climateprediction.net (CPDN) home page
Thread 'Oifs process statistics question.'

Thread 'Oifs process statistics question.'

Message boards : Number crunching : Oifs process statistics question.
Message board moderation

To post messages, you must log in.

AuthorMessage
Jean-David Beyer

Send message
Joined: 5 Aug 04
Posts: 1120
Credit: 17,202,915
RAC: 2,154
Message 66744 - Posted: 3 Dec 2022, 15:03:39 UTC

I am puzzled about some of the process statistics reported. For example, here is part of the the result of one that completed successfully.

Task 22245062

Server state 	Over
Outcome 	Success
Client state 	Done
Exit status 	0 (0x00000000)
Computer ID 	1511241

Validate state 	Valid
Credit 	0.00
Device peak FLOPS 	6.13 GFLOPS
Application version 	OpenIFS 43r3 Perturbed Surface v1.01
        x86_64-pc-linux-gnu
Peak working set size 	4,600.76 MB
Peak swap size 	4,974.33 MB    <---<<<
Peak disk usage 	1,221.20 MB


My machine has been up for over 11 days and has used, in total, only 164 Megabytes of swap space. On the other hand, it has 15992 MegaBytes of swap space allocated.

Since the "Peak swap size" is only slightly lager than the "Peak working set size" do I infer correctly that this number is only the amount of swap space that would be used if the process were completely swapped out? ... and that none of it has actually been used? I.e., should this line have a better name?
ID: 66744 · Report as offensive     Reply Quote
xii5ku

Send message
Joined: 27 Mar 21
Posts: 79
Credit: 78,322,658
RAC: 1,085
Message 66745 - Posted: 3 Dec 2022, 16:26:41 UTC

This figure most certainly shows peak virtual memory size, not swapped-out a.k.a. paged-out memory.
ID: 66745 · Report as offensive     Reply Quote
Jean-David Beyer

Send message
Joined: 5 Aug 04
Posts: 1120
Credit: 17,202,915
RAC: 2,154
Message 66746 - Posted: 3 Dec 2022, 17:23:29 UTC - in response to Message 66745.  

This figure most certainly shows peak virtual memory size, not swapped-out a.k.a. paged-out memory.


So they should rename it something like peak virtual memory size instead of its current name. But I do not suppose CPDN have anything to do with that name. Probably a Boinc (client) problem.
ID: 66746 · Report as offensive     Reply Quote
Glenn Carver

Send message
Joined: 29 Oct 17
Posts: 1052
Credit: 16,726,716
RAC: 12,672
Message 66747 - Posted: 3 Dec 2022, 17:44:49 UTC - in response to Message 66746.  
Last modified: 3 Dec 2022, 17:45:18 UTC

This figure most certainly shows peak virtual memory size, not swapped-out a.k.a. paged-out memory.
So they should rename it something like peak virtual memory size instead of its current name. But I do not suppose CPDN have anything to do with that name. Probably a Boinc (client) problem.
Correct, that number is the virtual memory size. As you explained nicely, it's the total size required if the process was completely swapped out of memory. It's a confusing choice of name which I think comes from the boinc client but I could be wrong.
ID: 66747 · Report as offensive     Reply Quote
Richard Haselgrove

Send message
Joined: 1 Jan 07
Posts: 1061
Credit: 36,767,175
RAC: 3,168
Message 66749 - Posted: 3 Dec 2022, 18:07:40 UTC - in response to Message 66747.  

ID: 66749 · Report as offensive     Reply Quote
xii5ku

Send message
Joined: 27 Mar 21
Posts: 79
Credit: 78,322,658
RAC: 1,085
Message 66750 - Posted: 3 Dec 2022, 18:38:00 UTC

Both client and server use the misleading nomenclature.
This gets onto the server's result.php web page:
https://github.com/BOINC/boinc/blob/master/html/inc/result.inc#L714

Boincmgr on the other hand says "Virtual memory size" in the [Properties] pop-up window of a task.
ID: 66750 · Report as offensive     Reply Quote
Richard Haselgrove

Send message
Joined: 1 Jan 07
Posts: 1061
Credit: 36,767,175
RAC: 3,168
Message 66751 - Posted: 3 Dec 2022, 18:52:47 UTC - in response to Message 66750.  

Come to that, it's buried pretty deeply in the fundamental design:

https://github.com/BOINC/boinc/blame/master/db/schema.sql#L321
ID: 66751 · Report as offensive     Reply Quote
Jean-David Beyer

Send message
Joined: 5 Aug 04
Posts: 1120
Credit: 17,202,915
RAC: 2,154
Message 66752 - Posted: 3 Dec 2022, 19:19:04 UTC - in response to Message 66747.  

If only this were the most important problem with the Boinc system. Then it would not matter whether they fixed it or not.
ID: 66752 · Report as offensive     Reply Quote

Message boards : Number crunching : Oifs process statistics question.

©2024 cpdn.org