Amicable Numbers has slowed down

Message boards : Number crunching : Amicable Numbers has slowed down

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
Allen Paschke

Send message
Joined: 27 Jan 18
Posts: 23
Credit: 8,750,328
RAC: 2,431
   
Message 959 - Posted: 31 Oct 2018, 16:24:26 UTC

I am running Amicable Numbers on 3 different computers, all with Microsoft Windows 10 Professional x64, in CPU Mode with 3 CPUs selected. Each computer has 4 Intel i5 processors. Historically, each task would take 10 - 12 hours to complete. Starting a week or two ago, each task started taking 20+ hours to complete. I did upgrade from BOINC 7.12.1 to BOINC 7.14.1, but I've made no other software changes to any of the 3 PCs. This seems suspicious that all 3 computers would run Amicable Numbers slower.
- Have other people has similar experiences?
- Have there been any changes to Amicable Numbers?
- Are there any parameters which I can control, so Amicable numbers will run faster.
ID: 959 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Brent

Send message
Joined: 6 Apr 18
Posts: 12
Credit: 20,200,952
RAC: 0
  
Message 960 - Posted: 31 Oct 2018, 17:24:38 UTC - in response to Message 959.  

Same here. I am running Amicable Numbers CPU mode with 7 CPUs selected. This is on a Dell XPS 8900 PC with INTEL Core i7-6700 @ 3.40GHz, Windows 10 Home on 1 PC. I was processing 13-15 units a day but am now down to 7-10 per day. The main change was a recent upgrade to the latest Boinc software, but the slowdown started before this upgrade.
ID: 960 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Brent

Send message
Joined: 6 Apr 18
Posts: 12
Credit: 20,200,952
RAC: 0
  
Message 961 - Posted: 31 Oct 2018, 17:32:38 UTC - in response to Message 960.  

I should add, the slowdown began around 10/20/2018
ID: 961 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Allen Paschke

Send message
Joined: 27 Jan 18
Posts: 23
Credit: 8,750,328
RAC: 2,431
   
Message 962 - Posted: 31 Oct 2018, 18:26:53 UTC - in response to Message 961.  

I agree with you, 20 October 2018 sounds about right. Yes, it was before BOINC 7.14.2.
ID: 962 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
bryan

Send message
Joined: 7 Nov 17
Posts: 14
Credit: 87,038,417
RAC: 0
   
Message 963 - Posted: 31 Oct 2018, 21:12:07 UTC

The applications were last published in September https://sech.me/boinc/Amicable/apps.php

You can see how long they are recorded to take at Tasks here https://sech.me/boinc/Amicable/home.php , and then see if there is an increase.

The only thing I've noticed is the number of new discoveries has for me, been non-existent for a time- explained in another thread.
ID: 963 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sergei Chernykh
Project administrator
Project developer

Send message
Joined: 5 Jan 17
Posts: 506
Credit: 72,451,573
RAC: 0
   
Message 964 - Posted: 1 Nov 2018, 8:33:20 UTC

My only guess is as numbers grow larger, internal buffers don't fit in CPU cache anymore. Hence the slowdown.
ID: 964 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
bryan

Send message
Joined: 7 Nov 17
Posts: 14
Credit: 87,038,417
RAC: 0
   
Message 968 - Posted: 2 Nov 2018, 18:35:44 UTC

Oh, if that bottleneck may be happening in primesieve the new versions has some more tweaks for that - (although last time they didn't have a noticeable impact)

https://github.com/kimwalisch/primesieve/blob/master/ChangeLog

Specifically 7.1:
primesieve-7.1 runs up to 30% faster on Intel Skylake-X CPUs!
The default sieve size is now (L2 cache size / 2). Using a sieve size
that is slightly smaller than the L2 cache size reduces the number
of L2 cache misses which improves performance on CPUs with slow L3
caches. primesieve-7.1 will also run slightly faster (< 3%) on most
other Intel CPUs.

Aside: is primesive used on OpenCL? I see that hasn't been updated on that branch for some time.
ID: 968 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sergei Chernykh
Project administrator
Project developer

Send message
Joined: 5 Jan 17
Posts: 506
Credit: 72,451,573
RAC: 0
   
Message 970 - Posted: 2 Nov 2018, 19:21:36 UTC - in response to Message 968.  

Oh my... I guess I'll need to update primesieve in all versions soon. This month probably because it's also used in OpenCL versions and will become a bottleneck as numbers grow larger.
ID: 970 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Brent

Send message
Joined: 6 Apr 18
Posts: 12
Credit: 20,200,952
RAC: 0
  
Message 972 - Posted: 2 Nov 2018, 19:59:57 UTC - in response to Message 964.  

Gee! If the numbers growing larger is causing the slowdown, way aren't the credits also growing larger. If it is taking twice as long to process a unit, shouldn't the credits be twice as much?
ID: 972 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sergei Chernykh
Project administrator
Project developer

Send message
Joined: 5 Jan 17
Posts: 506
Credit: 72,451,573
RAC: 0
   
Message 973 - Posted: 2 Nov 2018, 21:15:05 UTC - in response to Message 972.  

All CPU tasks should be balanced to roughly the same execution time, they were generated with primesieve performance taken into account. it's only OpenCL version that will slowdown eventually.
ID: 973 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
ccpmc

Send message
Joined: 28 May 18
Posts: 1
Credit: 34,447,580
RAC: 0
  
Message 976 - Posted: 15 Nov 2018, 15:50:08 UTC
Last modified: 15 Nov 2018, 15:50:45 UTC

I'm running Amicable Numbers in a machine which can only calculating about 8 hours per day.
But a workunit cost more than 20 hours.
So the period is too short for me to complete the workunit.

I think the period for workunits should be more longer.
ID: 976 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Brent

Send message
Joined: 6 Apr 18
Posts: 12
Credit: 20,200,952
RAC: 0
  
Message 977 - Posted: 15 Nov 2018, 17:26:00 UTC - in response to Message 973.  

All CPU tasks should be balanced to roughly the same execution time, they were generated with primesieve performance taken into account. it's only OpenCL version that will slowdown eventually.


Ok, so how do I avoid OpenCL so I am not wasting my computer time and electricity for less work?
ID: 977 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Matt Kowal
Avatar

Send message
Joined: 12 Feb 17
Posts: 14
Credit: 177,829,581
RAC: 0
   
Message 978 - Posted: 17 Nov 2018, 0:13:09 UTC - in response to Message 977.  
Last modified: 17 Nov 2018, 0:13:45 UTC

Ok, so how do I avoid OpenCL so I am not wasting my computer time and electricity for less work?


OpenCL tasks on this project are for GPUs. Your system (https://sech.me/boinc/Amicable/results.php?hostid=36634) is only utilizing the CPU on this project, so no changes are needed.
ID: 978 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Brent

Send message
Joined: 6 Apr 18
Posts: 12
Credit: 20,200,952
RAC: 0
  
Message 979 - Posted: 17 Nov 2018, 7:23:34 UTC - in response to Message 978.  

Ok, so how do I avoid OpenCL so I am not wasting my computer time and electricity for less work?


OpenCL tasks on this project are for GPUs. Your system (https://sech.me/boinc/Amicable/results.php?hostid=36634) is only utilizing the CPU on this project, so no changes are needed.


Then why am I experiencing the slowdown if only the Open CL tasks should be affected. I am spending almost the same computer time and electricity to process almost half the workunits. And others are reporting the same issue which began about Oct. 20.
ID: 979 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Jozef J

Send message
Joined: 24 Jan 17
Posts: 20
Credit: 1,180,716,010
RAC: 447
   
Message 981 - Posted: 17 Nov 2018, 19:12:58 UTC
Last modified: 17 Nov 2018, 19:23:45 UTC

hi, i see some slowing down on gpu nvidia 2080Ti ,,, on same setup and all... in win10... i see about 10 sec. more per task
most proably new nvidia drivers after 400.0 i have last always . now 416.94
Or increasing app diff.??'
I ask you for share your experience .
thank you
edit: i have also always disabled HW accelr. in web browser , clean OS from "bad bugs"
now i check task run time and i find some with "old good time" 85-92 second
but most of all other task have now 95-110 ..
Is it diff increase or nvidia..? that is big todays question..) windows 10 1803..
ID: 981 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Kellen

Send message
Joined: 14 Nov 17
Posts: 70
Credit: 1,000,005,236
RAC: 0
   
Message 982 - Posted: 18 Nov 2018, 0:56:01 UTC - in response to Message 981.  

Hi Jozef,

As we approach the end of the project the CPU time for each task will continue to increase, and for that reason you will need to run more than 1 task per GPU to reach full utilization. I switched to 2 concurrent tasks on my RTX 2070 today as the GPU utilization was down to 94%. The utilization went up to 99%-100% and now I am doing two tasks every 295 seconds instead of 1 task every 153 seconds, or a 3.7% increase in throughput.

I looked at your tasks and you can see that the CPU time and the Run Time for most of them are the same. This means that the Run Time is being limited by the CPU and not using the GPU completely.

To run two tasks concurrently, make a file called app_config.xml (make a text file, then change the extension to .xml from .txt) in your Amicable Numbers data directory and put the following text in it;

<app_config>
<app>
<name>amicable_10_20</name>
<gpu_versions>
<gpu_usage>0.5</gpu_usage>
<cpu_usage>1</cpu_usage>
</gpu_versions>
</app>
</app_config>

Once that is in there, select the Options menu in BOINC and click "Read Config Files" and two units should start running at the same time. Each unit will get a full CPU core and half of the GPU.

Once CPU usage becomes about 80-90% of the total run time of a task, then the GPU utilization will probably start to decrease. If that happens, then you can change the <gpu_usage> line to make more tasks run. 0.33 would be 3 tasks, 0.25 would be 4 tasks, 0.2 would be 5 tasks, etc. Just be aware that each task will use additional memory. At this time, each task is approximately 1400MB GPU memory, so 5 tasks would be 7GB.

Based on the potential total performance of your GPU compared to mine, 2 tasks might not be enough to fully utilize the RTX 2080ti and you may have to run 3 to get full use. The number of required tasks will continue to increase until the end of the project.

Hope this helps.

Regards,
Kellen
ID: 982 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Jozef J

Send message
Joined: 24 Jan 17
Posts: 20
Credit: 1,180,716,010
RAC: 447
   
Message 983 - Posted: 18 Nov 2018, 2:44:50 UTC

i remember just few day ago was utilizing on my 2080ti about 94-96% . But now is 85% . Booth on kernel size 23 ..
So is maybe time yo use kernel 24 , as we before discus, also some people with nv tesla V 100 need higher kernel.to get more from gpu.-.
ID: 983 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Kellen

Send message
Joined: 14 Nov 17
Posts: 70
Credit: 1,000,005,236
RAC: 0
   
Message 984 - Posted: 18 Nov 2018, 3:28:12 UTC - in response to Message 983.  

Hi Jozef,

I do not think that the kernel size will help anymore. I have not gone through all of the code in detail, however it is my understanding that the issue now is that the CPU must determine all of the prime numbers over a very large range of values and that range of values will continue to get larger towards the end of the project. Sergei would be able to confirm it, but I suspect that the only way to increase GPU utilization from now on will be to run more tasks concurrently.

Running multiple tasks will solve the problem for now and Sergei may have other solutions to the low utilization, but I am not familiar enough with the code to figure out anything else.

Regards,
Kellen
ID: 984 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Jozef J

Send message
Joined: 24 Jan 17
Posts: 20
Credit: 1,180,716,010
RAC: 447
   
Message 985 - Posted: 18 Nov 2018, 11:28:36 UTC

"make a text file, then change the extension to .xml from .txt" its not work in win 10
I try change but "file" always stay as "text file" and boinc manager not read this config file.. i try many times in other projects
did you try changing " extension" with "open in some program".. ? but this not work
can you put edited app_config.xml on some (megaupl) and share ..?
ID: 985 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Kellen

Send message
Joined: 14 Nov 17
Posts: 70
Credit: 1,000,005,236
RAC: 0
   
Message 986 - Posted: 18 Nov 2018, 13:20:02 UTC - in response to Message 985.  

Hi Jozef,

I am not home right now so I can't upload a file, but it sounds like the issue you are having is probably related to the file extensions being hidden and that you have a text file with ".xml" in the name, rather than having an .xml file. To fix this; in any folder window in Windows 10 click the "View" tab and make sure that the boxes for "File Name Extensions" and "Hidden Items" are checked. This will let you see the file extension so that you can change it to app_config.xml and also so that you can see the Program Data folder to get to the Amicable Numbers data directory, which should be located at C:/ProgramData/BOINC/projects/sech.me if I remember correctly.

In case my instructions are poor, I found a link to show what to click on to show file extensions.

http://kb.winzip.com/kb/entry/26/

Regards,
Kellen
ID: 986 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
1 · 2 · 3 · Next

Message boards : Number crunching : Amicable Numbers has slowed down


©2024 Sergei Chernykh