Hello All:

I have run into a particularly strange problem. I have been running my work units with one instance of the GPU (AMD Radeon RX560) and have had no problems. However, I decided to run multiple instances by way of the app_config.xml file, put into my Milkyway directory, under Boinc.

This is working fine on two other machines with the Radeon processors, but I have been having no luck on the other one.
Trouble machine is running thte latest Boinc software and is running under Windows 7. My app_config.xml file looks like this:

<app_config>
<app>
<name>milkyway</name>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>0.9</cpu_usage>
</gpu_versions>
</app>
</app_config>

The CPU is an AMD FX 8300, 8 cores.

Like I said, it is running one instance just fine, but it doesn't seem to want to acknowledge the app_config file. I have used the Options, run config files and the Event Log shows that it sees it, but not the program.

Any help, greatly appreciated!

Allen

My "app_config.xml" is working fine under Win7.
I used "text document" to edit it.

I have placed it in Boinc's ProgramData file into "projects > milkyway.cs.rpi.edu_milkyway"

Your "app_config.xml" syntax looks fine to me.

Cheers S-F-V

Hello All:

I have run into a particularly strange problem. I have been running my work units with one instance of the GPU (AMD Radeon RX560) and have had no problems. However, I decided to run multiple instances by way of the app_config.xml file, put into my Milkyway directory, under Boinc.

This is working fine on two other machines with the Radeon processors, but I have been having no luck on the other one.
Trouble machine is running thte latest Boinc software and is running under Windows 7. My app_config.xml file looks like this:

<app_config>
<app>
<name>milkyway</name>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>0.9</cpu_usage>
</gpu_versions>
</app>
</app_config>

The CPU is an AMD FX 8300, 8 cores.

Like I said, it is running one instance just fine, but it doesn't seem to want to acknowledge the app_config file. I have used the Options, run config files and the Event Log shows that it sees it, but not the program.

Any help, greatly appreciated!

Allen

AS SFV said the syntax is fine so it's probably that you are using a word processing program instead of Notepad and/or not saving it as the right kind of file, text type is correct. Word processing programs add stuff to the beginning and ends of the lines that causes Boinc to ignore them and the file won't work. Copy and paste what you posted into Notepad in Windows and then save it and tell Boinc to reread it and see if it works for you.

Hello All:

I have run into a particularly strange problem. I have been running my work units with one instance of the GPU (AMD Radeon RX560) and have had no problems. However, I decided to run multiple instances by way of the app_config.xml file, put into my Milkyway directory, under Boinc.

This is working fine on two other machines with the Radeon processors, but I have been having no luck on the other one.
Trouble machine is running thte latest Boinc software and is running under Windows 7. My app_config.xml file looks like this:

<app_config>
<app>
<name>milkyway</name>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>0.9</cpu_usage>
</gpu_versions>
</app>
</app_config>

The CPU is an AMD FX 8300, 8 cores.

Like I said, it is running one instance just fine, but it doesn't seem to want to acknowledge the app_config file. I have used the Options, run config files and the Event Log shows that it sees it, but not the program.

Any help, greatly appreciated!

Allen

AS SFV said the syntax is fine so it's probably that you are using a word processing program instead of Notepad and/or not saving it as the right kind of file, text type is correct. Word processing programs add stuff to the beginning and ends of the lines that causes Boinc to ignore them and the file won't work. Copy and paste what you posted into Notepad in Windows and then save it and tell Boinc to reread it and see if it works for you.

Thanks to both of you for responding. I did use Notepad to create the file. That is what is so strange. Guess I need to recheck everything and try again.
Thanks!

Allen

Just one other thing.... when files are downloaded to Boinc Mgr, the gpu files show that Boinc is recognizing the app_config file, since the GPU work units say .5 cpu and .33 gpu, yet it will only run 2 at a time.

Now any new ideas or possibilities?

Allen

Just wanted to say that I am not running under any host(Home, Work, School, etc.

Any thing just jump out at cha? Oh, BTW, I am now down to only being able to run one work unit under the app_config.xml.
What the hell did I do?????

Allen

Just one other thing.... when files are downloaded to Boinc Mgr, the gpu files show that Boinc is recognizing the app_config file, since the GPU work units say .5 cpu and .33 gpu, yet it will only run 2 at a time.

Now any new ideas or possibilities?

Allen

Ok when you go into the Boinc Manager and click Options, read config files what does the Event Log say about Milkyway? Does it just say 'found app_config file' or does it say something else? For instance when I have it read the config files for me under Einstein it says "9/12/2022 9:50:45 PM | Einstein@Home | Config: excluded GPU. Type: ATI. App: all. Device: all" which is what I want it to say, all the rest say ie "9/12/2022 9:50:45 PM | Milkyway@Home | Found app_config.xml"

Just one other thing.... when files are downloaded to Boinc Mgr, the gpu files show that Boinc is recognizing the app_config file, since the GPU work units say .5 cpu and .33 gpu, yet it will only run 2 at a time.

Now any new ideas or possibilities?

Allen

... are you (hopefully) using

<app_config>
<project_max_concurrent>x</project_max_concurrent>
...

or
<app>
<max_concurrent>x</max_concurrent>
...

where "x" is the number of "things" you want to do at the same time?

S-F-V

Just one other thing.... when files are downloaded to Boinc Mgr, the gpu files show that Boinc is recognizing the app_config file, since the GPU work units say .5 cpu and .33 gpu, yet it will only run 2 at a time.

Now any new ideas or possibilities?

Allen

Ok when you go into the Boinc Manager and click Options, read config files what does the Event Log say about Milkyway? Does it just say 'found app_config file' or does it say something else? For instance when I have it read the config files for me under Einstein it says "9/12/2022 9:50:45 PM | Einstein@Home | Config: excluded GPU. Type: ATI. App: all. Device: all" which is what I want it to say, all the rest say ie "9/12/2022 9:50:45 PM | Milkyway@Home | Found app_config.xml"

Hi Mikey,

It says the following....

9/13/2022 1:22:55 PM | | Re-reading cc_config.xml
9/13/2022 1:22:55 PM | | cc_config.xml not found - using defaults
9/13/2022 1:22:55 PM | | log flags: file_xfer, sched_ops, task
9/13/2022 1:22:55 PM | Milkyway@Home | Found app_config.xml

Also, local preferances are:

9/13/2022 1:35:52 PM | Milkyway@Home | General prefs: from Milkyway@Home (last modified 12-Sep-2022 16:27:18)
9/13/2022 1:35:52 PM | Milkyway@Home | Host location: none
9/13/2022 1:35:52 PM | Milkyway@Home | General prefs: using your defaults
9/13/2022 1:35:52 PM | | Reading preferences override file
9/13/2022 1:35:52 PM | | Preferences:
9/13/2022 1:35:52 PM | | max memory usage when active: 4078.57 MB
9/13/2022 1:35:52 PM | | max memory usage when idle: 8157.14 MB
9/13/2022 1:35:52 PM | | max disk usage: 20.00 GB
9/13/2022 1:35:52 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)

Hi SFV,

I'm using.....

<app_config>
<app>
<name>milkyway</name>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>0.9</cpu_usage>
</gpu_versions>
</app>
</app_config>

I am also using no location, such as Home, School etc.

Just one other thing.... when files are downloaded to Boinc Mgr, the gpu files show that Boinc is recognizing the app_config file, since the GPU work units say .5 cpu and .33 gpu, yet it will only run 2 at a time.

Now any new ideas or possibilities?

Allen

Ok when you go into the Boinc Manager and click Options, read config files what does the Event Log say about Milkyway? Does it just say 'found app_config file' or does it say something else? For instance when I have it read the config files for me under Einstein it says "9/12/2022 9:50:45 PM | Einstein@Home | Config: excluded GPU. Type: ATI. App: all. Device: all" which is what I want it to say, all the rest say ie "9/12/2022 9:50:45 PM | Milkyway@Home | Found app_config.xml"

Hi Mikey,

It says the following....

9/13/2022 1:22:55 PM | | Re-reading cc_config.xml
9/13/2022 1:22:55 PM | | cc_config.xml not found - using defaults
9/13/2022 1:22:55 PM | | log flags: file_xfer, sched_ops, task
9/13/2022 1:22:55 PM | Milkyway@Home | Found app_config.xml

Also, local preferances are:

9/13/2022 1:35:52 PM | Milkyway@Home | General prefs: from Milkyway@Home (last modified 12-Sep-2022 16:27:18)
9/13/2022 1:35:52 PM | Milkyway@Home | Host location: none
9/13/2022 1:35:52 PM | Milkyway@Home | General prefs: using your defaults
9/13/2022 1:35:52 PM | | Reading preferences override file
9/13/2022 1:35:52 PM | | Preferences:
9/13/2022 1:35:52 PM | | max memory usage when active: 4078.57 MB
9/13/2022 1:35:52 PM | | max memory usage when idle: 8157.14 MB
9/13/2022 1:35:52 PM | | max disk usage: 20.00 GB
9/13/2022 1:35:52 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)

I don't see anything wrong with that except maybe the 'memory when active' part, I don't think MW tasks use that much but it may depend on what else you are running at the same time. ie WCG ARP tasks take almost 1gb of memory per task so you would use that up very quickly limiting other tasks that could run. If you click on a running task in the Boinc Manager and then go to the left you will see a box that Properties, click on it and it will show you 2 things for the memory that task is using, Virtual Memory Size and Working set size. You care about the one that is the biggest as that's how much your system is reserving for that particular task.

alk44 AKA Allen:

I meant please try adding the following to your <app_config>:

<max_concurrent>3</max_concurrent>

between

<name> ...

and

<gpu_versions> ...

That helped me!

------------------------------------------------------------------------------------

BTW: As far as I can remeber, some changes to
<gpu_usage>
and/or to
<cpu_usage>
will only come into effect AFTER ALL of your "old" tasks are finished.

SFV

alk44 AKA Allen:

BTW: As far as I can remeber, some changes to
<gpu_usage>
and/or to
<cpu_usage>
will only come into effect AFTER ALL of your "old" tasks are finished.

SFV

Yes it applies to NEW tasks not existing tasks, but once you get those tasks you have to wait for them to be gone before the next change takes effect. So you might want to go to a zero resource share while testing and just wait out the 10 minutes inbetween tasks.

Just one other thing.... when files are downloaded to Boinc Mgr, the gpu files show that Boinc is recognizing the app_config file, since the GPU work units say .5 cpu and .33 gpu, yet it will only run 2 at a time.

Now any new ideas or possibilities?

Allen

Ok when you go into the Boinc Manager and click Options, read config files what does the Event Log say about Milkyway? Does it just say 'found app_config file' or does it say something else? For instance when I have it read the config files for me under Einstein it says "9/12/2022 9:50:45 PM | Einstein@Home | Config: excluded GPU. Type: ATI. App: all. Device: all" which is what I want it to say, all the rest say ie "9/12/2022 9:50:45 PM | Milkyway@Home | Found app_config.xml"

Hi Mikey,

It says the following....

9/13/2022 1:22:55 PM | | Re-reading cc_config.xml
9/13/2022 1:22:55 PM | | cc_config.xml not found - using defaults
9/13/2022 1:22:55 PM | | log flags: file_xfer, sched_ops, task
9/13/2022 1:22:55 PM | Milkyway@Home | Found app_config.xml

Also, local preferances are:

9/13/2022 1:35:52 PM | Milkyway@Home | General prefs: from Milkyway@Home (last modified 12-Sep-2022 16:27:18)
9/13/2022 1:35:52 PM | Milkyway@Home | Host location: none
9/13/2022 1:35:52 PM | Milkyway@Home | General prefs: using your defaults
9/13/2022 1:35:52 PM | | Reading preferences override file
9/13/2022 1:35:52 PM | | Preferences:
9/13/2022 1:35:52 PM | | max memory usage when active: 4078.57 MB
9/13/2022 1:35:52 PM | | max memory usage when idle: 8157.14 MB
9/13/2022 1:35:52 PM | | max disk usage: 20.00 GB
9/13/2022 1:35:52 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)

I don't see anything wrong with that except maybe the 'memory when active' part, I don't think MW tasks use that much but it may depend on what else you are running at the same time. ie WCG ARP tasks take almost 1gb of memory per task so you would use that up very quickly limiting other tasks that could run. If you click on a running task in the Boinc Manager and then go to the left you will see a box that Properties, click on it and it will show you 2 things for the memory that task is using, Virtual Memory Size and Working set size. You care about the one that is the biggest as that's how much your system is reserving for that particular task.

Mikey,
Yep, the memory thing says it's using a hundred or so megabytes, so that can't be a problem.
Virtual memory size
95.52 MB
Working set size
84.19 MB

Still working on it.

alk44 AKA Allen:

I meant please try adding the following to your <app_config>:

<max_concurrent>3</max_concurrent>

between

<name> ...

and

<gpu_versions> ...

That helped me!

------------------------------------------------------------------------------------

BTW: As far as I can remeber, some changes to
<gpu_usage>
and/or to
<cpu_usage>
will only come into effect AFTER ALL of your "old" tasks are finished.

SFV

Okay, I added your line to the app_config.xml and then read the config file and got nothing. I know what you said about having to run out of previous units before the new units would take hold of the new changes, so I tried temporarily pausing all of the old ones and then re-reading the config file and then doing an update, without getting results from that tactic either. I guess I will just wait an see what happens for now.

Thanks!

[

After reading config files, any NEW downloaded tasks will be identified by the new usage descriptor. So sort your tasks by received date in your cache and you see the new tasks with your 0.33 gpu usage.

---

BTW: As far as I can remeber, some changes to
<gpu_usage>
and/or to
<cpu_usage>
will only come into effect AFTER ALL of your "old" tasks are finished.

SFV

Okay, I added your line to the app_config.xml and then read the config file and got nothing. I know what you said about having to run out of previous units before the new units would take hold of the new changes, so I tried temporarily pausing all of the old ones and then re-reading the config file and then doing an update, without getting results from that tactic either. I guess I will just wait an see what happens for now.

Thanks!

One option is to abort all your existing tasks and then get new tasks, personally I would do that but also change the resource share to zero so you only get one task at a time so you see if it's working. Another option is to lower your cache size to 0.01 and 0.01 temporarily so you don't get a boatload of tasks that aren't doing what you want. Once you get it sorted out you can then change your cache size back to your normal settings.

BTW: As far as I can remeber, some changes to
<gpu_usage>
and/or to
<cpu_usage>
will only come into effect AFTER ALL of your "old" tasks are finished.

SFV

Okay, I added your line to the app_config.xml and then read the config file and got nothing. I know what you said about having to run out of previous units before the new units would take hold of the new changes, so I tried temporarily pausing all of the old ones and then re-reading the config file and then doing an update, without getting results from that tactic either. I guess I will just wait an see what happens for now.

Thanks!

One option is to abort all your existing tasks and then get new tasks, personally I would do that but also change the resource share to zero so you only get one task at a time so you see if it's working. Another option is to lower your cache size to 0.01 and 0.01 temporarily so you don't get a boatload of tasks that aren't doing what you want. Once you get it sorted out you can then change your cache size back to your normal settings.

Okay, I can hardy believe what I am going to tell you, but it just happened.

I went to Computing perferences, per your idea, and only changed the Store at least value to .1 days of work and the Store upto an additional 0 days of work and saved it and immediately my screen flickered and I was suddenly runninng 3 GPU units, instead of just one.

I give up trying to figure out why? If you would like to take a shot as to the answer though, I would be happy to listen.

Allen *;)

Just for the fun of it, I went back in and changed the .1 to 9.00 and immediately, when saved, it went back to running only 1 unit, but when I changed it back to .1 it started running the other 2 again. WOW!!!!