software priority

[SandStar]

I'm currently running version 25.11 (64bit) on linux and windows machines.
The process priority on both platforms seems wrong?

The linux version runs at nice 10 instead of 19.
The windows version runs at normal priority instead of lowest.

Was this intentionally and/or can this behaviour be altered? I'm having various issues with the software as it's stealing cpu cycles from more critical processes.

[TimSorbet]

I believe you are seeing the priorities of the thread handling the interface, not the thread(s) that run the actual work. This uses practically no CPU time, but is very important for responsiveness, so having it at a normal priority is a good thing, not a problem.
I know more details for Windows, (and not Linux) so I'll talk about that.

Originally Posted by SandStar

The windows version runs at normal priority instead of lowest.

Was this intentionally and/or can this behaviour be altered? I'm having various issues with the software as it's stealing cpu cycles from more critical processes.

What is the priority setting in Test > Worker Windows > Priority? 1 is the lowest, which is idle. If it's at this level, it shouldn't be taking any CPU time that other apps need. (assuming they aren't also at such a low priority, of course)

[SandStar]

Originally Posted by Mini-Geek

I believe you are seeing the priorities of the thread handling the interface, not the thread(s) that run the actual work. This uses practically no CPU time, but is very important for responsiveness, so having it at a normal priority is a good thing, not a problem.
I know more details for Windows, (and not Linux) so I'll talk about that.

What is the priority setting in Test > Worker Windows > Priority? 1 is the lowest, which is idle. If it's at this level, it shouldn't be taking any CPU time that other apps need. (assuming they aren't also at such a low priority, of course)

The priority is set at "1" but when I start taskmanager the actual priority is set to "normal".

As for the linux version:

top - 19:42:44 up 3 days, 19:58,  5 users,  load average: 2.28, 3.10, 3.51
Tasks: 160 total,   1 running, 158 sleeping,   0 stopped,   1 zombie
Cpu(s):  3.7%us,  2.5%sy, 88.2%ni,  5.1%id,  0.1%wa,  0.2%hi,  0.2%si,  0.0%st
Mem:   4057648k total,  4033268k used,    24380k free,    16344k buffers
Swap:   506036k total,      100k used,   505936k free,   539524k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 9883 erwin     30  10 1727m 1.5g  864 S  381 39.6   2:19.39 mprime

[joblack]

Originally Posted by SandStar

The priority is set at "1" but when I start taskmanager the actual priority is set to "normal".

As for the linux version:

top - 19:42:44 up 3 days, 19:58,  5 users,  load average: 2.28, 3.10, 3.51
Tasks: 160 total,   1 running, 158 sleeping,   0 stopped,   1 zombie
Cpu(s):  3.7%us,  2.5%sy, 88.2%ni,  5.1%id,  0.1%wa,  0.2%hi,  0.2%si,  0.0%st
Mem:   4057648k total,  4033268k used,    24380k free,    16344k buffers
Swap:   506036k total,      100k used,   505936k free,   539524k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 9883 erwin     30  10 1727m 1.5g  864 S  381 39.6   2:19.39 mprime

You only see the main thread which is responsible for administration and handling of the worker threads.

Btw. a priority of 30 is less then normal (nice level +10) so it should be no problem anyway.

[SandStar]

Originally Posted by joblack

You only see the main thread which is responsible for administration and handling of the worker threads.

Btw. a priority of 30 is less then normal (nice level +10) so it should be no problem anyway.

You're right. The threads themselves run at nice level 19.
I've had problems with the previous mprime version; the audio would stutter constantly. That seems solved in the latest version. I just looked at the process priority and thought the new version didnt solve it. *oops*

I guess the same can be applied to the windows version: the main thread runs at normal priority and the worker threads do run at idle priority.
That still leaves me with the problem that my mailserver is heavily affected whenever I start the client. Mail delivery goes from 1 second to 30 minutes. But only outgoing, incoming mail is unaffected.

I'll guess I have to look into it a bit more

Thanks for the help!

[mdettweiler]

Originally Posted by SandStar

That still leaves me with the problem that my mailserver is heavily affected whenever I start the client. Mail delivery goes from 1 second to 30 minutes. But only outgoing, incoming mail is unaffected.

What priority is your mail server at? If it's at priority 19 too, then it will be stomped all over by Prime95, which would cause what you're seeing. If that's the case, then you could try setting the mailserver to a higher priority, which should give it the proper precedence.

[SandStar]

Originally Posted by mdettweiler

What priority is your mail server at? If it's at priority 19 too, then it will be stomped all over by Prime95, which would cause what you're seeing. If that's the case, then you could try setting the mailserver to a higher priority, which should give it the proper precedence.

The mail server is a Windows 2008 server running Exchange 2007.

I think I've narrowed the problem down to Symantec. Currently updating it to the latest version. It seems the Exchange scanning part is not being friends with Prime95.

[Prime95]

Originally Posted by SandStar

I think I've narrowed the problem down to Symantec. Currently updating it to the latest version. It seems the Exchange scanning part is not being friends with Prime95.

Does that mean mail won't go out until it is scanned by Symantec AND Symantec is running at idle priority?

[lfm]

Originally Posted by SandStar

The priority is set at "1" but when I start taskmanager the actual priority is set to "normal".

As for the linux version:

top - 19:42:44 up 3 days, 19:58,  5 users,  load average: 2.28, 3.10, 3.51
Tasks: 160 total,   1 running, 158 sleeping,   0 stopped,   1 zombie
Cpu(s):  3.7%us,  2.5%sy, 88.2%ni,  5.1%id,  0.1%wa,  0.2%hi,  0.2%si,  0.0%st
Mem:   4057648k total,  4033268k used,    24380k free,    16344k buffers
Swap:   506036k total,      100k used,   505936k free,   539524k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 9883 erwin     30  10 1727m 1.5g  864 S  381 39.6   2:19.39 mprime

In "top" use the H command to see the individual threads. The compute bound thread(s) will then show up at their actual nice level instead of the level of the main thread.

[joblack]

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3448 joblack 39 19 745m 574m 3308 R 98.4 7.2 1948:52 mprime
3447 joblack 39 19 745m 574m 3308 R 96.7 7.2 1940:48 mprime
3564 root 20 0 582m 405m 52m R 62.5 5.1 338:09.55 Xorg
3446 joblack 39 19 745m 574m 3308 R 45.9 7.2 1297:14 mprime
3527 joblack 39 19 745m 574m 3308 R 45.5 7.2 1297:37 mprime

You're right it's LVL 39 which is normally no problem even if you have a mail server running.

[SandStar]

Originally Posted by Prime95

Does that mean mail won't go out until it is scanned by Symantec AND Symantec is running at idle priority?

Weirdest thing ever. It's solved after I upgraded to a new version (Symantec).
I'm still not sure what exactly happened but if I interpret my logs correctly it was a combination of Symantec exchange scanning, prime95 and DNS lookups. The DNS lookups took too long and caused Exchange to stop processing the mail and retry in 30min. I started looking for something like that when I found that all mail was exactly 30min late. Never 31 or 29 but always exact 30min. Since the internal DNS was not only used by Exchange but also various clients that did not have any issues with resolving, I don't think prime95 was interfering with Windows's DNS service.

Anyway, my head hurts when trying to figure out what went wrong. Important thing is it's fixed.
Thanks for all your help, couldnt have done it without your help! :)