.5M RAC, but I still only get about 4M RAC over the past 3 days. I took a look at the Event log and see that the server loads about 300 tasks each time it sends work - these complete in less than 20 min. Then the rig sits idle for 10-20 min with the relevant event log section c-p below: 12/15/2019 9:38:24 AM | Milkyway@Home | Computation for task de_modfit_14_bundle5_testing_4s3f_2_1574164502_14593798_0 finished 12/15/2019 9:39:52 AM | Milkyway@Home | [sched_op] Starting scheduler request 12/15/2019 9:39:52 AM | Milkyway@Home | Sending scheduler request: To fetch work. 12/15/2019 9:39:52 AM | Milkyway@Home | Reporting 4 completed tasks 12/15/2019 9:39:52 AM | Milkyway@Home | Requesting new tasks for NVIDIA GPU 12/15/2019 9:39:52 AM | Milkyway@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices 12/15/2019 9:39:52 AM | Milkyway@Home | [sched_op] NVIDIA GPU work request: 1814400.00 seconds; 3.00 devices 12/15/2019 9:39:53 AM | Milkyway@Home | Scheduler request completed: got 0 new tasks 12/15/2019 9:39:53 AM | Milkyway@Home | [sched_op] Server version 713 12/15/2019 9:39:53 AM | Milkyway@Home | Project requested delay of 91 seconds 12/15/2019 9:39:53 AM | Milkyway@Home | [sched_op] handle_scheduler_reply(): got ack for task de_modfit_82_bundle4_4s_south4s_bgset_2_1574164502_14359817_1 12/15/2019 9:39:53 AM | Milkyway@Home | [sched_op] handle_scheduler_reply(): got ack for task de_modfit_86_bundle4_4s_south4s_bgset_2_1574164502_14593783_0 12/15/2019 9:39:53 AM | Milkyway@Home | [sched_op] handle_scheduler_reply(): got ack for task de_modfit_14_bundle5_testing_4s3f_3_1574164502_14555903_1 12/15/2019 9:39:53 AM | Milkyway@Home | [sched_op] handle_scheduler_reply(): got ack for task de_modfit_14_bundle5_testing_4s3f_2_1574164502_14593798_0 12/15/2019 9:39:53 AM | Milkyway@Home | [sched_op] Deferring communication for 00:01:31 12/15/2019 9:39:53 AM | Milkyway@Home | [sched_op] Reason: requested by project 12/15/2019 9:52:34 AM | Milkyway@Home | [sched_op] Starting scheduler request 12/15/2019 9:52:34 AM | Milkyway@Home | Sending scheduler request: To fetch work. 12/15/2019 9:52:34 AM | Milkyway@Home | Requesting new tasks for NVIDIA GPU 12/15/2019 9:52:34 AM | Milkyway@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices 12/15/2019 9:52:34 AM | Milkyway@Home | [sched_op] NVIDIA GPU work request: 1814400.00 seconds; 3.00 devices 12/15/2019 9:52:36 AM | Milkyway@Home | Scheduler request completed: got 308 new tasks 12/15/2019 9:52:36 AM | Milkyway@Home | [sched_op] Server version 713 12/15/2019 9:52:36 AM | Milkyway@Home | Project requested delay of 91 seconds 12/15/2019 9:52:36 AM | Milkyway@Home | [sched_op] estimated total CPU task duration: 0 seconds 12/15/2019 9:52:36 AM | Milkyway@Home | [sched_op] estimated total NVIDIA GPU task duration: 17919 seconds 12/15/2019 9:52:36 AM | Milkyway@Home | [sched_op] Deferring communication for 00:01:31 12/15/2019 9:52:36 AM | Milkyway@Home | [sched_op] Reason: requested by project At 9:39:52 the last 4 tasks report as completed from the previous task-basket received. The server "sees" 3.00 Devices" and "got 0 new tasks" at 9:39:53, delays 91 sec... but the automatic request for more work fails to fetch any work until 9:52:36 when it gets 308 tasks (which complete in less than 20 min.). The system returns the work, and sits for at least 10-15min before more work arrives. So in effect, this rig only works about 50% of the time available to do work. It seems that adding a 3rd Titan V simply processed the task dump faster leading to a longer idle time? This is a home/personal PC, running windows 10 (https://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=776231 ). I'm not a programmer (just an enthusiast) but I would be able to do things like "make a B-tasks Rule" with specific instruction... Hoping the experienced users here can help: Is there anyway to get this 3rd Titan V to contribute in a meaningful way?? |
Send message Joined: 5 Jul 11 Posts: 990 Credit: 376,143,149 RAC: 0 |
To MW? Possibly with some fiddling. The others will let you know how. Myself, I can't be bothered. If MW aren't going to give out a sensible number of tasks, then when I have a machine sat idle, I add another project to it. You can even set the other project to 0 weight, which means it only does it when there are no MW tasks available. The only problem here are the MW admins who have set things up completely wrong. No other project has this difficulty. I thought it might be because the tasks finish so quick, that we're reporting them too often and the server won't give out tasks if we hassle it too much. But not so. I just changed my machine to want a buffer of 2+2 hours, which means it has the 3 hours queue (I only have one Radeon 7970), and hasn't spoken to the server for 1 hour when it hits the minimum buffer of 2 hours. At this point it reports 1 hour of completed tasks, and asks for an additional lot of tasks. The server gives me none. It's really weird, somehow they've managed to set up the server so you cannot report tasks and get new tasks at the same time. But that is always going to happen when the tasks are finished so quickly (another oddity to this project - tiny tasks) - they finish faster than the time the server wants us to wait between each communication, so on every single communication, no matter what your settings, you will have at least 1 task done, which means no new tasks. And then to make matters worse, for some reason when you run out of tasks, some kind of backoff timer is given to us by the server so instead of just waiting the usual 1.5 minutes, it waits 10 minutes? What the hell? Why do they do that?! I don't even understand how that could be possible. When I run out of tasks, why isn't my client asking the server 1.5 minutes later? Why does the timer change to 10 minutes just because I'm not processing now? If anything it should be more likely to ask! |
Send message Joined: 29 Apr 17 Posts: 33 Credit: 7,041,502,264 RAC: 0 |
I appreciate your sentiment, though I don't (yet) share it. Still looking for a FIX to this problem.... |
Send message Joined: 5 Jul 11 Posts: 990 Credit: 376,143,149 RAC: 0 |
I appreciate your sentiment, though I don't (yet) share it. I added another paragraph in my last message after you replied. I'd like someone to explain where this extra timer comes into play. Normally my client connects every 1.5 minutes, and tries to report completed tasks and get more work. Why is my client then just dozing off for TEN minutes instead of 1.5 when the work runs out? It's contacting the server for the purpose of getting new work, NOT to report work. So it should still have the same desire for new work as it did before. So I should only have it idle for 1.5 minutes, not 10!! What is going on here?! |
Send message Joined: 18 Nov 08 Posts: 291 Credit: 2,461,693,501 RAC: 0 |
I appreciate your sentiment, though I don't (yet) share it. I have a "fix" for the "milkyway" problem and in addition it can be used to get more than the 200 max per GPU if so desired. https://github.com/JStateson/BoincMasterSlave I posted windows executable for just the milkyway fix on another thread. For Linux you will have to build the above as I don't have an install for various Linux systems |
Send message Joined: 29 Apr 17 Posts: 33 Credit: 7,041,502,264 RAC: 0 |
Thank you ! I'll give it a try tonight! |
Send message Joined: 29 Apr 17 Posts: 33 Credit: 7,041,502,264 RAC: 0 |
I appreciate your sentiment, though I don't (yet) share it. Okay, extracted, and ... then what? Is there an installer or a readme file in there somewhere? :) |
Send message Joined: 29 Apr 17 Posts: 33 Credit: 7,041,502,264 RAC: 0 |
I appreciate your sentiment, though I don't (yet) share it. Which thread did you post the windows exec just to fix MW? ... plz. ;) |
Send message Joined: 18 Nov 08 Posts: 291 Credit: 2,461,693,501 RAC: 0 |
Which thread did you post the windows exec just to fix MW? ... plz. ;) https://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4532&postid=69272#69272 This only fixes the Milkyway problem, it does not give a bigger GPU count. |
Send message Joined: 29 Apr 17 Posts: 33 Credit: 7,041,502,264 RAC: 0 |
I posted back in that thread ... please take a look. :) |
Send message Joined: 23 Oct 11 Posts: 8 Credit: 539,713 RAC: 0 |
I came after ages. Hello everyone. This thread is hilarious if you go through it from start to end in one sitting. Enjoy yourselves. |
Send message Joined: 2 Aug 11 Posts: 13 Credit: 44,453,057 RAC: 0 |
I already explained it here few month ago: - 1.5 min back off timer is set by server after each communication. Its OK - protection from SPAMming/DDOS - 10 min back off time is set by BOINC client if getting new work request failed. It is expected behavior set by BOINC programmers and works from client side, so it can be overridden by user (manually or by script) - getting new work request fails due to errors in MW server software which can not correctly handle combined request (reporting completed work + requesting new work in one request) - server sends new work only if client does NOT report completed work in the same communication. This error is a root of the problem and it also trigger 10 min backoff timer in client/ |
Send message Joined: 5 Jul 11 Posts: 990 Credit: 376,143,149 RAC: 0 |
So isn't it simply a case of me setting my client to use a time other than 10 minutes? Or can that value not be set easily? |
Send message Joined: 7 May 14 Posts: 57 Credit: 206,540,646 RAC: 328 |
hi all made vid on youtube for multiple instances instruction's and at full load on a Radeon VII RADEON VII GIGABYTE// 3 Instances_ Milkyway@home WUs BOINC_ 3_instances https://www.youtube.com/watch?v=4xKy9wGKmz4 all the best and welcome to earth |
©2024 Astroinformatics Group