Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummers-lowrise
|
1)
Message boards :
Generalized Fermat Prime Search :
DO YOU FEEL LUCKY?
(Message 135892)
Posted 1258 days ago by  KarmannGaz
Ah, thanks. Just unlucky then! I'll give it another go one day when my computer is looking more stable.
|
2)
Message boards :
Generalized Fermat Prime Search :
DO YOU FEEL LUCKY?
(Message 135890)
Posted 1258 days ago by KarmannGaz
I enjoy participating in the PrimeGrid challenges and try my best to get in the top 300 each time, and thought I was doing ok in the latest (Aussie Aussie Ossie Oi Oi Oi) until I started a DYFL task - it had taken just over 3 days of processing on my Radeon RX580 and said it was about 59% completed, when my computer decided to freeze and I had to restart it. When BOINC restarted, the DYFL task went back to zero percent and started again, so I gave up and aborted it. What a waste of processing. My other running tasks (Einstein & LHC at home) all continued from where they were before ok, it was just the GFN DYFL task which did this.
Is there some way to make these take checkpoints and continue from the latest one, like other BOINC tasks seem to, or maybe they do and it was just my odd one being very unlucky? I'm not going to try another long running task like this in the meantime just in case.
|
3)
Message boards :
AP26 - AP27 Search :
Scheduler Wait
(Message 101168)
Posted 2383 days ago by KarmannGaz
Thanks for the info Iain, I'll have a look into this. I remember having a nightmare getting it working at all with various Linux drivers and thought I had the right AMD one, but maybe not. Will have a check tomorrow with a fresh head and cup of coffee! Cheers. :-)
|
4)
Message boards :
AP26 - AP27 Search :
Scheduler Wait
(Message 101082)
Posted 2386 days ago by KarmannGaz
Thanks, yes, that does look a little unusual! Ah well, guess my gpu won't be contributing to the current challenge then, not to worry, I'll clock up a few cpu tasks instead. It would be nice to know what's wrong at some point though. Will wait and see if any app authors respond.
Cheers!
|
5)
Message boards :
AP26 - AP27 Search :
Scheduler Wait
(Message 100996)
Posted 2388 days ago by KarmannGaz
Also forgot to mention, this is on BOINC Client version 7.2.42 for x86_64-pc-linux-gnu on an AMD Athlon dual core 64-bit cpu with AMD Radeon HD6670 gpu coprocessor.
|
6)
Message boards :
AP26 - AP27 Search :
Scheduler Wait
(Message 100994)
Posted 2388 days ago by KarmannGaz
I don't seem to have those options under any of the BOINC Manager menus, but I managed to turn them on via my cc_config.xml file and there's a menu option to reread the config file(s), so fairly easy to turn them on & off that way. I didn't get much from it with just cpu_sched, so added the others you mentioned, and a couple more which looked interesting... there's a bit more in the log now, but I'm still not seeing what the problem is :
Thu 17 Nov 2016 09:25:00 GMT | | [mem_usage] All others: RAM 1424.62MB, page 24331.59MB, user 59.920, kernel 47.770
Thu 17 Nov 2016 09:25:00 GMT | | [mem_usage] non-BOINC CPU usage: 2.98%
Thu 17 Nov 2016 09:25:07 GMT | PrimeGrid | task ap27_155075_0 resumed by user
Thu 17 Nov 2016 09:25:07 GMT | | [cpu_sched_debug] Request CPU reschedule: task suspended, resumed or aborted by user
Thu 17 Nov 2016 09:25:08 GMT | | [cpu_sched_debug] schedule_cpus(): start
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [cpu_sched_debug] earliest deadline: 1479978854 ap27_155075_0
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [cpu_sched_debug] scheduling ap27_155075_0 (coprocessor job, EDF) (prio -1.000000)
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [cpu_sched_debug] reserving 1.000000 of coproc ATI
Thu 17 Nov 2016 09:25:08 GMT | | [cpu_sched_debug] enforce_schedule(): start
Thu 17 Nov 2016 09:25:08 GMT | | [cpu_sched_debug] preliminary job list:
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [cpu_sched_debug] 0: ap27_155075_0 (MD: yes; UTS: yes)
Thu 17 Nov 2016 09:25:08 GMT | | [cpu_sched_debug] final job list:
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [cpu_sched_debug] 0: ap27_155075_0 (MD: yes; UTS: no)
Thu 17 Nov 2016 09:25:08 GMT | | [mem_usage] enforce: available RAM 3754.82MB swap 1464.91MB
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [coproc] Assigning ATI instance 0 to ap27_155075_0
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [cpu_sched_debug] scheduling ap27_155075_0
Thu 17 Nov 2016 09:25:08 GMT | | [cpu_sched_debug] using 0.06 out of 2 CPUs
Thu 17 Nov 2016 09:25:08 GMT | Einstein@Home | [cpu_sched_debug] LATeah0006L_1008.0_0_0.0_15110025_0 sched state 1 next 1 task state 0
Thu 17 Nov 2016 09:25:08 GMT | Einstein@Home | [cpu_sched_debug] LATeah0006L_1040.0_0_0.0_2077950_0 sched state 1 next 1 task state 0
Thu 17 Nov 2016 09:25:08 GMT | Einstein@Home | [cpu_sched_debug] p2030.20160120.G176.30-00.40.N.b4s0g0.00000_192_1 sched state 1 next 1 task state 0
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [cpu_sched_debug] ap27_155075_0 sched state 1 next 2 task state 0
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [task] ACTIVE_TASK::start(): forked process: pid 5145
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [task] task_state=EXECUTING for ap27_155075_0 from start
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [cpu_sched] Restarting task ap27_155075_0 using ap26 version 201 (opencl_ati_AP27) in slot 3
Thu 17 Nov 2016 09:25:08 GMT | PrimeGrid | [css] running ap27_155075_0 (0.0572 CPUs + 1 ATI GPU)
Thu 17 Nov 2016 09:25:08 GMT | | [cpu_sched_debug] enforce_schedule: end
Thu 17 Nov 2016 09:25:10 GMT | PrimeGrid | [mem_usage] ap27_155075_0: WS 120.64MB, smoothed 83.20MB, page 271.52MB, 0.00 page faults/sec, user CPU 1.900, kernel CPU 0.140
Thu 17 Nov 2016 09:25:10 GMT | | [mem_usage] All others: RAM 1424.70MB, page 24331.65MB, user 60.450, kernel 48.080
Thu 17 Nov 2016 09:25:10 GMT | | [mem_usage] non-BOINC CPU usage: 4.16%
Thu 17 Nov 2016 09:25:15 GMT | PrimeGrid | [task] Process for ap27_155075_0 exited, status 0, task state 1
Thu 17 Nov 2016 09:25:15 GMT | PrimeGrid | [task] task called temporary_exit(60.000000)
Thu 17 Nov 2016 09:25:15 GMT | PrimeGrid | [task] task_state=UNINITIALIZED for ap27_155075_0 from temporary exit
Thu 17 Nov 2016 09:25:15 GMT | | [cpu_sched_debug] Request CPU reschedule: application exited
Thu 17 Nov 2016 09:25:15 GMT | | [cpu_sched_debug] schedule_cpus(): start
Thu 17 Nov 2016 09:25:15 GMT | PrimeGrid | [cpu_sched_debug] thrashing prevention: mark ap27_155075_0 as deadline miss
Thu 17 Nov 2016 09:25:15 GMT | | [cpu_sched_debug] enforce_schedule(): start
Thu 17 Nov 2016 09:25:15 GMT | | [cpu_sched_debug] preliminary job list:
Thu 17 Nov 2016 09:25:15 GMT | | [cpu_sched_debug] final job list:
Thu 17 Nov 2016 09:25:15 GMT | | [mem_usage] enforce: available RAM 3754.82MB swap 1464.91MB
Thu 17 Nov 2016 09:25:15 GMT | | [cpu_sched_debug] using 0.00 out of 2 CPUs
Thu 17 Nov 2016 09:25:15 GMT | Einstein@Home | [cpu_sched_debug] LATeah0006L_1008.0_0_0.0_15110025_0 sched state 1 next 1 task state 0
Thu 17 Nov 2016 09:25:15 GMT | Einstein@Home | [cpu_sched_debug] LATeah0006L_1040.0_0_0.0_2077950_0 sched state 1 next 1 task state 0
Thu 17 Nov 2016 09:25:15 GMT | Einstein@Home | [cpu_sched_debug] p2030.20160120.G176.30-00.40.N.b4s0g0.00000_192_1 sched state 1 next 1 task state 0
Thu 17 Nov 2016 09:25:15 GMT | PrimeGrid | [cpu_sched_debug] ap27_155075_0 sched state 2 next 1 task state 0
Thu 17 Nov 2016 09:25:15 GMT | | [cpu_sched_debug] enforce_schedule: end
The Einstein tasks mentioned are all suspended, as I suspended the project so that there's only my one PrimeGrid AP27 task active. I don't have any other tasks on that computer at the moment.
Does any of this look unusual to you perhaps?
|
7)
Message boards :
AP26 - AP27 Search :
Scheduler Wait
(Message 100990)
Posted 2388 days ago by KarmannGaz
Thanks for the suggestions JimB. My memory usage options are set at 95% when in use and 99% when idle (I don't tend to do much else with this particular computer, so just let it sit doing BOINC work most of the time and have most of the memory & cpu to itself). The BOINC log reports 3.86Gb of physical memory available (non-GPU) and 1.91Gb virtual. Plenty of disk space too (48.99Gb free). I've ticked the boxes to run cpu and gpu tasks while the computer is in use as well, so it shouldn't be suspending them unless non-BOINC work exceeds 80%, which it occasionally does when background admin tasks are running, but then ALL the tasks get suspended, not just one. I've even tried manually suspending all other tasks so there's only the AP27 one running, but it still does the same thing. I did notice that if I clicked on Properties for the task while it was in Scheduler Wait state, the virtual memory size and working set size were both only a few hundred megabytes each, but I don't know if that relates to the GPU's VRAM or just normal computer RAM, or even if that's relevant. All I can think of is that the task wants to allocate more than the 1.9Gb available, but I can't find any messages to support this or say anything else which might give me a clue what's up.
|
8)
Message boards :
AP26 - AP27 Search :
Scheduler Wait
(Message 100941)
Posted 2389 days ago by KarmannGaz
Hi,
I've been trying to get my computer to run AP27 GPU tasks, but they only run for a few seconds then go into Scheduler Wait state for a while before the task reinitialises and starts again from zero percent done. This repeats forever and I end up having to abort the tasks. From what I can gather on other BOINC sites, Scheduler Wait means it's waiting for more memory, though I can't find any exact matches on the forum here.
My GPU is an AMD/ATI Radeon (Turks) HD6670 with 2Gb RAM on board. BOINC reports it with CAL version 1.4.1848, 2048MB, 1970MB available, 1632GFlops peak, driver version 1800.11, OpenCL 1.2 AMD-APP.
Other PrimeGrid GPU tasks run fine, it only seems to be AP27 which is having this problem. I know it says AP27 tasks need 1.5Gb+ of VRAM, but as mine has more than this I thought it might be ok. I've tried this in the past and given up, but thought I'd revisit it with the Wallis is Born Challenge coming up, I was hoping it might have been a bug which has been fixed by now, but apparently not. Is there a setting I'm missing somewhere to get this working, or is the HD6670 just a non-starter?
Thanks,
Gary.
|
9)
Message boards :
Number crunching :
Summer Paralympics Challenge
(Message 98472)
Posted 2463 days ago by KarmannGaz
I like reading about all the speedy computers out there and what you're doing to make them so fast. I have two rather slower AMD based machines under Linux Mint and have been doing some case mods prior to this latest challenge, and am pleased that the A8 quad-core box with a new hole at the top and extra fan in it is now looking like completing these WUs in 54 hours, so I'm hoping to get 8 through from there by the end of the challenge period. My other is an AMD Athlon dual-core which I've installed a huge fan inside and some ducting so it blows directly on the RAM and CPU cooler fins, plus an extra exhaust fan too, and that's looking like it'll complete in 105 hours, so should also get two WUs through by the end, just. It might not sound much, but I'm a very happy bunny! :-)
|
|
