App_Config

[OldChap]

I don't uninstall either.... Never had a problem so never bothered.

[jeanguy2]

Sound advise, as usual, from the fly boy. If I had the ability, I'd give you a Thank You to jeanguy2 For This Useful Post.

Dumb question: jeanguy2, could/would you please provide a link to the "new version release notes" area? That will provide a logical completion to this issue.

Edit: As to the if it ain't broke, I get an occasional calculation error. How do I know if it was cause by my current revision or not? The errors are not occurring often enough to justify modifying the app_config file.

Gandaff, Do not worry about the official thanking for me. By helping you, I am just paying back my dues for all the help I received from other XS members... (I owe a lot of thanks to the like of SAM, OC, Nano, Snowcrash and many other XS members...) Once your knowledge picks up you"ll just have to do the same for future XS WCG NFG...

Here is the link to the "BOINC 7 Change Log and news" . Just scroll down to view older alpha versions. When downloading a version, double check to insure that you are getting the correct one for your OS

The BOINC updating experiences of MUMAK and OC suggest that an uninstall of the old BOINC version may not be required prior to updating to a newer BOINC version. Here is a link to "Updating a BOINC Client Software Installation With The Windows Installer Program" . Just scroll down to "General Outline of the Upgrade Process" and read on. You can then select the update path that you wish to follow...

As stated by Mumak few errors will be caused by the actual BOINC application. Some errors may not require corrective action either if they are not too numerous and do not overly affect production efficiency. For example: the Xfire rigs I operate will usually produce 1 to 3 errors daily having to do with "No process down the pipe" these cost no time and have a GPU/CPU expanded time of 0.0/0.0 . If a BOINC error worries you use the "Results Status" option on your WCG page to ID it. You can then Google or post here to find out more about it.

Have a great day,

[Gandalf]

I have another question (I go a million of them! - anyone remember what entertainer used to say that? I do. ).

As I understand it, if I want to have some of my GPUs task on a different Project other than hcc1, all I need to do is change the <name>hcc1</name> to that Projects' initials. Everything else can stays the same. Right?

Is there a list of the Projects with their associated <name> ?
If so, what is its URL?

Thanks

[Evantaur]

Is there a list of the Projects with their associated <name> ?

Thanks

inside your boinc data folder
(C:\ProgramData\BOINC)

client_state.xml

find line that says <app>

[Gandalf]

inside your boinc data folder
(C:\ProgramData\BOINC)

client_state.xml

find line that says <app>

This is all I found there...

<app>
<name>cep2</name>
<user_friendly_name>The Clean Energy Project - Phase 2</user_friendly_name>
<non_cpu_intensive>0</non_cpu_intensive>
</app>
<app>
<name>hcc1</name>
<user_friendly_name>Help Conquer Cancer</user_friendly_name>
<non_cpu_intensive>0</non_cpu_intensive>
</app>
<app>
<name>gfam</name>
<user_friendly_name>GO Fight Against Malaria</user_friendly_name>
<non_cpu_intensive>0</non_cpu_intensive>
</app>
<app>
<name>dsfl</name>
<user_friendly_name>Drug Search for Leishmaniasis</user_friendly_name>
<non_cpu_intensive>0</non_cpu_intensive>
</app>
<app>
<name>hpf2</name>
<user_friendly_name>Human Proteome Folding - Phase 2</user_friendly_name>
<non_cpu_intensive>0</non_cpu_intensive>
</app>
<app>
<name>sn2s</name>
<user_friendly_name>Say No to Schistosoma</user_friendly_name>
<non_cpu_intensive>0</non_cpu_intensive>
</app>
<app>
<name>faah</name>
<user_friendly_name>FightAIDS@Home</user_friendly_name>
<non_cpu_intensive>0</non_cpu_intensive>
</app>

To my knowledge hcc1 is the only one that uses GPU tasking.
Anyone have a better understanding of what I'm looking for?
The above didn't even include any of the non-WCG Projects.

[Gandalf]

Jeanguy2 - I just received and swapped out a 7770ATI GPU for one of the NV GPUs in my i7-3930K.

The app_config.xml currently reads...

<app_config>
<app>
<name>hcc1</name>
<gpu_versions>
<max_concurrent>6</max_concurrent>
<gpu_usage>0.50</gpu_usage>
<cpu_usage>0.50</cpu_usage>
</gpu_versions>
</app>
</app_config>

The resulting tasking indicates...
0.5C + 0.5NV (d0)
0.5C + 0.5NV (d0)
0.5C + 0.5NV (d1)
0.5C + 0.5NV (d1)
1C + 1ATI GPU

How do I modify the app_config to run both ATI and NV GPUs in the same computer or do I have to remove all the NV GPUs in order to use the ATI GPU?

Thanks

[jeanguy2]

Gandalf,

I am going to have to do a Douglas Adams on this one....!! Sorry... I must refuse to answer your question on the ground that I have zero expertize with Nvidia and Nvidia / ATI crunching on the same rig...

Do not desperate.. OC, SnowCrash, Nano, Sam or another Nvidia XS WCG high wizard will eventually pick up your post and answer it...

[Gandalf]

Update -

0.5C + 0.5NV (d0)
0.5C + 0.5NV (d0)
0.5C + 0.5NV (d1)
0.5C + 0.5NV (d1)
0.5C + 0.5ATI
0.5C + 0.5ATI

I forgot to reboot. However, I would like to make better use of the ATI GPU. However, the NV can't, in my experience, handle anything more than 0.5NV. Is there a way to configure one set for the NV and another for ATI within the same app_config file? I'm will to experiment if one of the local Wizards has an idea.

I learned a long time ago, when one person asks a question, many people await the answer.

[Mumak]

I don't think it's a good idea to mix nV/AMD GPUs in one machine for BOINC, but maybe your experience will be different.
AFAIK, there's no way how to specify different GPU usages for different GPUs. There's a trick (I learned from OC) which can be used for asymmetric configs to set max_concurrent to total number of WUs for all GPUs and gpu_usage per GPU. So based on how you set these values, the 1st GPU in system (device 0) will be used and the rest of GPU(s) will use the remaining units (of max_concurrent). But since your config is very specific, I think that won't be possible. Maybe if you would swap the ATI to be the primary card...

[OldChap]

I'm not sure I understand how many cards you are running in total but if 2 then:

change the app config <gpu usage>to run the right number for the amd card.

change the <cpu usage> for that number times 2 cards

Now change <maximum concurrent> for the total gpu wu's you want to run.

Example:

<app_config>
<app>
<name>hcc1</name>
<gpu_versions>
<max_concurrent>10</max_concurrent>
<gpu_usage>0.125</gpu_usage>
<cpu_usage>0.25</cpu_usage>
</gpu_versions>
</app>
</app_config>

want to run 10, 8 on amd and 2 on nv.
gpu usage =8 (for the amd)
cpu usage = 16 (but can be set for 10 too if you wish) assumed that this is a 4 core only cpu.
max concurrent set for 10

Provided that your rig sees the amd card as the primary card this should fill the amd card first the give another 2 to the nv card.

Off to work now any other q +a later

[Evantaur]

Btw managed to fix problems i mentioned earlier

Had to install libraries from debian experimental repo and update boinc

[Gandalf]

Sorry OldChap, but I got so frustrated, I pulled all the NVIDIA cards. I've currently have just the one 7770 card in it now, but I've got more 7770s coming. No matter in what order I arranged the cards, the NVs were always referred to as device 0 and device 1. No matter how I modified the app_config file, both types of GPU worked on the same setting. So, I yanked them. Sometime even a wizard has to know when to attack in a different direction.

I set the ATI to 0.25C + 0.0625ATI. The 6 pack now is running 16 GPU tasks and 8 CPU tasks. What is interesting is the quad could not handle the 0.0625ATI setting, but, so far, the six pack is happy as a pig in mud. I think I'll hold off any addition app_config modifications for now to see how the 6 pack settles in with the ATI GPU.

Thanks for the suggestion on the configuration, but I just couldn't make it work.

[Evantaur]

You could always run 2 boincs, one for nv and one for ati

[Gandalf]

Evantaur - I am converting over from NV to ATI. I have learned that since the world is going to OpenCL, the ATI is the GPU to use.

Since the WCG does not seem to want or is not able to continue providing GPU tasks, I am learning that WCG is not the only fish in the sea. I'm planning to continue assigning my CPU tasks to WCG, but I am currently looking for a GPU oriented Project for my GPUs. That is why I placed those posts (reference #78 and #80). A friend referred me to BOINC@Austrialia, but I haven't even been accepted into their forum much less onto their team. Well, I guess I have 23 days to find a worthy team to join. I'm to old to be playing politics. I'm having enough trouble just typing this post.

[Snow Crash]

join XS and be done ... we participate in a number of different projects not just WCG. Before you dump NV take a look around at what the different projects suupport ...a person favorite of mine is GPUGrid ... thye are coded for CUDA only as OpenCL is 10 times slower and does not have the primitives and error handling necessary to the science they are doing (molecular biology).

[Gandalf]

BOINC reports it can't find this in the app_config.xml

<app_config>
<app>
<name>cep2</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>c4cw</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>cfsw</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>dddt2</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>dsfl</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>faah</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>gfam</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>hcc1</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>hcmd2</name>
<max_concurrent>0</max_concurrent>
</app>
<app>
<name>hpf2</name>
<max_concurrent>0</max_concurrent>
</app>
<app>
<name>rice</name>
<max_concurrent>0</max_concurrent>
</app>
<app>
<name>sn2s</name>
<max_concurrent>0</max_concurrent>
</app>
</app_config>


What did I do wrong?

Thanks for the help!
Gandalf

[Evantaur]

BOINC reports it can't find this in the app_config.xml

<app_config>
<app>
<name>cep2</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>c4cw</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>cfsw</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>dddt2</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>dsfl</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>faah</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>gfam</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>hcc1</name>
<max_concurrent>1</max_concurrent>
</app>
<app>
<name>hcmd2</name>
<max_concurrent>0</max_concurrent>
</app>
<app>
<name>hpf2</name>
<max_concurrent>0</max_concurrent>
</app>
<app>
<name>rice</name>
<max_concurrent>0</max_concurrent>
</app>
<app>
<name>sn2s</name>
<max_concurrent>0</max_concurrent>
</app>
</app_config>


What did I do wrong?

Thanks for the help!
Gandalf

What version are you running?

[Gandalf]

What version are you running?

7.0.62 (x86)

[Evantaur]

You have it in your projects/www.world... Folder?

[Gandalf]

You have it in your projects/www.world... Folder?

Yes.

[Evantaur]

Should detect it just fine then :S

[Gandalf]

An interesting development. I upgraded to the latest and greatest version of BOINC and suddenly my app_config.xml configuration errors went from 12 to only 6.
I find that very interesting.

[Snow Crash]

What errors get displayed in the log?

[Gandalf]

What errors get displayed in the log?

4/22/2013 2:26:14 AM | World Community Grid | app c4cw not found in app_config.xml
4/22/2013 2:26:14 AM | World Community Grid | app hcmd2 not found in app_config.xml
4/22/2013 2:26:14 AM | World Community Grid | app cfsw not found in app_config.xml
4/22/2013 2:26:14 AM | World Community Grid | app dddt2 not found in app_config.xml
4/22/2013 2:26:14 AM | World Community Grid | app rice not found in app_config.xml

The other projects are running.

[Snow Crash]

Sounds like you have those projects selected in your WCG profile online and BOINC is alerting you to that fact ... nothing more.