Message boards :
Number crunching :
Raspberry Pi results
Message board moderation
Author | Message |
---|---|
elec999 Send message Joined: 24 Nov 02 Posts: 375 Credit: 416,969,548 RAC: 141 |
Good Day, I am very curious since Seti is been released for the pi, if anyone is using one of these and the kinds of results they are getting. If they are decent I would get 10x and build a little seti farm based on the pies vs I would be better off getting an ~$400-450 GPU. Thank you |
Mark Wyzenbeek Send message Joined: 28 Jun 99 Posts: 134 Credit: 6,203,079 RAC: 0 |
According to the estimate, one unit will take just over 7 days on my Pi B+. Of course the Pi 2 or Pi 3 will crunch them faster. I think GPUs would be more cost-effective power wise. The Universe is not only stranger than you imagine, it's stranger than you can imagine. SETI@home classic workunits 1,405 CPU time 57,318 hours |
Sidewinder Send message Joined: 15 Nov 09 Posts: 100 Credit: 79,432,465 RAC: 0 |
My Pi 2 Model B: http://setiathome.berkeley.edu/results.php?hostid=7925555 Results in beta: http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=78147&offset=0&show_names=0&state=4&appid= Looking at the currently running tasks, it took ~42 hours to get 70% (I'm thinking they're ~100 credits each). So a 100 credit WU might be ~54 hours. My Pi is running stock 900 MHz Raspian 8 (lite) with no gui. According to Kill-a-watt, it's consuming ~3.3 watts (running 4 WUs). |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Side query: What's the rough range on Rasberry Pi 3 802.11n Wifi ? (assume reasonable access points). I have a number of control applications around the yard these would be ideal for, and have a solar panel in the yard probably sufficient to run about a dozen of them, with surplus for battery-charge/nighttime-crunching just because. Anyone using these in applications for (sparse) monitoring/control, and crunching on them as well ? "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
Sidewinder Send message Joined: 15 Nov 09 Posts: 100 Credit: 79,432,465 RAC: 0 |
Probably depends on what kind of LoS they would have to the APs. I've heard others saying anywhere from 5-30 meters for the Pi 3 (not sure on what kind of obstacles though). Since they're low power I probably wouldn't put the limit much higher than that. You could probably get a decent idea of coverage by doing a site survey with a smartphone if you have one. |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
...You could probably get a decent idea of coverage by doing a site survey with a smartphone if you have one. Good idea. Will probably pace the yard next weekend, with the Nexus 9 and Wifi Analyzer. Adding a dedicated AP would be an option, and likely within 10m for all places I'd want a unit. Minimising the support infrastructure required would be good though, if only it could mean I transfer the cablemodem/gateway to off grid, though would have to figure out how much power that thing uses first as well. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
[A.S.a.M.]koschi Send message Joined: 8 Jan 01 Posts: 15 Credit: 5,947,861 RAC: 15 |
103.000-126.000 seconds on the Raspberry Pi 3 at 1.2GHz https://setiathome.berkeley.edu/results.php?hostid=7974154 78.000-80.000 seconds on the Odroid C2 at 2GHz https://setiathome.berkeley.edu/results.php?hostid=7974151 |
Sidewinder Send message Joined: 15 Nov 09 Posts: 100 Credit: 79,432,465 RAC: 0 |
Adding a dedicated AP would be an option, and likely within 10m for all places I'd want a unit. Minimising the support infrastructure required would be good though, if only it could mean I transfer the cablemodem/gateway to off grid, though would have to figure out how much power that thing uses first as well. You could look at configuring the Pi's for a mesh network. It would take some CPU, but as long as they're all sufficiently spaced and powered it might allow you forego an additional AP. http://hackaday.com/2012/11/14/mesh-networking-with-multiple-raspberry-pi-boards/ http://scalabilly.com/2015/08/mesh-ad-hoc-network-on-multiple-raspberry-pis/ |
Dr Grey Send message Joined: 27 May 99 Posts: 154 Credit: 104,147,344 RAC: 21 |
My Pi 2 B has just returned its first workunits after 55 hrs. Still waiting for validation. |
elec999 Send message Joined: 24 Nov 02 Posts: 375 Credit: 416,969,548 RAC: 141 |
This calls for an fun project. I think I will purchase an pi. |
ML1 Send message Joined: 25 Nov 01 Posts: 20334 Credit: 7,508,002 RAC: 20 |
This calls for an fun project. I think I will purchase an pi. Or two or four or more for your own cluster?... ;-) Happy cool crunchin', Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
Raspberry Pi - Brian Send message Joined: 22 Oct 99 Posts: 12 Credit: 164,328 RAC: 0 |
There has been some discussion on "if you should use a fan" with the Raspberry Pi 3 Model B. Simple answer: YES. If the core temperature goes above 80 degrees C, it will throttle back the the cpu to protect itself. You can use a command line command or write a script to check this for yourself. I put together a simple script: speedtemp.sh #!/bin/bash vcgencmd measure_clock arm && vcgencmd measure_temp that will result in a report like: frequency(45)=900000000 temp=42.2'C Expanding the edge of Science. |
Sidewinder Send message Joined: 15 Nov 09 Posts: 100 Credit: 79,432,465 RAC: 0 |
I would agree with the fan as well; especially if you're using them as dedicated crunchers. It's pretty easy to set up a separate fan with multiple Pi's. A heatsink also helps a bit on the CPU and RAM (not as necessary on the RAM though). You can set the temp to an alias to make it a little easier: ~/.bash_aliases alias pitemp='/opt/vc/bin/vcgencmd measure_temp' Mine (so far): |
Dr Grey Send message Joined: 27 May 99 Posts: 154 Credit: 104,147,344 RAC: 21 |
Thanks for the code :) pi@raspberrypi:~ $ #!/bin/bash pi@raspberrypi:~ $ vcgencmd measure_clock arm && vcgencmd measure_temp frequency(45)=900094000 temp=71.3'C That's with a couple of little heatsinks but in a closed case. Not as bad as I thought it would be though. |
Dr Grey Send message Joined: 27 May 99 Posts: 154 Credit: 104,147,344 RAC: 21 |
Mine (so far): Looking good :) |
Raspberry Pi - Brian Send message Joined: 22 Oct 99 Posts: 12 Credit: 164,328 RAC: 0 |
I would agree with the fan as well; especially if you're using them as dedicated crunchers. It's pretty easy to set up a separate fan with multiple Pi's. A heatsink also helps a bit on the CPU and RAM (not as necessary on the RAM though). Thanks for the tip on aliases. I can always learn new ways of The Force... ...err.. I mean Linux. One of my machines is a Raspberry Pi Zero with Jessie lite. (Ugh! Just one core!) I'll use rlogin to get into it and run a script with #! /bin/bash echo Project Tasks: sudo boinccmd --get_state |grep -E 'project URL|fraction done|remaining|active_task_state' echo New Work Setting: sudo boinccmd --get_state |grep -E 'master URL|request more work' to get: Project Tasks: project URL: http://setiathome.berkeley.edu/ active_task_state: EXECUTING fraction done: 0.113847 estimated CPU time remaining: 175900.348237 New Work Setting: master URL: http://einstein.phys.uwm.edu/ don't request more work: no master URL: http://setiathome.berkeley.edu/ don't request more work: no (yeah, that would be around 11% completed and I like to see where I set the "more work flag" double negative for each project. Currently, only SETI & Einstein are installed.) Sometimes I see that boinc has stalled out and no additional work is being done. There should be nothing else running, so I'll just execute a "reboot", but there should be a better way. |
Sidewinder Send message Joined: 15 Nov 09 Posts: 100 Credit: 79,432,465 RAC: 0 |
What do you mean by stalled? No tasks are executing? I'm no Linux guru, but you could probably rig up a quick shell script like: #!/bin/bash # Infinite loop while : do x = $(sudo boinccmd --get_state | grep EXECUTING) # check if x is null if [[ -z $x ]] then # nothing is processing, restart boinc-client sudo /etc/init.d/boinc-client restart fi # clear variable x = sleep 60 done It's been awhile so you'll want to double check that code. You could run the above script in the background (sudo nohup script.sh &) or modify it and run it as a cron job (would allow you to take out the sleep commands). |
Raspberry Pi - Brian Send message Joined: 22 Oct 99 Posts: 12 Credit: 164,328 RAC: 0 |
PERFECT .. Oh, Wait.. when it's stalled, the status shows "EXECUTING" :-( I can come back an hour later and it will show the same "seconds". Even now, it shows: active_task_state: EXECUTING fraction done: 0.118937 estimated CPU time remaining: 174890.009414 To Quote Dr McCoy: "He's dead, Jim" (lame Star Trek reference) <edit> OK, not so fast... it isn't dead, but it trickels... it pops in and of a "top" command: top - 23:50:10 up 5:56, 1 user, load average: 0.00, 0.01, 0.05 Tasks: 60 total, 1 running, 58 sleeping, 0 stopped, 1 zombie %Cpu(s): 0.1 us, 0.6 sy, 0.2 ni, 98.2 id, 0.9 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 445376 total, 172952 used, 272424 free, 31984 buffers KiB Swap: 102396 total, 0 used, 102396 free. 45488 cached Mem PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 331 boinc 30 10 33560 10132 7392 S 0.6 2.3 1:40.37 boinc 1137 pi 20 0 5104 2336 2020 R 0.1 0.5 0:00.29 top 443 boinc 39 19 0 0 0 Z 0.1 0.0 30:54.20 setiathome_+ 1135 root 20 0 0 0 0 S 0.1 0.0 0:00.27 kworker/0:1 40 root 20 0 0 0 0 S 0.0 0.0 0:01.85 jbd2/mmcblk+ 292 root 20 0 0 0 0 S 0.0 0.0 0:10.38 RTW_CMD_THR+ (It should be locked in at the top!) |
Sidewinder Send message Joined: 15 Nov 09 Posts: 100 Credit: 79,432,465 RAC: 0 |
You could try something like this to see if you can spot a pattern or do some metrics on the WUs: #!/bin/bash # Infinite loop while : do echo "----- $(date +%Y-%m-%d:%H:%M:%S) -----" >> /tmp/boinc_task.log sudo boinccmd --get_state | grep -B12 -A4 'EXECUTING' | grep -E 'WU|fraction' >> /tmp/boinc_task.log echo "" >> /tmp/boinc_task.log sleep 60 done |
Raspberry Pi - Brian Send message Joined: 22 Oct 99 Posts: 12 Credit: 164,328 RAC: 0 |
hmmm .. Leaving my telnet session open, I noticed that I would get random syslog messages: Message from syslogd@raspberrypi00lite at Apr 27 09:12:45 ... kernel:[50302.140893] 3fe0: 020d08a0 beaa86b8 0002b3b8 0002b410 80000010 ffffffff 00000000 00000000 Fortunately, this isn't a "1202 alarm" that might suggest that we scrub the mission to land (Apollo 11 reference) Digging into /var/log/syslog .. I find: Apr 27 09:12:45 raspberrypi00lite kernel: [50301.833349] BUG: unsupported FP instruction in kernel mode Apr 27 09:12:45 raspberrypi00lite kernel: [50301.841611] Internal error: Oops - undefined instruction: 0 [#3] ARM Apr 27 09:12:45 raspberrypi00lite kernel: [50301.850750] Modules linked in: cfg80211 rfkill 8192cu snd_bcm2835 bcm2835_gpiomem bcm2835_rng snd_pcm snd_timer snd uio_pdrv_genirq uio ipv6 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.868980] CPU: 0 PID: 16380 Comm: setiathome_8.02 Tainted: G D 4.1.19+ #858 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.882541] Hardware name: BCM2708 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.888536] task: d8ed0000 ti: da4d2000 task.ti: da4d2000 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.896510] PC is at vfp_save_state+0x0/0x28 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.903300] LR is at vfp_sync_hwstate+0x3c/0x48 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.910286] pc : [<c000a948>] lr : [<c000a58c>] psr: 60000113 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.910286] sp : da4d3e78 ip : da4d3e90 fp : da4d3e8c Apr 27 09:12:45 raspberrypi00lite kernel: [50301.926498] r10: 002efe08 r9 : da4d2000 r8 : beaa83c8 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.934013] r7 : beaa84b8 r6 : da4d20f8 r5 : beaa85c0 r4 : c0000788 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.942786] r3 : da4d20f8 r2 : c0842364 r1 : c0000788 r0 : da4d20f8 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.951528] Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user Apr 27 09:12:45 raspberrypi00lite kernel: [50301.960902] Control: 00c5387d Table: 1a4e8008 DAC: 00000015 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.968874] Process setiathome_8.02 (pid: 16380, stack limit = 0xda4d2188) Apr 27 09:12:45 raspberrypi00lite kernel: [50301.977975] Stack: (0xda4d3e78 to 0xda4d4000) Apr 27 09:12:45 raspberrypi00lite kernel: [50301.984506] 3e60: da4d3e90 da4d2000 Apr 27 09:12:45 raspberrypi00lite kernel: [50301.996889] 3e80: da4d3eac da4d3e90 c000a6f0 c000a55c 00000000 beaa83c8 00000000 d8ed0468 Apr 27 09:12:45 raspberrypi00lite kernel: [50302.009213] 3ea0: da4d3ec4 da4d3eb0 c0012d20 c000a6c8 da4d3fb0 da4d3ec8 da4d3f8c da4d3ec8 Apr 27 09:12:45 raspberrypi00lite kernel: [50302.021626] 3ec0: c0013150 c0012bac 000d5378 14000000 0021b940 00000000 00000000 0000000e Apr 27 09:12:45 raspberrypi00lite kernel: [50302.034201] 3ee0: 00000000 00000080 00000000 00000000 00000000 00000000 00000001 da4d3f08 Apr 27 09:12:45 raspberrypi00lite kernel: [50302.046924] 3f00: c005b474 c005d5e0 0000007d 0000000e 00000011 c07d6e1c 00000001 c07d6e1c Apr 27 09:12:45 raspberrypi00lite kernel: [50302.059749] 3f20: c081b9b8 00000000 00000000 c005dc4c da4d3f4c da4d3f40 c005dc4c c005a9b8 Apr 27 09:12:45 raspberrypi00lite kernel: [50302.072767] 3f40: da4d3f5c da4d3f50 c001ff64 c002741c da4d3f74 da4d3f60 c002741c 0000000e Apr 27 09:12:45 raspberrypi00lite kernel: [50302.085987] 3f60: c005df80 00000001 da4d2000 00000000 da4d3fb0 00000000 da4d2000 002efe08 Apr 27 09:12:45 raspberrypi00lite kernel: [50302.099435] 3f80: da4d3fac da4d3f90 c0013458 c0012e8c 0002b410 80000010 f200b200 00c5387d Apr 27 09:12:45 raspberrypi00lite kernel: [50302.113119] 3fa0: 00000000 da4d3fb0 c000f7e4 c001339c 020d0400 beaa872c 00000013 020d0400 Apr 27 09:12:45 raspberrypi00lite kernel: [50302.126964] 3fc0: 020d0300 0000002d 020d0858 00000040 002f1f60 002eff31 002efe08 002e8958 Apr 27 09:12:45 raspberrypi00lite kernel: [50302.140893] 3fe0: 020d08a0 beaa86b8 0002b3b8 0002b410 80000010 ffffffff 00000000 00000000 Apr 27 09:12:45 raspberrypi00lite kernel: [50302.154990] [<c000a948>] (vfp_save_state) from [<c000a6f0>] (vfp_preserve_user_clear_hwstate+0x34/0xc0) Apr 27 09:12:45 raspberrypi00lite kernel: [50302.170501] [<c000a6f0>] (vfp_preserve_user_clear_hwstate) from [<c0012d20>] (setup_sigframe+0x180/0x190) Apr 27 09:12:45 raspberrypi00lite kernel: [50302.186059] [<c0012d20>] (setup_sigframe) from [<c0013150>] (do_signal+0x2d0/0x3e0) Apr 27 09:12:45 raspberrypi00lite kernel: [50302.199611] [<c0013150>] (do_signal) from [<c0013458>] (do_work_pending+0xc8/0xd8) Apr 27 09:12:45 raspberrypi00lite kernel: [50302.213039] [<c0013458>] (do_work_pending) from [<c000f7e4>] (work_pending+0xc/0x20) Apr 27 09:12:45 raspberrypi00lite kernel: [50302.226591] Code: e12fff1e e1a0200d e1a0e009 eafffebe (eca00b20) Apr 27 09:12:45 raspberrypi00lite rsyslogd-2007: action 'action 17' suspended, next retry is Wed Apr 27 09:13:15 2016 [try http://www.rsyslog.com/e/2007 ] Apr 27 09:12:45 raspberrypi00lite kernel: [50302.261347] ---[ end trace fa7d22edd3fd7799 ]--- I stopped running BOINC with sudo /etc/init.d/boinc-client stopand let it sit for a while, and I got nothing (making sure that it was or was not boinc related) Executing dmesg |grep seti... I get [ 636.537114] CPU: 0 PID: 441 Comm: setiathome_8.02 Not tainted 4.1.19+ #858 [ 636.618585] Process setiathome_8.02 (pid: 441, stack limit = 0xd8e98188) [ 2852.301898] CPU: 0 PID: 1052 Comm: setiathome_8.02 Tainted: G D 4.1.19+ #858 [ 2852.401553] Process setiathome_8.02 (pid: 1052, stack limit = 0xda798188) [50301.868980] CPU: 0 PID: 16380 Comm: setiathome_8.02 Tainted: G D 4.1.19+ #858 [50301.968874] Process setiathome_8.02 (pid: 16380, stack limit = 0xda4d2188) My Pi 2B doesn't get this message. Chasing after that "rsyslog-2007" hint, results in an older message in http://kb.monitorware.com/kbeventdb-detail-id-6925.html .. but I'm not sure that helps. (This Pi Zero is running Jessie Lite) .. doing on more "Googling"... and finding many hits. Seeing references to "stack limit"... is there an adjustment that I can make? Expanding the edge of Science. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.