Help Wanted - Danger - Lunatic at work

It will make a difference; the platform is referenced in the cient_state.xml file for each task, which is why I'm interested in what 'blahblah' means....

I think the Fermi error is a red herring, and the actual problem IS the platform title for the CPU units....
 
Good to see you've come full circle and have done no real damage. This is going to take a little while to work through, and I'm about to hit the sack so I will have to continue tomorrow. Don't rush this and loose processing time, or worse, really FUBAR things up!

We need to back up just a little. When you installed the lunatics client, what options did you select for the GPU (if any)? Did the lunatics installer touch the client_state.xml file (check the date/time stamp, you'll find the file in the C:\Program Data\BOINC sub-directory)?

Can you post the app_info you currently use on this machine (prior to installing the sse version)?

Can you translate the 'blahblah'?

Was there any reason you decided to borrow an app_info from somewhere else, and not edit in place the existing one?

Nope i have double and triple backed up and just gone back to 1 task per gpu and all is ok.
I did not install any option for the gpu as its a fermi gpu and just added the fermi cuda sections myself.
There is no app_info used at present it is just the stock 6.10.56 no lunatics and use resched to catch vlars thats it.
There was avery good reason i used another app_info from elsewhere, through trial and error i got it wrong more than once and asked on the boards and the result was what i am using currently on 2 machines.
Hope thats clear for you Area 51, get yourself to sleep hopefully i may have resolved it by tommorrow - and pigs wil develope wings :D
 
Bloody hot to be honest. Summer is all well and good, but I wish it'd bugger off at night and let me sleep!

I thought I had a possible solution to your woes, but alas I did not. One thing intrigues me though:

The platform suggests 32-bit, yet the app is for 64-bit. Doubt this will solve your CUDA-related issues, but was just wondering if the above is correct.

Get yourself a fan and leave all the windows open - oops too late hear comes the rain lol
Yeah the 32 bit thing had me puzzled but it works on 2 64 bit machines so must not grumble :D
 
That app_info you posted is missing the cuda entries Toxic, you have the cuda_fermi entry but not the cuda :) Also if you want to use Rescheduler with this then you need to double up your CUDA entries so Rescheduler understands it.

I've taken the CPU part of yours and the CUDA part of mine and joined them together.

Here:
Code:
[COLOR="Cyan"]<app_info> 
    <app>
        <name>setiathome_enhanced</name>
    </app>
    <file_info>
        <name>AK_v8b_win_x64_SSSE3x.exe</name>
        <executable/>
    </file_info>
    <app_version>
        <app_name>setiathome_enhanced</app_name>
        <version_num>603</version_num>
        <platform>windows_intelx86</platform>
        <file_ref>
            <file_name>AK_v8b_win_x64_SSSE3x.exe</file_name>
             <main_program/>
        </file_ref>
    </app_version>
    <app_version>
        <app_name>setiathome_enhanced</app_name>
        <version_num>603</version_num>
        <platform>windows_x86_64</platform>
        <file_ref>
            <file_name>AK_v8b_win_x64_SSSE3x.exe</file_name>
             <main_program/>
        </file_ref>
    </app_version>[/COLOR]
    [COLOR="Red"]<app>
        <name>setiathome_enhanced</name>
    </app>
    <file_info>
	<name>libfftw3f-3-1-1a_upx.dll</name>
	<executable/>
    </file_info>
    <file_info>
	<name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</name>
	<executable/>
    </file_info>
    <file_info>
	<name>cudart32_30_14.dll</name>
	<executable/>
    </file_info>
    <file_info>
	<name>cufft32_30_14.dll</name>
	<executable/>
    </file_info>
    <app_version>
	<app_name>setiathome_enhanced</app_name>
	<version_num>610</version_num>
	<avg_ncpus>0.200000</avg_ncpus>
	<max_ncpus>0.200000</max_ncpus>
	<plan_class>cuda_fermi</plan_class>
	<file_ref>
	    <file_name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</file_name>
	    <main_program/>
	</file_ref>
	<file_ref>
	    <file_name>cudart32_30_14.dll</file_name>
	</file_ref>
	<file_ref>
	    <file_name>cufft32_30_14.dll</file_name>
	</file_ref>
	<file_ref>
	    <file_name>libfftw3f-3-1-1a_upx.dll</file_name>
	</file_ref>
	<coproc>
	    <type>CUDA</type>
	    <count>0.5</count>
	</coproc>
    </app_version>
    <app_version>
	<app_name>setiathome_enhanced</app_name>
	<version_num>610</version_num>
	<avg_ncpus>0.200000</avg_ncpus>
	<max_ncpus>0.200000</max_ncpus>
	<plan_class>cuda</plan_class>
	<file_ref>
	    <file_name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</file_name>
	    <main_program/>
	</file_ref>
	<file_ref>
	    <file_name>cudart32_30_14.dll</file_name>
	</file_ref>
	<file_ref>
	    <file_name>cufft32_30_14.dll</file_name>
	</file_ref>
	<file_ref>
	    <file_name>libfftw3f-3-1-1a_upx.dll</file_name>
	</file_ref>
	<coproc>
	    <type>CUDA</type>
	    <count>0.5</count>
	</coproc>
    </app_version>
    <app_version>
	<app_name>setiathome_enhanced</app_name>
	<version_num>608</version_num>
	<avg_ncpus>0.200000</avg_ncpus>
	<max_ncpus>0.200000</max_ncpus>
	<plan_class>cuda_fermi</plan_class>
	<file_ref>
	    <file_name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</file_name>
	    <main_program/>
	</file_ref>
	<file_ref>
	    <file_name>cudart32_30_14.dll</file_name>
	</file_ref>
	<file_ref>
	    <file_name>cufft32_30_14.dll</file_name>
	</file_ref>
	<file_ref>
	    <file_name>libfftw3f-3-1-1a_upx.dll</file_name>
	</file_ref>
	<coproc>
	    <type>CUDA</type>
	    <count>0.5</count>
	</coproc>
    </app_version>
    <app_version>
	<app_name>setiathome_enhanced</app_name>
	<version_num>608</version_num>
	<avg_ncpus>0.200000</avg_ncpus>
	<max_ncpus>0.200000</max_ncpus>
	<plan_class>cuda</plan_class>
	<file_ref>
	    <file_name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</file_name>
	    <main_program/>
	</file_ref>
	<file_ref>
	    <file_name>cudart32_30_14.dll</file_name>
	</file_ref>
	<file_ref>
	    <file_name>cufft32_30_14.dll</file_name>
	</file_ref>
	<file_ref>
	    <file_name>libfftw3f-3-1-1a_upx.dll</file_name>
	</file_ref>
	<coproc>
	    <type>CUDA</type>
	    <count>0.5</count>
	</coproc>
    </app_version>
</app_info>[/COLOR]

Get that working, and I have another question for you. Are you using i7 CPU's and if so why aren't you using the SSSE4.1 enchanced app rather than the SSE3 enhanced app?
 
Last edited:
Morning all,

Biffa, apparently there is very little difference between the SSE3 and the 4.1 in performance. Im sure I read somewhere that the best setting for i7 was SSSE3...

I will post my mangled app_info when I get to work...

Toxic - look at the seti forums cuda thread as that has the steps you need to go through...

The process was meant to be:

1) Stop BOINC
2) ReSchedule tasks
3) Examine client_state
4) Modify app_info
5) Start BOINC

You will lose work if you allow #5 without successfully completing #3 and #4.

I had to add this to the end of my app info

<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>610</version_num>
<avg_ncpus>0.200000</avg_ncpus>
<max_ncpus>0.200000</max_ncpus>
<plan_class>cuda</plan_class>
<file_ref>
<file_name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart32_30_14.dll</file_name>
</file_ref>
<file_ref>
<file_name>cufft32_30_14.dll</file_name>
</file_ref>
<file_ref>
<file_name>libfftw3f-3-1-1a_upx.dll</file_name>
</file_ref>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
</app_version>

</app_info>


If you are still stuck I can send you my app_info and you can compare...
The key bit too look at to get rescheduler to work is the <plan_class>cuda</plan_class> bit, that has to match the entry in your client_state.xml file after you have rescheduled units
 
Last edited:
I see Biffa's <platform> entry is correct, which is why your CPU units barfed. Every task has a platform entry in client_state, so when the client tried to run, it couldn't find a corresponding <platform> entry in the app_info. With Biffa's explanation of the CUDA entry, I think this is good to go now.......
 
Thanks for that Mr Biffa sah much appreciated, am going to sit and give it a try in a while and will let you know what happens....got to finish some work first....booo :)

Edge i will have a good read through the message boards again but as far as those steps go i am not sure what i am looking for in the client state xml but will back up before i start to be sure.

Thanks for your input too Area 51 i think without you showing me how to back up i could well have fubarred the lot....much appreciated sah! :)
 
I used the app_info you posted Biffa but still when i restart Boinc it discards all Cuda Fermi tasks stating there is no application for it, i should be happy it does a task in 10 mins but another machine does 2 in just over 15 so i have to go for it :D
I understand that entries have to be added to reschedule tasks from gpu to cpu as per Edges app_info. But i simply cannot understand why another machine is running a 480 with optimised apps and can edit app_info to run 1-3 tasks concurrently with the following app_info

<app_info>
<app>
<name>setiathome_enhanced</name>
</app>
<file_info>
<name>AK_v8b_win_x64_SSSE3x.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>603</version_num>
<platform>windows_intelx86</platform>
<file_ref>
<file_name>AK_v8b_win_x64_SSSE3x.exe</file_name>
<main_program/>
</file_ref>
</app_version>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>603</version_num>
<platform>windows_x86_64</platform>
<file_ref>
<file_name>AK_v8b_win_x64_SSSE3x.exe</file_name>
<main_program/>
</file_ref>
</app_version>
<app>
<name>setiathome_enhanced</name>
</app>
<file_info>
<name>libfftw3f-3-1-1a_upx.dll</name>
<executable/>
</file_info>
<file_info>
<name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</name>
<executable/>
</file_info>
<file_info>
<name>cudart32_30_14.dll</name>
<executable/>
</file_info>
<file_info>
<name>cufft32_30_14.dll</name>
<executable/>
</file_info>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>610</version_num>
<avg_ncpus>0.750000</avg_ncpus>
<max_ncpus>0.750000</max_ncpus>
<plan_class>cuda_fermi</plan_class>
<file_ref>
<file_name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart32_30_14.dll</file_name>
</file_ref>
<file_ref>
<file_name>cufft32_30_14.dll</file_name>
</file_ref>
<file_ref>
<file_name>libfftw3f-3-1-1a_upx.dll</file_name>
</file_ref>
<coproc>
<type>CUDA</type>
<count>0.33</count>
</coproc>
</app_version>
</app_info>

I guess its lots of coffee, a few cigarettes and an afternoon on the message boards. I do appreciate everyones help and i love sinking my teeth into problems and learning something new but realise by keeness will end up driving you all insane, so please feel free to tell me to sod orff at any time :D
 
I have mailed you my app_info, but here is a summary

To get rescheduler working I had to duplicate my Fermi section, but change the plan_class from cuda_fermi to cuda
If you look at the two last sections you will see they are identical except for the plan_class and platform bit

I have highlighted the two sections and changed them to capitals so you know the bits to look at

<PLAN_CLASS>CUDA_FERMI</PLAN_CLASS> - tells boinc to use the Fermi app

But reschedule brands the units as CUDA and Boinc doesn’t think it has an app for that

So <PLAN_CLASS>CUDA_FERMI</PLAN_CLASS> becomes <PLAN_CLASS>CUDA</PLAN_CLASS>

If you are running a 64 bit OS then you will need to change
<PLATFORM>WINDOWS_X86_64</PLATFORM>

to

<PLATFORM>WINDOWS_INTELX86</PLATFORM>

To get more than 1 task on the GPU change

<COUNT>1</COUNT>
to
<COUNT>0.33</COUNT>
Which will give you 3 tasks
Or use 0.5 to have 2 tasks


editing an app_info to run more tasks is different to editing an app_info to use rescheduler..
 
Thank you Edge very much h'appreciated. Have printed out the above stuck the coffee pot on and will get it going one way or the other today :D
 
:D:D:D
Finally, with the help of Edgedemon Biffa and Area 51 i have tried errored and finally succeeded.
It should be common sense i know but have truly learnt the value of "back it up" thanks to Area 51.
With the help of Biffa and Edges app_info i finally see why multiple cuda entries are neccesary especially when it comes to rescheduler.
Line upon line of text has allways bothered me after a while my eyes hurt like heck but now i know why and how it works i will even try Astropulse, even though there does not seem much astropulse work what kind of credit is involved?
Anyways all seems ok on that machine so will very carefully go around and apply changes to other rigs.
One machines down to 50 another 110 cuda tasks but if the extended outages do become a regular feature i think its a good idea. For one they will get a chance to sift through all the data they have allready harvested, and 2 if its down and as long as we have a decent cache i wont feel the need to update every 2 minutes to try and get more work etc, leaving me a bit of extra free time to find something else to sink my teeth into :D
Thank you chaps TBag towers is awash with cheesy grins this afternoon :D:D
 
Glad you're up and running. The one thing I would say about restoring from backups is that if you crunch for multiple projects, a little more finesse is required - you'd have to pack the blunderbuss away for those situations ;) !!!!! Still, all's well that ends well :D.
 
Don't think i could have handled multiple projects, i am allready a gibbering wreck from the last 24 hours, there is a great sense of satisfaction in learning something new...you just can't beat it.
 
Then two days later you will have forgotten :D

Post your current xml in
Code:
[code]
tags for a gander
 
Not sure code tags? If you mean the way you posted the 2 joined xml's am not sure how to do that....i'm a copy n paste person unless you want to tell me i will give it a go :)
 
Just put
PHP:
[code][/code]
tags around what you copy and paste (at the top and bottom), like you do when you put
PHP:
[img][/img]
tags around an image :)

That way it maintains the spacing and layout of the xml file
 
Code:
<app_info>
- <app>
  <name>setiathome_enhanced</name> 
  </app>
- <file_info>
  <name>AK_v8b_win_x64_SSSE3x.exe</name> 
  <executable /> 
  </file_info>
- <app_version>
  <app_name>setiathome_enhanced</app_name> 
  <version_num>603</version_num> 
  <platform>windows_intelx86</platform> 
- <file_ref>
  <file_name>AK_v8b_win_x64_SSSE3x.exe</file_name> 
  <main_program /> 
  </file_ref>
  </app_version>
- <app_version>
  <app_name>setiathome_enhanced</app_name> 
  <version_num>603</version_num> 
  <platform>windows_x86_64</platform> 
- <file_ref>
  <file_name>AK_v8b_win_x64_SSSE3x.exe</file_name> 
  <main_program /> 
  </file_ref>
  </app_version>
- <app>
  <name>astropulse_v505</name> 
  </app>
- <file_info>
  <name>ap_5.05r409_SSE.exe</name> 
  <executable /> 
  </file_info>
- <app_version>
  <app_name>astropulse_v505</app_name> 
  <version_num>505</version_num> 
  <platform>windows_intelx86</platform> 
- <file_ref>
  <file_name>ap_5.05r409_SSE.exe</file_name> 
  <main_program /> 
  </file_ref>
  </app_version>
- <app_version>
  <app_name>astropulse_v505</app_name> 
  <version_num>505</version_num> 
  <platform>windows_x86_64</platform> 
- <file_ref>
  <file_name>ap_5.05r409_SSE.exe</file_name> 
  <main_program /> 
  </file_ref>
  </app_version>
- <app>
  <name>setiathome_enhanced</name> 
  </app>
- <file_info>
  <name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</name> 
  <executable /> 
  </file_info>
- <file_info>
  <name>cudart32_30_14.dll</name> 
  <executable /> 
  </file_info>
- <file_info>
  <name>cufft32_30_14.dll</name> 
  <executable /> 
  </file_info>
- <file_info>
  <name>libfftw3f-3-1-1a_upx.dll</name> 
  <executable /> 
  </file_info>
- <app_version>
  <app_name>setiathome_enhanced</app_name> 
  <version_num>610</version_num> 
  <avg_ncpus>0.330000</avg_ncpus> 
  <max_ncpus>0.330000</max_ncpus> 
  <platform>windows_intelx86</platform> 
  <plan_class>cuda_fermi</plan_class> 
- <file_ref>
  <file_name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</file_name> 
  <main_program /> 
  </file_ref>
- <file_ref>
  <file_name>cudart32_30_14.dll</file_name> 
  </file_ref>
- <file_ref>
  <file_name>cufft32_30_14.dll</file_name> 
  </file_ref>
- <file_ref>
  <file_name>libfftw3f-3-1-1a_upx.dll</file_name> 
  </file_ref>
- <coproc>
  <type>CUDA</type> 
  <count>0.5</count> 
  </coproc>
  </app_version>
- <app_version>
  <app_name>setiathome_enhanced</app_name> 
  <version_num>610</version_num> 
  <avg_ncpus>0.330000</avg_ncpus> 
  <max_ncpus>0.330000</max_ncpus> 
  <platform>windows_x86_64</platform> 
  <plan_class>cuda_fermi</plan_class> 
- <file_ref>
  <file_name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</file_name> 
  <main_program /> 
  </file_ref>
- <file_ref>
  <file_name>cudart32_30_14.dll</file_name> 
  </file_ref>
- <file_ref>
  <file_name>cufft32_30_14.dll</file_name> 
  </file_ref>
- <file_ref>
  <file_name>libfftw3f-3-1-1a_upx.dll</file_name> 
  </file_ref>
- <coproc>
  <type>CUDA</type> 
  <count>0.5</count> 
  </coproc>
  </app_version>
- <app_version>
  <app_name>setiathome_enhanced</app_name> 
  <version_num>610</version_num> 
  <avg_ncpus>0.330000</avg_ncpus> 
  <max_ncpus>0.330000</max_ncpus> 
  <platform>windows_intelx86</platform> 
  <plan_class>cuda</plan_class> 
- <file_ref>
  <file_name>setiathome_6.10_windows_intelx86__cuda_fermi.exe</file_name> 
  <main_program /> 
  </file_ref>
- <file_ref>
  <file_name>cudart32_30_14.dll</file_name> 
  </file_ref>
- <file_ref>
  <file_name>cufft32_30_14.dll</file_name> 
  </file_ref>
- <file_ref>
  <file_name>libfftw3f-3-1-1a_upx.dll</file_name> 
  </file_ref>
- <coproc>
  <type>CUDA</type> 
  <count>0.5</count> 
  </coproc>
  </app_version>
  </app_info>
 
Heh, you don't have the xml spacing layout in yours anyway :D thats ok just makes it harder to follow :)

Uh.. wtf are all those dashes in there? I assume the are not in the real file?
 
Back
Top Bottom