Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummers-lowrise
|
Message boards :
Generalized Fermat Prime Search :
Because I am on a weird "let's try this" mode
| Author |
Message |
|---|
|
|
|
GEN21 MT CPU, I am wondering if my 10980XE do better with two units and 9 cores, or one unit and 18 cores?
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 | |
|
 Michael Goetz Volunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14037
ID: 53948
Credit: 477,144,468
RAC: 289,049
                               
|
GEN21 MT CPU, I am wondering if my 10980XE do better with two units and 9 cores, or one unit and 18 cores?
I am trying, and failing, to recall if MT still works on GFN-21.
____________
My lucky number is 75898524288+1 | |
|
MonkeydeeVolunteer tester
Send message
Joined: 8 Dec 13
Posts: 548
ID: 284516
Credit: 1,707,334,225
RAC: 3,253,968
                    
|
|
According to this post from Yves Gallot a GFN21 CPU task should use about 16MB of cache. The 10980XE has 24.75MB total.
So it is likely best to run one task on all 18 cores, and it might be better again to run one task on all 36 threads. Something to try anyway.
____________
My Primes
Badge Score: 4*2 + 6*2 + 7*1 + 8*11 + 9*1 + 11*3 + 12*1 = 169
 | |
|
|
|
I am trying, and failing, to recall if MT still works on GFN-21.
GFN-21 is still advertised on home screen, and there are stats of CPU, so it looks like it to me.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 | |
|
Nick
Send message
Joined: 11 Jul 11
Posts: 2301
ID: 105020
Credit: 10,035,148,120
RAC: 29,345,057
                         
|
|
If it is similar to other ~16 MB tasks - run 2 each of 9 threads.
It should be totally OK to run 2 x 8 thread tasks and keep a couple of threads for yourself. | |
|
Nick
Send message
Joined: 11 Jul 11
Posts: 2301
ID: 105020
Credit: 10,035,148,120
RAC: 29,345,057
|
|
I am going to amend previous - I don't know.
What happens with the task on the GPU?
Does that use CPU cache? | |
|
|
|
|
This does not seem to work
<app_config>
<app>
<app_version>
<name>genefer</name>
<plan_class>cpuGFN21</plan_class>
<cmdline>-nt 18</cmdline>
<avg_ncpus>18</avg_ncpus>
</app_version>
</app>
</app_config>
The error I receive is:
Your app_config.xml file refers to an unknown application ''. Known applications: 'genefer20', 'genefer16', 'genefer17mega', 'ww', 'llrPSP', 'llrTRP', 'llrCUL', 'llrSOB', 'genefer15', 'genefer_extreme', 'llrGCW', 'genefer18', 'genefer'
Which show the genefer name that I am using.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 | |
|
MonkeydeeVolunteer tester
Send message
Joined: 8 Dec 13
Posts: 548
ID: 284516
Credit: 1,707,334,225
RAC: 3,253,968
|
This does not seem to work
<app_config>
<app>
<app_version>
<name>genefer</name>
<plan_class>cpuGFN21</plan_class>
<cmdline>-nt 18</cmdline>
<avg_ncpus>18</avg_ncpus>
</app_version>
</app>
</app_config>
The error I receive is:
Your app_config.xml file refers to an unknown application ''. Known applications: 'genefer20', 'genefer16', 'genefer17mega', 'ww', 'llrPSP', 'llrTRP', 'llrCUL', 'llrSOB', 'genefer15', 'genefer_extreme', 'llrGCW', 'genefer18', 'genefer'
Which show the genefer name that I am using.
Try this one:
<app_config>
<app_version>
<app_name>genefer</app_name>
<cmdline>-nt 4</cmdline>
<plan_class>cpuGFN21</plan_class>
<avg_ncpus>4</avg_ncpus>
</app_version>
</app_config>
That is one I've successfully used in the past. Change the 4's to 18's and you should be good to go.
You may still get an error like the one you've mentioned if you've never run GFN21 on that system before. But that's okay and you can ignore it.
____________
My Primes
Badge Score: 4*2 + 6*2 + 7*1 + 8*11 + 9*1 + 11*3 + 12*1 = 169
| |
|
|
|
Try this one:
<app_config>
<app_version>
<app_name>genefer</app_name>
<cmdline>-nt 4</cmdline>
<plan_class>cpuGFN21</plan_class>
<avg_ncpus>4</avg_ncpus>
</app_version>
</app_config>
That is one I've successfully used in the past. Change the 4's to 18's and you should be good to go.
You may still get an error like the one you've mentioned if you've never run GFN21 on that system before. But that's okay and you can ignore it.
That is the one I had originally, but it downloaded and was running 18 tasks.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 | |
|
|
|
|
Figured it out. Had to go before the GPU ones I had in the list, now it worked. Still downloaded 18 the first time, but says running 18 CPU, so I aborted the other 17 and it should hopefully only grab one at a time afterward.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 | |
|
|
|
|
Running 18 cores, and it is running along well. I finish them faster than the DYFL ones on my GPU. This rig currently is only doing crunching (yes, I know such a high end machine just for crunching, but circumstances...). I am amazed at the speed of this machine and the work it puts out.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 | |
|
|
|
|
Reusing a thread instead of creating a new one.
Curiosity has me wondering why more GFN cannot be run CPU MT? If we have one that can do it, and it runs really well, what makes the others not as desirable to do so? I'd love to be able to push out some of the larger GFN quicker on CPU.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 | |
|
|
|
Reusing a thread instead of creating a new one.
Curiosity has me wondering why more GFN cannot be run CPU MT? If we have one that can do it, and it runs really well, what makes the others not as desirable to do so? I'd love to be able to push out some of the larger GFN quicker on CPU.
Curiosity has me wondering why this has not been answered?
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 | |
|
|
|
|
https://www.primegrid.com/forum_thread.php?id=9821&nowrap=true#153138
Yves Gallot said:
Multithreading is not needed for GFN-17. The data size is 2 MB. On Intel processors the L3 cache size is #cores * 2MB and on AMD CPU it is #cores * 2MB or * 4MB or more.
I guess it's meant to utilize as much cache available on your CPU to improve your throughput. | |
|
|
|
I guess it's meant to utilize as much cache available on your CPU to improve your throughput.
I was looking at 19 and 20 mainly, I know 17 and 18 are fairly quick.
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1
My movie https://vimeo.com/manage/videos/502242 | |
|
Message boards :
Generalized Fermat Prime Search :
Because I am on a weird "let's try this" mode |
