Memory requirements for apps of E@h

Raistmer*
Raistmer*
Joined: 20 Feb 05
Posts: 208
Credit: 177642743
RAC: 220383
Topic 222156

I'm still learning current apps behavior and looking through completed results found this one:

 

Run time (sec):8,912.38


CPU time (sec):8,761.12


Peak working set size (MB):451.89


Peak swap size (MB):15881.12


Peak disk usage (MB):4.47


Состояние проверки:Valid


Предоставлено очков:2,000


Приложение:Gravitational Wave search O2 Multi-Directional GPU v2.07 (GW-opencl-nvidia)
x86_64-pc-linux-gnu


 


So, ~15GB of SWAP space was used, really?


Is this normal/usual for this app?


 


It's NV GPU app, btw.


 


EDIT: and here is result for CPU app of same type from same host:


 



Run time (sec):36,855.66

CPU time (sec):28,208.21

Peak working set size (MB):2158.96

Peak swap size (MB):2161.07

Peak disk usage (MB):4.48

Состояние проверки:Valid

Предоставлено очков:1,000

Приложение:Gravitational Wave search O2 Multi-Directional v2.07 (GWnew)
x86_64-pc-linux-gnu


 


Big memory usage, but not 15GB still...


I'm quite curious why GPU app takes so much of SWAP 

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 4701
Credit: 17545549726
RAC: 6411938

You'll probably have to get a

You'll probably have to get a reply from some of the old-timers that really understand the applications.  Maybe Bernd will chime in.

I'm guessing that because the gpu is processing in parallel, that when it comes time to push the interim calculated results back to the cpu for final FP64 processing, that the arrays used in the gpu are much larger in size than the serial arrays used in the cpu application.

Just wild arse guesses on my part.  Take with salt.

 

petri33
petri33
Joined: 4 Mar 20
Posts: 117
Credit: 3341045819
RAC: 0

I'm testing with 3 per gpu,

I'm testing with 3 per gpu, 12 at a time total. Main memory usage is about 8 Gb total including OS, browser etc. The virt column says 22g but my guess is that NV-drivers map some of GPU memory to virtual main memory for the transfers.

 

and

top - 08:33:18 up 7 days, 22:41,  1 user,  load average: 10,41, 9,72, 9,67
Tasks: 359 total,   8 running, 265 sleeping,   0 stopped,   0 zombie
%Cpu(s):  1,3 us,  9,4 sy, 50,3 ni, 38,9 id,  0,1 wa,  0,0 hi,  0,0 si,  0,0 st
KiB Mem : 32795180 total, 12859544 free,  8144252 used, 11791384 buff/cache
KiB Swap:  1003516 total,  1003516 free,        0 used. 23821712 avail Mem 


13616 boinc     30  10 22,771g 423692 131272 R  63,9  1,3  11:16.21 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13635 boinc     30  10 22,774g 426840 130936 S  63,9  1,3  11:10.28 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13644 boinc     30  10 22,763g 415696 131356 R  63,9  1,3  11:12.44 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13607 boinc     30  10 22,710g 393284 131140 R  63,6  1,2  11:13.75 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13626 boinc     30  10 22,766g 418592 131420 R  63,2  1,3  11:11.62 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13653 boinc     30  10 22,766g 418984 131124 R  62,3  1,3  11:06.17 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13571 boinc     30  10 19,017g 440440 139576 S  54,0  1,3   9:35.56 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13566 boinc     30  10 18,998g 420460 139508 R  53,6  1,3   9:39.09 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13569 boinc     30  10 18,998g 419640 139392 S  53,6  1,3   9:37.34 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13580 boinc     30  10 18,808g 381332 105212 R  42,4  1,2   7:44.82 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13589 boinc     30  10 18,847g 389024 104944 R  42,1  1,2   7:41.30 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.07_x+
13598 boinc     30  10 18,845g 387500 105108 R  41,7  1,2   7:37.56 ../../projects/einstein.phys.uwm.edu/einstein_O2MDF_2.0
Raistmer*
Raistmer*
Joined: 20 Feb 05
Posts: 208
Credit: 177642743
RAC: 220383

So you think those 15GB are

So you think those 15GB are reserved but not committed ones?

Would be interesting to try the same under Windows, where I would have additional means for analysis.

 

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.