Mac os Catalina

Message boards : Number crunching : Mac os Catalina
Message board moderation

To post messages, you must log in.

AuthorMessage
ugfbill Project Donor

Send message
Joined: 31 Mar 00
Posts: 22
Credit: 110,888,542
RAC: 18
United States
Message 2014592 - Posted: 8 Oct 2019, 3:59:53 UTC

I've just upgraded on one of my crunchers. Upon restart, I looked at Boinc to see what was happening. I saw that many units were held up from processing due to Shared Memory problems.

This web site tells you how to set up Shared memory to prevent these problems.

http://www.spy-hill.net/help/apple/SharedMemory.html

The procedure calls you to use "Emacs". But Emacs is not available in Catalina. So my solution was to update the Shared memory as noted in this document. Before you update. After Update, you can use VIM to create/edit the file needed.

Bill
ID: 2014592 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22115
Credit: 416,307,556
RAC: 380
United Kingdom
Message 2014607 - Posted: 8 Oct 2019, 7:12:44 UTC

You have 7 computers listed - which one are you talking about. When asking questions like this it is useful to say what the SETI "computer id" is so others can quickly focus on that one rather than having to look at all seven, guess wrongly and give you a misleading answer.
Generally shared memory issues are associated with built-in graphics modules, bu not always.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 2014607 · Report as offensive
ugfbill Project Donor

Send message
Joined: 31 Mar 00
Posts: 22
Credit: 110,888,542
RAC: 18
United States
Message 2014619 - Posted: 8 Oct 2019, 11:04:40 UTC - in response to Message 2014607.  

sorry, it's the Mac pro 6,1 ID: 7988896 . Now update. The units are crunching for 7 + hours, instead of the 1.5 hours Before the update.

Bill
ID: 2014619 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 2014626 - Posted: 8 Oct 2019, 15:43:40 UTC - in response to Message 2014619.  

You appear to be using the Old CPU App built in Snow Leopard for Core2 CPUs without SSE 4.1. What happens when you use a newer CPU App such as the AVX App from here, https://arkayn.us/forum/index.php?topic=191.msg4369#msg4369
ID: 2014626 · Report as offensive
ugfbill Project Donor

Send message
Joined: 31 Mar 00
Posts: 22
Credit: 110,888,542
RAC: 18
United States
Message 2014656 - Posted: 8 Oct 2019, 21:40:49 UTC - in response to Message 2014626.  

Thanks much for your assistance. I've downloaded and installed the AVX2+cpur3605 for this client. It has definitely fixed the run time issue I was seeing. Now another question... The multithreading before upgrade to Catalina, used to use all 12 cpu. Now it uses 11. Any thoughts?

Thanks ,
bill
ID: 2014656 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34575
Credit: 261,360,520
RAC: 489
Australia
Message 2014658 - Posted: 8 Oct 2019, 21:55:13 UTC

If you use all 12 cores you will be overcommitting your CPU and that will result in longer run times. ;-)

Cheers.
ID: 2014658 · Report as offensive
ugfbill Project Donor

Send message
Joined: 31 Mar 00
Posts: 22
Credit: 110,888,542
RAC: 18
United States
Message 2014728 - Posted: 9 Oct 2019, 11:02:24 UTC - in response to Message 2014656.  

I must have been premature with the first report. I've installed the AVX2 as noted, and there is still little improvement in the completion. I started overnight, and the units started good, hence my earlier note, but over night, they are at 96% and have been running for 10 hours. These units run in 1.5 or so hours. The main difference is Catalina. I would warn you about upgrading to this new OS.

Something is wrong.
ID: 7988896 Mac Pro 6.1.

Thanks,

Bill
ID: 2014728 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14636
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2014735 - Posted: 9 Oct 2019, 13:11:09 UTC - in response to Message 2014728.  

That sounds like the usual situation when an application fails even to start running properly: BOINC displays 'pseudo-progress' until eventually timing out with a 'maximum time exceeded' error.

I don't know the mantra for checking that on a Mac (of any version): could somebody talk him through checking stderr in the slot directory, please?
ID: 2014735 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 2014737 - Posted: 9 Oct 2019, 13:21:19 UTC - in response to Message 2014735.  

Almost no knowledge on Apple but the WU was manually aborted while are running:

<core_client_version>7.14.2</core_client_version>
<![CDATA[
<message>
aborted by user</message>
<stderr_txt>
Not using mb_cmdline.txt-file, using commandline options.

Build features: SETI8 Non-graphics FFTW FFTOUT JSPF SSSE3 64bit 
 System: Darwin  x86_64  Kernel: 18.7.0
CPU : Intel(R) Xeon(R) CPU E5-1650 v2 @ 3.50GHz 
 GenuineIntel x86, Family 6 Model 62 Stepping 4
 Features : FPU TSC PAE APIC MTRR MMX SSE  SSE2 HT  SSE3 SSSE3 SSE4.1 SSE4.2 AVX1.0  

ar=0.073527  NumCfft=100989  NumGauss=0  NumPulse=147150212041  NumTriplet=147215533171
In v_BaseLineSmooth: NumDataPoints=1048576, BoxCarLength=8192, NumPointsInChunk=32768
OS X optimized setiathome_v8 application
Version info: SSSE3xjf (Intel, Core 2-optimized v8-nographics) V5.13 by Alex Kan
SSSE3xjf OS X 64bit Build 3365 , Ported by : Raistmer, JDWhale, Urs Echternacht

Work Unit Info:
...............
Credit multiplier is :  2.85
WU true angle range is :  0.073527
Triplet: peak=11.53191, time=23.67, period=13.67, d_freq=5354539476.6, chirp=-10.378, fft_len=128 
Autocorr: peak=18.75994, time=62.99, delay=4.3505, d_freq=5354540512.31, chirp=-20.102, fft_len=128k
Spike: peak=24.50605, time=17.18, d_freq=5354543501.52, chirp=22.064, fft_len=128k
Spike: peak=24.65897, time=17.18, d_freq=5354543501.54, chirp=22.065, fft_len=128k
Spike: peak=25.45293, time=17.18, d_freq=5354543501.52, chirp=22.069, fft_len=128k
Spike: peak=25.85124, time=17.18, d_freq=5354543501.54, chirp=22.07, fft_len=128k
Spike: peak=25.13275, time=17.18, d_freq=5354543501.52, chirp=22.074, fft_len=128k
Spike: peak=25.89419, time=17.18, d_freq=5354543501.54, chirp=22.075, fft_len=128k
Spike: peak=24.75974, time=17.18, d_freq=5354543501.54, chirp=22.08, fft_len=128k

</stderr_txt>


Someone else could guide better than me in this case.
ID: 2014737 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14636
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2014740 - Posted: 9 Oct 2019, 13:36:49 UTC - in response to Message 2014737.  

Well, that one obviously got started. I must have checked a slightly different one, but it was similar - found some signals, but not as many as the wingmates which completed successfully. Sounds possibly like it's stalling on pulsefinding?
ID: 2014740 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 2014752 - Posted: 9 Oct 2019, 15:24:46 UTC - in response to Message 2014728.  
Last modified: 9 Oct 2019, 15:28:08 UTC

I must have been premature with the first report. I've installed the AVX2 as noted, and there is still little improvement in the completion. I started overnight, and the units started good, hence my earlier note, but over night, they are at 96% and have been running for 10 hours. These units run in 1.5 or so hours. The main difference is Catalina. I would warn you about upgrading to this new OS.

Something is wrong.
ID: 7988896 Mac Pro 6.1.

Thanks,
Bill
I just checked, that CPU doesn't have AVX2, http://www.cpu-world.com/CPUs/Xeon/Intel-Xeon%20E5-1650%20v2.html As I remember it, the Trash Can Macs only have AVX. Try installing the AVX CPU App and see if that changes anything. The tasks should have been ending in an Error in just a few seconds though, not running for hours. I think the Upgrade didn't go as planned as others running Catalina aren't having any trouble. If the AVX App doesn't fix it, try booting to recovery mode and reinstall the OS. If that doesn't work, rename the app_info.xml to app_info1.xml to go back to stock. Going back to stock will trash ALL your existing tasks, so, I'd try reinstalling the OS first. See if any of that works.
ID: 2014752 · Report as offensive
ugfbill Project Donor

Send message
Joined: 31 Mar 00
Posts: 22
Credit: 110,888,542
RAC: 18
United States
Message 2014773 - Posted: 9 Oct 2019, 18:40:21 UTC - in response to Message 2014752.  
Last modified: 9 Oct 2019, 18:42:17 UTC

Tbar, it looks like there was an issue with the update to Catalina. Something in Catalina didn't like my Boinc installation. So, while bowling this morning, I figured I would go back to basics. I came home - stopped Boinc. renamed the BOINC_data directory. Then uninstalled Boinc.

Next step was to download Boinc,and reinstall.

When it started, it appeared that it was running well now. I had 12 slots filled and running. When looking at Activity monitor, each CPU now shows. I realize that when these units were running forever, there was no listing in the Activity monitor.

All is good now.

Thanks for all your help.

Bill
ID: 2014773 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 2015439 - Posted: 14 Oct 2019, 18:19:58 UTC - in response to Message 2014773.  

It looks like you have a mix of Stock & Anonymous platform tasks on your machines, is there some problem with the AVX CPU App in Catalina? It seems to be working OK on your other machines. If you have a HD5 or higher AMD GPU you need to run the package ATi5r3710&CPU-AVX which has both the GPU App & AVX CPU App. If you need to change the CPU part of the package just add only the CPU App to the GPU package in the setiathome.berkeley.edu folder and change the CPU App name in the two locations where it's named in the app_info.xml. That way you can use the GPU App with any of the three CPU Apps. If you are running a GPU App you need to reserve at least one CPU core to feed the GPU, otherwise both the GPU and CPU will run slower, and you might have screen lag.
ID: 2015439 · Report as offensive

Message boards : Number crunching : Mac os Catalina


 
©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.