Write your own Einstein@home screensaver

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 700103081
RAC: 272224

OK, prototype seems to work

OK, prototype seems to work under Linux:

Will try to complile under Mac OS and for Windows next.

CU
Bikeman

Oliver Behnke
Oliver Behnke
Moderator
Administrator
Joined: 4 Sep 07
Posts: 982
Credit: 25170813
RAC: 0

Hi guys, This is great!

Hi guys,

This is great! Nice to see that the framework seems to work for you guys so far. Please let me know your opinions and ideas for improvements you'd like to see.

@Bikeman: I suggest you turn on the high quality mode to get proper anti-aliasing and the fog effect for depth perception. You can do so either in your account preferences (when using live data) or by admending your local init_data.xml (standalone use) like this:


Cheers,
Oliver

Einstein@Home Project

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 700103081
RAC: 272224

RE: Hi guys, This is

Message 78000 in response to message 77999

Quote:

Hi guys,

This is great! Nice to see that the framework seems to work for you guys so far. Please let me know your opinions and ideas for improvements you'd like to see.

@Bikeman: I suggest you turn on the high quality mode to get proper anti-aliasing and the fog effect for depth perception. You can do so either in your account preferences (when using live data) or by admending your local init_data.xml (standalone use) like this:


Cheers,
Oliver

Hmm... interesting. I'm testing on a remote desktop (vnc) and it would not work on "high" (graphics program just exits). I guess this is caused by a lack of feature support by the remote desktop server, but anyway, I wonder whether it might be useful to have an automativ fallback to lower settings if the gfx engine fails to init with high settings.

CU
Bikeman

Oliver Behnke
Oliver Behnke
Moderator
Administrator
Joined: 4 Sep 07
Posts: 982
Credit: 25170813
RAC: 0

RE: Hmm... interesting.

Message 78001 in response to message 78000

Quote:

Hmm... interesting. I'm testing on a remote desktop (vnc) and it would not work on "high" (graphics program just exits). I guess this is caused by a lack of feature support by the remote desktop server, but anyway, I wonder whether it might be useful to have an automativ fallback to lower settings if the gfx engine fails to init with high settings.

That's interesting indeed and most likely caused by limited hardware support. Can you please have a look for warnings and error messages on STDERR (or its file redirects respectively)? Any hint? Does it work with "medium"?

In fact, there already is minimum fallback support. The problem with an automatic fallback is that it's not always clear which feature request (size, depth, buffers, FSAA, etc.) caused the video mode request to fail. One would have to test all combinations separately, because SDL only allows you to set a video mode and test whether it worked afterwards - there's no such thing like an upfront capabilities check.

Thanks,
Oliver

Einstein@Home Project

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 700103081
RAC: 272224

RE: RE: Hmm...

Message 78002 in response to message 78001

Quote:
Quote:

Hmm... interesting. I'm testing on a remote desktop (vnc) and it would not work on "high" (graphics program just exits). I guess this is caused by a lack of feature support by the remote desktop server, but anyway, I wonder whether it might be useful to have an automativ fallback to lower settings if the gfx engine fails to init with high settings.

That's interesting indeed and most likely caused by limited hardware support. Can you please have a look for warnings and error messages on STDERR (or its file redirects respectively)? Any hint? Does it work with "medium"?

In fact, there already is minimum fallback support. The problem with an automatic fallback is that it's not always clear which feature request (size, depth, buffers, FSAA, etc.) caused the video mode request to fail. One would have to test all combinations separately, because SDL only allows you to set a video mode and test whether it worked afterwards - there's no such thing like an upfront capabilities check.

Thanks,
Oliver

Medium works, will check output this evening.

CU
Bikeman

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 700103081
RAC: 272224

RE: RE: RE: Hmm...

Message 78003 in response to message 78002

Quote:
Quote:
Quote:

Hmm... interesting. I'm testing on a remote desktop (vnc) and it would not work on "high" (graphics program just exits). I guess this is caused by a lack of feature support by the remote desktop server, but anyway, I wonder whether it might be useful to have an automativ fallback to lower settings if the gfx engine fails to init with high settings.

That's interesting indeed and most likely caused by limited hardware support. Can you please have a look for warnings and error messages on STDERR (or its file redirects respectively)? Any hint? Does it work with "medium"?

In fact, there already is minimum fallback support. The problem with an automatic fallback is that it's not always clear which feature request (size, depth, buffers, FSAA, etc.) caused the video mode request to fail. One would have to test all combinations separately, because SDL only allows you to set a video mode and test whether it worked afterwards - there's no such thing like an upfront capabilities check.

Thanks,
Oliver

Medium works, will check output this evening.

CU
Bikeman

I get this in stderr :

Quote:

Could not acquire rendering surface: Couldn't find matching GLX visual
Window manager could not be initialized!
Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 700103081
RAC: 272224

Now, I managed to compile the

Now, I managed to compile the E@H screensaver under OS X "Leopard" with Xcode 3.1 , but only after setting the following environment variables:

Quote:


export LDFLAGS="-isysroot /Developer/SDKs/MacOSX10.4u.sdk -Wl,-syslibroot,/Developer/SDKs/MacOSX10.4u.sdk -arch i386 $LDFLAGS"
export CPPFLAGS="-isysroot /Developer/SDKs/MacOSX10.4u.sdk -arch i386 $CPPFLAGS"
export CFLAGS="-isysroot /Developer/SDKs/MacOSX10.4u.sdk -arch i386 $CFLAGS"
export CXXFLAGS="-isysroot /Developer/SDKs/MacOSX10.4u.sdk -arch i386 $CXXFLAGS"
export SDKROOT="/Developer/SDKs/MacOSX10.4u.sdk"
export MACOSX_DEPLOYMENT_TARGET=10.4
export MKDIRPROG="mkdir -p"

All but the last line were borrowed from the science app build scripts, the last line avoided a strange build error I got in the install phase of the BOINC part of the build process.

Anyway, works like a charm on my Core Duo Mac mini, except for "high" graphics mode again. This time the "screensaver" start up ok, but doesn't display most of the layers (only constellation and observatories), and is unresponsive to keyboard input. Go figure.. I have to do more debugging here, but at least up to "med" level eveything works pretty well on my mac mini now.

CU
Bikeman

Oliver Behnke
Oliver Behnke
Moderator
Administrator
Joined: 4 Sep 07
Posts: 982
Credit: 25170813
RAC: 0

Hi everyone, Please note

Hi everyone,

Please note that the MinGW cross-compile build is currently broken (again). I already found the culprit (upstream) and will push yet another workaround into the repo in the next couple of hours...

Update: workaround deployed!
2nd Update: fixed upstream!

Sorry for any inconvenience,

Oliver

Einstein@Home Project

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 700103081
RAC: 272224

Hi all! If you want to try

Hi all!

If you want to try out the visual effect of my screensaver modification (the one from the screenshots posted earlier), here's how to do it:

The code needs some polishing here and there and some more testing under different platforms, but for those who want to have a look, here's a patch that will add some functionality to the standard starsphere code:

* inside the starsphere, there will be a graphical representation of the results (up to 10.000 of the best scoring search templates), indicating the sky position of the candidate and it's score wrt to one of the metrics collected in the output. Points closer to the surface of the sphere (more "outward", color coded red) are better than those close to the center of the sphere.

* This display can be toggled on and off using the 'u' key.

To build the modified starsphere executable, first download the original code using

git clone git://git.aei.uni-hannover.de/shared/einsteinathome/graphicsframework.git

then apply the patch using

git apply starsphere_20090318.patch


where starsphere_20090318.patch is the saved copy of the file downloaded above.

You can ignore some warnings about whitespaces while performing the patch.

Feedback is welcome,

CU
Bikeman

Oliver Behnke
Oliver Behnke
Moderator
Administrator
Joined: 4 Sep 07
Posts: 982
Credit: 25170813
RAC: 0

FYI, for all of you who make

FYI, for all of you who make modifications/contributions to the official code base I suggest you publish your changes using public git repositories instead of supplying patches. It's much easier to track, try and integrate changes this way. That's what's so nice about distributed SCM in general and git in particular.

If you can't host those repositories yourself, I recommend GitHub as a free (for open source software) git hosting service.

Cheers,
Oliver

Einstein@Home Project

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.