Server maintenance on 07.10.2018
Message boards :
News :
Server maintenance on 07.10.2018
Message board moderation
Author | Message |
---|---|
Send message Joined: 4 Apr 15 Posts: 4 Credit: 140,884,136 RAC: 0 |
|
Send message Joined: 23 Sep 15 Posts: 15 Credit: 42,196,320 RAC: 0 |
Last modified: 10 Sep 2019, 12:23:56 UTC Seems to me as the admins wait for nearly a week after running empty before releasing a new batch of WUs. That could be seen when looking at the "Credit per day" diagram on the BoincStats page: https://www.boincstats.com/stats/134/project/detail/credit Maybe this is to clear out the rather large cache of "Tasks in progress", which tends to be several hundreds thousands of tasks!! I can believe that is not very nice for the servers to handle. However, I think that such issues are better dealt with by setting down the deadlines for reissued tasks. Most of the tasks take less than two hours on a modern i5/i7 machine, and even the longest tasks are crunched in about 5 hours on an old mobile Core-II Duo. Therefore I think that the current deadline of 10.5 days is unnecessary long. //Gunnar |
Send message Joined: 28 Sep 13 Posts: 29 Credit: 120,387,757 RAC: 9,943 |
Last modified: 10 Sep 2019, 14:35:29 UTC Gunnar It would not surprise me if Kyong did want to clear out the batch. And it is ridiculous the members here fill their buffers to the max. I use .1 days of ALL buffers - no extra time, so that I am working "current". It just amazes me, even though the project is clearing, I STILL have 649 WUs "waiting for validation"! Looking in on the "hogger" machines that are slowing down this process, and I can see turnaround times measured in 8 to 9 days! Mine is ONLY .12 days. AND, I have also seen machines with 100s of WUs buffered go down - never to reappear - and ALL of those WUs have to be recycled. I would hope that Kyong would set a reasonable time frame - 3 days max. Although I may run close to 60 cores on the project, it can take over two weeks for an actual effective RAC - because of the LONG validation times caused by "hoggers". When there are new WUs available, my machines will automatically kick back to this project, as I have set this project as a higher priority. This also keeps my machines from being "down" during these "hogger" clearing periods. James |
Send message Joined: 23 Sep 15 Posts: 15 Credit: 42,196,320 RAC: 0 |
Hi James! I do agree with you there - there sure are some dreadful hoggers, hoarding lots of tasks and creating this situation. Personally I don't bite off more than I can crunch in a few days. Now there seems to be new WUs created however! :-) I don't know if it would help to limit the amount of tasks fetched for each CPU core as I know many other projects are doing. While I was doing a lot of TN-Grid crunching this spring I saw some very strange computers with 3072 cores!! Obviously it was an attempt to fool that limiting system in connection with some form of "Boinc challenge". //Gunnar |
Send message Joined: 9 Sep 19 Posts: 2 Credit: 3,360 RAC: 0 |
|
Send message Joined: 17 Feb 17 Posts: 13 Credit: 44,071,565 RAC: 0 |
|
Send message Joined: 23 Sep 15 Posts: 15 Credit: 42,196,320 RAC: 0 |
Hi Wolfman! I do agree with you in both these questions: The WU bin at the server seems to go empty every 14:th day or so, and often not to be refilled until the next Thursday. That discourage many from running this project at all. On the other hand it encourage many existing users to setting their work buffer to at least 5 days or more. I think that is a bit bad habit and have negative consequences both for the server which has to keep track of about 350,000 WUs "in flight", and it is also bad for the users who gets very big pending bins before being credited for the work. I have also noticed that the Cuda application doesn't seem to be too efficient. I only have a GTX 750Ti and a GTX 960 in my computers, and they are just about 3 times as efficient as a single thread on a slow hyper-threaded XEON E3-1245 (not the sharpest knife in the box). I know far too little about Cuda programming to be able to tell why, or how the application should be rewritten to better utilize the massively parallel GPU architecture. I hope the admins can find time to address these issues soon, and also to write a bit more about the scientific progress of the project, which I am quite curious about. Happy Crunching to you all!!! //Gunnar |
Send message Joined: 23 Sep 15 Posts: 15 Credit: 42,196,320 RAC: 0 |
More crunchy WU-food!, yummy yummy!!! :-) (see also news-thread with link to new interesting paper: http://asteroidsathome.net/boinc/forum_thread.php?id=772 ) //Gunnar |
Send message Joined: 18 Mar 18 Posts: 2 Credit: 3,700,320 RAC: 0 |
|
Send message Joined: 18 Mar 18 Posts: 2 Credit: 3,700,320 RAC: 0 |
Last modified: 2 Oct 2019, 9:09:08 UTC |
Send message Joined: 23 Sep 15 Posts: 15 Credit: 42,196,320 RAC: 0 |
Hi David! Seems that we're temporarily out of work for the moment, but I have PM:ed one of the admins so hopefully there will be new WUs soon. (This usually happens twice a month or so.) Just keep the computer running and it will ask the servers regularly and get tasks as soon as they appear on the server. Have a nice week!! //Gunnar |
Send message Joined: 17 Feb 17 Posts: 13 Credit: 44,071,565 RAC: 0 |
|
Send message Joined: 16 Mar 17 Posts: 3 Credit: 552,000 RAC: 0 |
Hi David! Still nothing. What was the last peak of active users within 24 hours, something like 4300? Now it's 1300. 3000 people that would compute for Asteroids@Home are now doing something else. So much wasted computing power. Instead of computing for something as important as asteroid research, the computing power now goes towards solving completely pointless mathematical stuff ... like stupid prime numbers, aliquot sequences or calculating Pi to 150 million decimal places ... or something similarly pointless. Since those are some of the few projects that yield a decent reward for CPUs. |
Send message Joined: 23 Sep 15 Posts: 15 Credit: 42,196,320 RAC: 0 |
Hi Nick! I certainly agree with you there!! It's a pity that the admins cannot supply with the WUs in time, or AT LEAST TELL US WHY there is a shortage! I'm mostly into physics BOINC projects, and just now I had to switch some of my starving computers over to Milkyway@home instead. In just a couple of days I'll have to switch all the rest of them too, and then it will take a LOOOONG time before I try on the asteroids again! :-( Hope that there will be more WUs before Monday. Have a nice weekend!!! //Gunnar |
Send message Joined: 17 Feb 17 Posts: 13 Credit: 44,071,565 RAC: 0 |
Hi David! I just ranted a little. Sorry. Oh man I seriously can't even begin to agree with you enough there. I'm not into math, at all, yet it pays the best out of everything for no other reason than to attract crunchers- and attract it does. Then there are the biological and medicine related projects that pay a pittance compared to even this project. 2 years of crunching regularly and exclusively got me to 25 million credits over at WCG. It's disheartening that I've already, after only crunching regularly here for less than a month, nearly gained 10 million credits. I'm well aware that it's all about the science. But when you see those numbers crawling up and taking months and months to get into the hundreds of thousands (yes, Seti, I'm looking at you) it just doesn't seem worth it. And don't get me started on project participation by the admins. Most of the numbers projects have fantastic administrative and GPU/CPU support and optimizations. I joined this project because my fleet is primarily CPUs and very few GPUs worth crunching with (and this project doesn't really give much of a reward for using a gpu). This lack of work and more frustrating yet, lack of word from any admin or developer is making me want to move 100 plus threads varying from Atom to Ryzen elsewhere where I will feel more appreciated. My electric bill goes up because I crunch - we could have a bit of transparency to explain why work isn't forthcoming. Is it because the server reaches a certain amount then gets overwhelmed? Do folks take vacations on the weekend? DO folks recommend any other CPU projects similar to this that pay reasonably well? Milkyway? Einstein? I'd like to support Climate prediction, but they have such small batches of work that I can never seem to get any recently. Sorry for the long message. |
Send message Joined: 24 Aug 13 Posts: 111 Credit: 31,652,006 RAC: 1,989 |
Hi Admins What's going on? No WUs for days! By all means link me if I've missed a relevant thread..... Cheers Mark Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10 2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7 |
Send message Joined: 24 Aug 13 Posts: 111 Credit: 31,652,006 RAC: 1,989 |
And WUs are available now! Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10 2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7 |
Send message Joined: 23 Sep 15 Posts: 15 Credit: 42,196,320 RAC: 0 |
And WUs are available now! But not any more... The latest batch was only half the size and lasted less than two days!!! I cannot but wonder if the project isn't coming to its very end now? When "theSkyNet/POGS" ended last year it happened totally without any warnings from the admins! (Probably to minimize the set of remaining tasks circulating out in the grid.) Is this project maybe heading in the same direction? //Gunnar |
Send message Joined: 17 Feb 17 Posts: 13 Credit: 44,071,565 RAC: 0 |
And WUs are available now! Hope not, but if this continues I will be pulling my 100 plus threads out. I get outages, but having them without a word as to why is disheartening. |
Send message Joined: 23 Sep 15 Posts: 15 Credit: 42,196,320 RAC: 0 |
And WUs are available now! More tasty WU-food! :-) Seems like this batch is at least a good million if not even more. That's really nice!, Thank you admins for that! Let's hope the Asteroids project carries on, and gives good scientific results too! Have a good, crunchy weekend!! :-) //Gunnar |
Message boards :
News :
Server maintenance on 07.10.2018