We are releasing a new version of separation! This version of separation will be run under a different application name because it runs a modified fit and we would like to be able to work using both the previous and new fits for the time being. If there are any issues with the work units from milkyway_separation__modified_fit please post them here.

The following platforms are supported with this update:
32-bit Linux: OpenCL devices with double precision and CPU
64-bit Linux: OpenCL devices with double precision and CPU
64-bit Windows: OpenCL devices with double precision and CPU
MacOSX: OpenCL devices with double precision and CPU

We hope to add 32-bit Windows support in the near future.

Thank you,

Jake W

Then I suggest you give it a unique 'friendly_name' so we can tell the difference in the project specific prefs and the other BOINC generated info pages.

Preferably before you actually roll it out.

what exactly is the "near future"?
Since the depreciation fiasco there has been no work for my 32bit cpu machine.

...
Since the depreciation fiasco there has been no work for my 32bit cpu machine.

I did post a workaround for that in NC.

In the project preferences, there are two separation applications! How do we know which is which?

You could just do what malariacontrol.net does. They have applications alongside "beta" applications (with same name, just plus "beta"). Users can run either or both alongside the regular app.

Are there any WU in the field?

Can´t see any in the server stats.

---

There aren't currently any work units up yet. I will be pushing out the newest version today as well as some work units. The new separation should be labeled Milkyway Separation Modified Fit. As far as when we will have a 32bit Windows application, I am not sure. It is the next thing on my list to do after getting this released and stable.

Jake W.

There aren't currently any work units up yet. I will be pushing out the newest version today as well as some work units. The new separation should be labeled Milkyway Separation Modified Fit. As far as when we will have a 32bit Windows application, I am not sure. It is the next thing on my list to do after getting this released and stable.

Jake W.

Hmmmm...

When I look at the projects application page, the Modified Fit part doesn't show up in section title line. Likewise, it's missing from the text on the app selection line on the project specific preferences.

Not a big deal, but I think we've had enough confusion lately. I imagine you'd like to minimize that going forward. ;-)

I found some time to look into the problems with Win32 and it looks like I found a temporary solution. I have the binaries built now, just want to do some testing with it and make sure it actually works. Hopefully will have it released in a timely manner.

Jake W.

I don't know if this belongs here but every work unit I get is calcualted until 100% and then results in 'calculation error'. All of those are either de_separation or ps_separation.

Application is Milkyway@home 1.02 (opencl_amd_ati).

I tried it ~1 week ago with one batch of workunits and I tried it again today.

This is one of the results: http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=504866748

I guess this is the problem

Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4'
Error reading astronomy parameters from file 'astronomy_parameters.txt'
Trying old parameters file

but I can't find such a file on my PC (I tried sudo updatedb && locate astronomy_parameters.txt with no result)

I don't know if this belongs here but every work unit I get is calcualted until 100% and then results in 'calculation error'. All of those are either de_separation or ps_separation.

Application is Milkyway@home 1.02 (opencl_amd_ati).

I tried it ~1 week ago with one batch of workunits and I tried it again today.

This is one of the results: http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=504866748

I guess this is the problem

Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4'
Error reading astronomy parameters from file 'astronomy_parameters.txt'
Trying old parameters file

but I can't find such a file on my PC (I tried sudo updatedb && locate astronomy_parameters.txt with no result)

This isn't the application being discussed in this thread, and it would be better to post this in Number Crunching.

Anyway, the 'error' message in the stderr isn't the problem in this case, the problem is the project started sending you work for the openCL version rather than the CAL based one.

My guess would be most likely a driver version issue.

OK, what's the story here?

There doesn't seem to be any of the MWSMF in the field. The server status page shows 500 ready and zero sent and has been that way for over 24 hours.

I am currently talking to Travis about it to see what is going on. I don't know much about the inner workings of the server so I am hoping it will be fixed soon.

Jake W.

OK, what's the story here?

There doesn't seem to be any of the MWSMF in the field. The server status page shows 500 ready and zero sent and has been that way for over 24 hours.

They should be getting sent out now.

--Travis

---

All is working now!

Thanks all,

Jake W

Confirmed.

I have a couple now on one of my hosts, so now the next adventure begins! ;-)

<edit> I take it giving it a unique name did the trick?

Has the app_name been changed for the modified fit version or are mods required to the app_config? I have an app_config file and specify GPU usage at .5 for 2x and the modified fit only runs 1x. The app_config works running 2x with the earlier versions.

Edit: I found the name in client_state and added the following to the app_config file. All works as expected now.

<app>
<name>milkyway_separation__modified_fit</name>
<gpu_versions>
<gpu_usage>0.50</gpu_usage>
<cpu_usage>0.20</cpu_usage>
</gpu_versions>
</app>

Yes, and if you're running an app_info (to keep nBody under better control for example) you'll need to edit the <name> element for MWSMF as well.

Confirmed.

I have a couple now on one of my hosts, so now the next adventure begins! ;-)

I take it giving it a unique name did the trick?

Actually we just needed to restart the boinc daemons. :P

---

All of the Modified Fit Separation tasks that I download run for about 2 seconds and then show 100% Progress along with a "Computation Error" Status. I have reset the project a couple of times with the same results. Unmodified separations appear to be running normally.

I have a Mac laptop with a 2.3 GHz Intel Core i7 processor running OS X 10.8.4. I am using Boinc Manager ver. 7.0.65.

Emmett