Anyone still using a raspberry pi 3

Message boards : Number crunching : Anyone still using a raspberry pi 3
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Profile Jeearr

Send message
Joined: 14 Nov 00
Posts: 1
Credit: 719,126
RAC: 2,392
United States
Message 1891614 - Posted: 24 Sep 2017, 4:53:26 UTC
Last modified: 24 Sep 2017, 4:55:00 UTC

Running 12 of them(Pi 3's), adding in 3 more soon. In a month or so, will add 5 more for a total of 20. These are currently the only devices I am running Seti on. You can see the rig on my profile page, although I am running 2x fans now, and replaced the 2x Pi 2s with 3s.
ID: 1891614 · Report as offensive     Reply Quote
elec999

Send message
Joined: 24 Nov 02
Posts: 220
Credit: 200,889,837
RAC: 319,083
Canada
Message 1893649 - Posted: 6 Oct 2017, 17:05:56 UTC

Is it worth running seti on the pis?
$50 each and producing 300 results each, 6000 results costs you 20*50 = $1k
ID: 1893649 · Report as offensive     Reply Quote
Profile MarkJProject Donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 08
Posts: 1044
Credit: 50,380,824
RAC: 2,437
Australia
Message 1893687 - Posted: 6 Oct 2017, 21:12:23 UTC - in response to Message 1893649.  
Last modified: 6 Oct 2017, 21:20:23 UTC

Is it worth running seti on the pis?
$50 each and producing 300 results each, 6000 results costs you 20*50 = $1k

If you want to use them as crunchers you’ll need to add a heatsink to the SoC and a fan for air flow, and then there is a power supply.

You can have one larger fan cool a stack of them. Some of the multi-port USB chargers can be used to run 4 or more Pi’s off a single power point. The SD cards and fans wear out too so you’ll need to replace them over time, assume 1 to 2 years out of the SD card. There are some people running them off solar/battery but that adds quite a bit to the initial cost.

See Mod my Pi case for what I did to mine.
BOINC blog
ID: 1893687 · Report as offensive     Reply Quote
Profile Gary CharpentierCrowdfunding Project Donor
Volunteer tester
Avatar

Send message
Joined: 25 Dec 00
Posts: 21118
Credit: 30,655,903
RAC: 23,924
United States
Message 1893747 - Posted: 7 Oct 2017, 3:46:31 UTC - in response to Message 1891114.  

Using 5 Raspberry Pi 3s crunching SETI@Home 100% CPU time. I have them in a 3D printed rack and laid on their side with a big 24V fan pushing hair through each slot in the rack and keeping the CPU packet temperature at a nice chilly 38 degrees Celcius.

Works great, currently crunching almost 1500 average credit and am in the top 25% of average credit contributors.


My Raspberry Pi 2s are crunching SETI using only a single core (project_max_concurrent set to 1) and running other projects on the other cores.
Even though they are only using a single core they have a RAC of 80-90. If your Pi 3s are running on all cores 24/7 they should have a significantly higher RAC.
As I mentioned earlier I also have a Pi 3, but it gets very hot (> 75 degrees Celcius) when running SETI, so I am using it to run other projects now. When I did run SETI, it was significantly faster that the Pi 2's I have.
Based on my experience a Pi 3 running 24/7 on all cores should be able to have a RAC of at least 450.
Are you sure you Pi 3s are running full speed on all 4 cores ?

I'd echo the RAC of 450+. My Pi3 is getting 350-375 on Seti, while pumping out an additional 15-20 on Beta and 80-90 on Einstein and the Pi is also doing other non-BOINC compute intensive tasks. Ratios are 6000 Seti, 600 Einstein, 100 Beta. YMMV
ID: 1893747 · Report as offensive     Reply Quote
Tom Rinehart
Volunteer tester

Send message
Joined: 12 Dec 01
Posts: 109
Credit: 1,511,630
RAC: 4,968
United States
Message 1893750 - Posted: 7 Oct 2017, 5:06:04 UTC - in response to Message 1893747.  

The beta app is the same app as on main, so there is no need to run it for testing purposes.

I run S@H on two Pi3s (309 & 295 RAC), three Pi2s (298, 293, 276 RAC), and one Pi1B(29 RAC). I also run an Orange Pi One (267 RAC). They all have heat sinks, but no fans and are located in a cool basement.
ID: 1893750 · Report as offensive     Reply Quote
Tom Rinehart
Volunteer tester

Send message
Joined: 12 Dec 01
Posts: 109
Credit: 1,511,630
RAC: 4,968
United States
Message 1893751 - Posted: 7 Oct 2017, 5:19:17 UTC

Is RAC a good measure? It is variable based on how quickly tasks are validated. If I look at Average processing rate by clicking on Details for a computer and the Application Details I get:

Pi3 - 1.55 GFLOPS & 1.52 GFLOPS
Pi2 - 1.47 GFLOPS, 1.42 GFLOPS, & 1.42 GFLOPS
Pi1 - 0.69 GFLOPS
Orange Pi One -1.61 GFLOPS
ID: 1893751 · Report as offensive     Reply Quote
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 8893
Credit: 115,251,013
RAC: 70,859
Australia
Message 1893755 - Posted: 7 Oct 2017, 5:57:31 UTC - in response to Message 1893751.  
Last modified: 7 Oct 2017, 5:57:59 UTC

Is RAC a good measure? It is variable based on how quickly tasks are validated.

The variability in Credit is almost impossible to fathom, however one factor that doesn't come in to play is how long it takes to validate.
Hardware used, software used, hardware used by your wingman, software used by your wingman. Maybe even how you sit in your chair and use your keyboard all have an effect on what Credit you do (or don't) get. But how quickly tasks validate doesn't impact on the Credit awarded.


If I look at Average processing rate by clicking on Details for a computer and the Application Details

As long as only 1 WU is being run at a time, and there is no re-scheduling taking place, then APR is a very good method of comparing performance between systems.

Even so, the distribution of work from the Scheduler is often less than random- i have 2 systems and the slower one frequently picks up a higher proportion of GBT work (which takes longer to process) than Arecibo work. So it's APR is lower than it would be if it got the same proportion of work as the other system (it wouldn't even come close to the to the other system's APR, but it would be higher than it is).
Grant
Darwin NT
ID: 1893755 · Report as offensive     Reply Quote
BetelgeuseFiveProject Donor
Volunteer tester

Send message
Joined: 6 Jul 99
Posts: 115
Credit: 12,210,548
RAC: 6,962
Netherlands
Message 1893790 - Posted: 7 Oct 2017, 12:12:05 UTC

My 2 cents:

Pi2: https://setiathome.berkeley.edu/show_host_detail.php?hostid=8235893

Measured floating point speed 0.44 billion ops/sec
Measured integer speed 1.72 billion ops/sec

Pi3: https://setiathome.berkeley.edu/show_host_detail.php?hostid=8250996

Measured floating point speed 0.75 billion ops/sec
Measured integer speed 2.47 billion ops/sec

My general impression: Pi3 is about 50% faster than Pi2

Tom
ID: 1893790 · Report as offensive     Reply Quote
Profile Chris Feyen

Send message
Joined: 1 Jan 01
Posts: 22
Credit: 5,840,550
RAC: 22,457
United States
Message 1894548 - Posted: 11 Oct 2017, 2:26:01 UTC - in response to Message 1893790.  

I have 2 Raspberry pi 3s. Both are set to run 2 procs only to keep temps down - even thou I have a case that acts as a heat sink, and to keep a couple of days work. I haven't touched them for months - I forget they are even there doing work.
~Chris

ID: 1894548 · Report as offensive     Reply Quote
Profile Connor BaileyProject Donor

Send message
Joined: 9 Jan 17
Posts: 5
Credit: 313,944
RAC: 1,563
United States
Message 1895878 - Posted: 18 Oct 2017, 1:24:14 UTC - in response to Message 1891114.  

Using 5 Raspberry Pi 3s crunching SETI@Home 100% CPU time. I have them in a 3D printed rack and laid on their side with a big 24V fan pushing hair through each slot in the rack and keeping the CPU packet temperature at a nice chilly 38 degrees Celcius.

Works great, currently crunching almost 1500 average credit and am in the top 25% of average credit contributors.


My Raspberry Pi 2s are crunching SETI using only a single core (project_max_concurrent set to 1) and running other projects on the other cores.
Even though they are only using a single core they have a RAC of 80-90. If your Pi 3s are running on all cores 24/7 they should have a significantly higher RAC.
As I mentioned earlier I also have a Pi 3, but it gets very hot (> 75 degrees Celcius) when running SETI, so I am using it to run other projects now. When I did run SETI, it was significantly faster that the Pi 2's I have.
Based on my experience a Pi 3 running 24/7 on all cores should be able to have a RAC of at least 450.
Are you sure you Pi 3s are running full speed on all 4 cores ?



Yep I'm positive I'm running them on all cores. A single work unit takes about a day of CPU time to complete which knocks my RAC way down. Each of my Pi's have a RAC between 250 and 400 and it fluctuates depending on the work they get.

If you're going to be running SETI@Home for a long period of time, you're going to want some sort of fan on them. There are some on Amazon that are only $5 or so and plug right into the GPIO on the Pi. I used to run it like that, but the power draw was becoming too much. The 24V fan I have right now blowing on all 5 of them does the trick, but I had to split and wire up a DC24V power supply to work with the fan.
"I'd love to change the world, but they wouldn't give me the source code" - Unknown
ID: 1895878 · Report as offensive     Reply Quote
Profile Chris Feyen

Send message
Joined: 1 Jan 01
Posts: 22
Credit: 5,840,550
RAC: 22,457
United States
Message 1896218 - Posted: 19 Oct 2017, 22:00:27 UTC - in response to Message 1895878.  

When I went to setup a 3rd RPI 3, Seti got the 1 confused with the other, and 1 high jacked the backend stats of the other. So I stopped at just the 2 running.
~Chris

ID: 1896218 · Report as offensive     Reply Quote
Williamgcastro

Send message
Joined: 24 Oct 08
Posts: 7
Credit: 6,559,191
RAC: 32,660
United States
Message 1896359 - Posted: 20 Oct 2017, 16:30:43 UTC

Hi Guys

I'm currently running two Pi3s headless with heat sinks and 40x40mm x 10mm cooling fans. Rasbian as the operating system on both and my temps are usually steady in the high 50C range when I bother to check them. They are both running SETI version 8 100% of the time on all 4 cores with a 5 day buffer of work units.

My recent average credit for both has been around 575 and 625 individually for a long time now. I have noticed that the slower Pi has linux 4.4.50 installed on it while the faster on has 4.4.21 but I am uncertain if that explains the difference in RAC.

When I started crunching the PIs seemed to top out at a RAC around 300 using application 8.02. When application 8.06 was released the RAC increased to the current levels. Processing rate increased from 1.62 GFlops to 2.89 GFlops when the application was updated.

Regards

William
ID: 1896359 · Report as offensive     Reply Quote
Tom Rinehart
Volunteer tester

Send message
Joined: 12 Dec 01
Posts: 109
Credit: 1,511,630
RAC: 4,968
United States
Message 1896654 - Posted: 21 Oct 2017, 17:43:22 UTC - in response to Message 1896359.  

When I started crunching the PIs seemed to top out at a RAC around 300 using application 8.02. When application 8.06 was released the RAC increased to the current levels. Processing rate increased from 1.62 GFlops to 2.89 GFlops when the application was updated.


8.06 included additional code that uses the NEON functions on the ARM processors. It looks like it made a big difference on your Pi3s.
ID: 1896654 · Report as offensive     Reply Quote
Profile Eric GiganteProject Donor
Avatar

Send message
Joined: 26 Jun 99
Posts: 5
Credit: 1,782,073
RAC: 8,515
United States
Message 1899091 - Posted: 4 Nov 2017, 14:49:00 UTC - in response to Message 1896654.  

Sorry if I should not be asking this here but start a new post. I figured with the recent talk of version I would ask this here.

I am currently running SETI on two Raspberry Pi's version 2. I have been running one for a while and according to my dashboard on SETI's site it says I am using boinc version 7.4.23.

Recently I had to reformat my SD card in the other Raspberry Pi and I reinstalled boinc. According to my dashboard that version is 7.6.33.
I have ran sudo apt-get update, apt-get upgrade, apt-get dist-upgrade on the older version of boinc and it has not updated the files. Is there anyway I can update boinc without having to reformat the SD card and loose my current work units?
ID: 1899091 · Report as offensive     Reply Quote
Tom Rinehart
Volunteer tester

Send message
Joined: 12 Dec 01
Posts: 109
Credit: 1,511,630
RAC: 4,968
United States
Message 1899103 - Posted: 4 Nov 2017, 15:37:22 UTC - in response to Message 1899091.  

The difference between the two is that the one using boinc 7.4.23 is running Raspbian Jessie. The one running 7.6.33 is running Raspbian Stretch (the latest). It is recommended that you reformat the SD card, but there are also instructions for upgrading. See:

https://www.raspberrypi.org/blog/raspbian-stretch/

Scroll down to How to get Raspbian Stretch
ID: 1899103 · Report as offensive     Reply Quote
Profile Eric GiganteProject Donor
Avatar

Send message
Joined: 26 Jun 99
Posts: 5
Credit: 1,782,073
RAC: 8,515
United States
Message 1900105 - Posted: 9 Nov 2017, 23:57:39 UTC - in response to Message 1899103.  

Thank you. I just wanted to be able to continue using my current work units. Since when you reformat the SD card you also get a new computer on your list of computers. I'm running the update right now.
ID: 1900105 · Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : Number crunching : Anyone still using a raspberry pi 3


 
©2017 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.