einsteinbinary_BRP4_1.01_i686-apple-darwin__BRP3cuda32OSX was downloaded via the boinc client and works out of the box with my NVIDIA GTS450, OS X10.6.8 / 10.7 and latest Nvidia CUDA Driver and toolkit.
so it have to with an app_info.xml too!
behind i want to try the OpenCL file.
in my app_info.xml i fixed the line graphics_app , delete the collatz comantline and checked 1024 mb vram as 1073741824.00 from client_state.xml
do i have to bind any other files like stochastic_full.bank or the slideshowfiles?
here is my message output for BRP4 CUDA:
Einstein@Home | Sending scheduler request: To fetch work.
Einstein@Home | Requesting new tasks for NVIDIA
Einstein@Home | Scheduler request completed: got 1 new tasks
Einstein@Home | Resent lost task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0
Einstein@Home | Started download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152.binary
Einstein@Home | Started download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_153.binary
Einstein@Home | Finished download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_153.binary
Einstein@Home | Started download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_154.binary
Einstein@Home | Finished download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152.binary
Einstein@Home | Started download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_155.binary
Einstein@Home | Finished download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_154.binary
Einstein@Home | Started download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_156.binary
Einstein@Home | Finished download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_155.binary
Einstein@Home | Started download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_157.binary
Einstein@Home | Finished download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_156.binary
Einstein@Home | Started download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_158.binary
Einstein@Home | Finished download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_157.binary
Einstein@Home | Started download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_159.binary
Einstein@Home | Finished download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_158.binary
Einstein@Home | Started download of stochastic_full.bank
Einstein@Home | Finished download of stochastic_full.bank
Einstein@Home | Started download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000.zap
Einstein@Home | Finished download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000.zap
Einstein@Home | Finished download of p2030.20111005.G35.37+01.20.C.b6s0g0.00000_159.binary
Einstein@Home | Starting task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 using einsteinbinary_BRP4 version 101 (BRP3cuda32OSX) in slot 1
Einstein@Home | Computation for task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 finished
Einstein@Home | Output file p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0_0 for task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 absent
Einstein@Home | Output file p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0_1 for task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 absent
Einstein@Home | Output file p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0_2 for task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 absent
Einstein@Home | Output file p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0_3 for task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 absent
Einstein@Home | Output file p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0_4 for task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 absent
Einstein@Home | Output file p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0_5 for task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 absent
Einstein@Home | Output file p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0_6 for task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 absent
Einstein@Home | Output file p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0_7 for task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 absent
message output for BRP4 OpenCL:
Einstein@Home | Restarting task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 using einsteinbinary_BRP4 version 110 (NVOpenCL) in slot 1
Einstein@Home | Task p2030.20111005.G35.37+01.20.C.b6s0g0.00000_152_0 exited with zero status but no 'finished' file
Einstein@Home | If this happens repeatedly you may need to reset the project.
could i use einsteinbinary_BRP4_1.01_i686-apple-darwin__BRP3cuda32OSX5 too ?
what i have to change if i want two WU´s for one GPU?
would be great if you could give me some more help
Do you really think it's a good idea to take test versions of apps from the test project site and transport them across to the main site and try to force them to run under AP? Will you take responsibility for any damage you cause to results stored in the main project database through this very ill-advised action? You should only run the official apps as would be supplied automatically to your client when it contacts the server. If you want to run these apps under AP, it is your responsibility to set up the appropriate app_info.xml file. Running under AP is not encouraged by the project Admins and if you choose to go down this path, you need to make sure you know what you are doing. There is no 'one size fits all' version of app_info.xml. It needs to be customised for your particular requirements. It's easy to trash lots of work and to waste more resources than you are likely to save just to get things running in the first place. You need to pay close attention in case the app version is changed by the project and you need a proper plan for how to back out of AP when circumstances change. It's easy to waste resources at both these points.
If you want to run any app other than an official one, it would be common courtesy to contact one of the project Admins (eg Bernd) and let him know what you propose to do and why you think you should be allowed to do it. I would think it's extremely unlikely that you would be granted permission, simply because of the risk it would pose to the integrity of the project's results.
The OpenCL app is still under test at the Albert@Home site and should only be run there at present. There is certain to be a full announcement when it is ready for release here. Until that time you should wait patiently like everybody else.
hello, thx for reply.
first, i want to create an app_info.xml for the official projekt to edit the cmdline to get more GPU load. (right now only 80 %)
i did the same with the collatz opencl app and it is now 2,5 times faster
behind as i wrote before i would like to test the opencl version!
bernd ist talking here about BRP4_1.07 (July 2011) and this is not official too!
so what do you think as newbie when you could download these files?
why could anybody download these files and could damage the project?
i don´t think that everybody takes time to write here in and ask for help to support the project.
but still, you are right :)
nevermind, here is my working app_info.xml for the offical einsteinbinary_BRP4_1.01_i686-apple-darwin__BRP3cuda32OSX
I may be missing the point or something,
And i do not know if this will work here at Einstein,
But if all you want is more GPU load why not run two tasks at the same time using the standard application by changing the value.
CUDA
1.000000
To . . .
CUDA
0.500
I dont remember exactly how it is set out,
It is done that way over on SETI for running two, three, fore or more work units at the same time, if your GPU has the ram needed to cope with the load.
And SETI is `down for maintenance` at the moment so i am unable to link to the thread that spell`s out the way to do it.
hello, thx for reply. i
)
hello, thx for reply.
i download einsteinbinary_BRP4_1.10_i686-apple-darwin__NVOpenCL from:
http://albert.phys.uwm.edu/EinsteinAtHome/download/
as you could download all other files.
einsteinbinary_BRP4_1.01_i686-apple-darwin__BRP3cuda32OSX was downloaded via the boinc client and works out of the box with my NVIDIA GTS450, OS X10.6.8 / 10.7 and latest Nvidia CUDA Driver and toolkit.
so it have to with an app_info.xml too!
behind i want to try the OpenCL file.
in my app_info.xml i fixed the line graphics_app , delete the collatz comantline and checked 1024 mb vram as 1073741824.00 from client_state.xml
do i have to bind any other files like stochastic_full.bank or the slideshowfiles?
here is my message output for BRP4 CUDA:
message output for BRP4 OpenCL:
could i use einsteinbinary_BRP4_1.01_i686-apple-darwin__BRP3cuda32OSX5 too ?
what i have to change if i want two WU´s for one GPU?
would be great if you could give me some more help
thx in advance
RE: i download
)
Do you really think it's a good idea to take test versions of apps from the test project site and transport them across to the main site and try to force them to run under AP? Will you take responsibility for any damage you cause to results stored in the main project database through this very ill-advised action? You should only run the official apps as would be supplied automatically to your client when it contacts the server. If you want to run these apps under AP, it is your responsibility to set up the appropriate app_info.xml file. Running under AP is not encouraged by the project Admins and if you choose to go down this path, you need to make sure you know what you are doing. There is no 'one size fits all' version of app_info.xml. It needs to be customised for your particular requirements. It's easy to trash lots of work and to waste more resources than you are likely to save just to get things running in the first place. You need to pay close attention in case the app version is changed by the project and you need a proper plan for how to back out of AP when circumstances change. It's easy to waste resources at both these points.
If you want to run any app other than an official one, it would be common courtesy to contact one of the project Admins (eg Bernd) and let him know what you propose to do and why you think you should be allowed to do it. I would think it's extremely unlikely that you would be granted permission, simply because of the risk it would pose to the integrity of the project's results.
The OpenCL app is still under test at the Albert@Home site and should only be run there at present. There is certain to be a full announcement when it is ready for release here. Until that time you should wait patiently like everybody else.
Cheers,
Gary.
hello, thx for reply. first,
)
hello, thx for reply.
first, i want to create an app_info.xml for the official projekt to edit the cmdline to get more GPU load. (right now only 80 %)
i did the same with the collatz opencl app and it is now 2,5 times faster
behind as i wrote before i would like to test the opencl version!
bernd ist talking here about BRP4_1.07 (July 2011) and this is not official too!
so what do you think as newbie when you could download these files?
why could anybody download these files and could damage the project?
i don´t think that everybody takes time to write here in and ask for help to support the project.
but still, you are right :)
nevermind, here is my working app_info.xml for the offical einsteinbinary_BRP4_1.01_i686-apple-darwin__BRP3cuda32OSX
einsteinbinary_BRP4 Binary Radio Pulsar Search (Arecibo) 0 einsteinbinary_BRP4_1.01_i686-apple-darwin__BRP3cuda32OSX 1 einsteinbinary_BRP3_1.00_graphics_i686-apple-darwin 1 libcudart32_32.dylib 1 libcufft32_32.dylib 1 libtlshook32_32.dylib 1 dbhs.dev.mac.3d35195e 1 db.dev.mac.3d35195e 1 einstein_icon.png 1 eah_slide_11.png 1 eah_slide_02.png 1 eah_slide_12.png 1 eah_slide_13.png 1 eah_slide_05.png 1 eah_slide_07.png 1 eah_slide_08.png 1 einsteinbinary_BRP4 101 i686-apple-darwin 0.200000 1.000000 BRP3cuda32OSX 6.13.0 einsteinbinary_BRP4_1.01_i686-apple-darwin__BRP3cuda32OSX einsteinbinary_BRP3_1.00_graphics_i686-apple-darwin graphics_app libcudart32_32.dylib libcudart.dylib libcufft32_32.dylib libcufft.dylib libtlshook32_32.dylib libtlshook.dylib dbhs.dev.mac.3d35195e dbhs.dev db.dev.mac.3d35195e db.dev CUDA 1.000000would be great if you could tell me the cmdline for more GPU load
before i spent more time, do you need testers for the new einsteinbinary_BRP4_1.22_i686-apple-darwin__BRP3cuda32OSX or OpenCL files?
I may be missing the point or
)
I may be missing the point or something,
And i do not know if this will work here at Einstein,
But if all you want is more GPU load why not run two tasks at the same time using the standard application by changing the value.
CUDA
1.000000
To . . .
CUDA
0.500
I dont remember exactly how it is set out,
It is done that way over on SETI for running two, three, fore or more work units at the same time, if your GPU has the ram needed to cope with the load.
And SETI is `down for maintenance` at the moment so i am unable to link to the thread that spell`s out the way to do it.