| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 14
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
7259 Charlie 5 Dec 2012 |
||
|
|
9maMSSuNWXgttyKdZhMemeXmEx8
Senior Cruncher Puerto Rico Joined: Feb 20, 2008 Post Count: 191 Status: Offline Project Badges:
|
Downloaded the latest BOINC client (7.0.40). It detects my Intel HD4000 as an OpenCL ATI GPU
----------------------------------------12/8/2012 9:45:33 AM | | NVIDIA GPU 0: GeForce GT 650M (driver version 310.64, CUDA version 5.0, compute capability 3.0, 2048MB, 1982MB available, 641 GFLOPS peak) 12/8/2012 9:45:33 AM | | OpenCL: NVIDIA GPU 0: GeForce GT 650M (driver version 310.64, device version OpenCL 1.1 CUDA, 2048MB, 1982MB available) 12/8/2012 9:45:33 AM | | OpenCL: ATI GPU 0: Intel(R) HD Graphics 4000 (driver version 9.17.10.2867, device version OpenCL 1.1, 1624MB, 1624MB available) 12/8/2012 9:45:33 AM | | Version change (7.0.38 -> 7.0.40) EDIT: Here's the scheduler trying to fetch Intel GPU tasks: 12/8/2012 9:51:41 AM | World Community Grid | Requesting new tasks for CPU and intel_gpu ![]() ![]() [Edit 1 times, last edit by moisesmcardona at Dec 8, 2012 1:52:59 PM] |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges:
|
Here's the scheduler trying to fetch Intel GPU tasks: Did you get any?
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
![]() ![]() |
||
|
|
9maMSSuNWXgttyKdZhMemeXmEx8
Senior Cruncher Puerto Rico Joined: Feb 20, 2008 Post Count: 191 Status: Offline Project Badges:
|
nope. I wrote on the BOINC forums to seek for an Intel GPU project. I'm awating an answer.
----------------------------------------![]() ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Let's emphasize, this is very much a test client. An Intel HD4000 identified as a ATI_GPU is already a set up for failing tasks, to which I have little doubt. Reporting to the developers is surely appreciated.
----------------------------------------edit: When mentioning "Reporting to the developers", I mean their alpha mail list. They hardly read the BOINC forums. [Edit 1 times, last edit by Former Member at Dec 8, 2012 4:39:47 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
To add to previous post edit, those interested to feed back to developers [those running the latest greatest, or not so great alpha clients], please consider this post on actions for alpha testers: http://boinc.berkeley.edu/dev/forum_thread.php?id=6258#36237
----------------------------------------And on the note of testing, and the new feature started with 7.0.40 to be able to control the number of tasks of a science [see change log notes on app_config.xml http://boinc.berkeley.edu/dev/forum_thread.php?id=6698&nowrap=true#46649] with this it would now be possible to set how many concurrent tasks on a GPU or a CPU of a specific science app. E.g. if you load up with 4 CEP2 in your buffer, but specify, for not more than 2 to run concurrent, this client would only start a maximum of 2, even if the other 2 would have their turn in the FIFO order of processing. This e.g. makes sure that you can maximize on a particular science without having to micro-manage to prevent crippling during use. The app_config.xml file also has to be in the projects/* folder. I'll be hoping to test this in the coming week. edit: FTM not getting app_config.xml to work. Start 1 CEP2, with a limit to only run 1, took copy of the application mask from Task Manager, pasted into the proper line. Restarted BOINC and forced a second to start. 2 CEP2 are running. Put a copy in both the data_dir and the project dir to make sure [the wiki does not mention where to place the file], restart BOINC... 2 running. Then per the wiki put the "short name" i.e. cep2 [per the client_state.xml info], still 2 running on restart... going to sleep on this. [Edit 1 times, last edit by Former Member at Dec 8, 2012 8:44:27 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Downloaded the latest BOINC client (7.0.40). It detects my Intel HD4000 as an OpenCL ATI GPU 12/8/2012 9:45:33 AM | | NVIDIA GPU 0: GeForce GT 650M (driver version 310.64, CUDA version 5.0, compute capability 3.0, 2048MB, 1982MB available, 641 GFLOPS peak) 12/8/2012 9:45:33 AM | | OpenCL: NVIDIA GPU 0: GeForce GT 650M (driver version 310.64, device version OpenCL 1.1 CUDA, 2048MB, 1982MB available) 12/8/2012 9:45:33 AM | | OpenCL: ATI GPU 0: Intel(R) HD Graphics 4000 (driver version 9.17.10.2867, device version OpenCL 1.1, 1624MB, 1624MB available) 12/8/2012 9:45:33 AM | | Version change (7.0.38 -> 7.0.40) EDIT: Here's the scheduler trying to fetch Intel GPU tasks: 12/8/2012 9:51:41 AM | World Community Grid | Requesting new tasks for CPU and intel_gpu The bolded bit has already been fixed for the next test release: http://boinc.berkeley.edu/trac/changeset/5c13...403c645baddd47c122/boinc# (HT to Claggy on the BOINC forums]. The GPU-Z program tells me the 2670QM CPU has a Intel(R) HD Graphics Family on board but no version/series number, except one box mentions GT2. |
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1403 Status: Offline Project Badges:
|
edit: FTM not getting app_config.xml to work. Start 1 CEP2, with a limit to only run 1, took copy of the application mask from Task Manager, pasted into the proper line. Restarted BOINC and forced a second to start. 2 CEP2 are running. Put a copy in both the data_dir and the project dir to make sure [the wiki does not mention where to place the file], restart BOINC... 2 running. Then per the wiki put the "short name" i.e. cep2 [per the client_state.xml info], still 2 running on restart... going to sleep on this. For me it's working. My WCG prefs are 4 CEPs. I loaded 2 CEPs, waited that 1 was running and suspended all 'ready to start' SIMAP's and 1 running SIMAP. The second CEP didn't start. On my quad 2 SIMAP's and 1 CEP2 running and 1 idle core. Perhaps you oversaw the error on the BOINC-wiki page, where an s is missing in the close tag </gpu_version> |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks for checking that. Not sure how a wrong tag for the GPU affects the functioning for CPU, but I'll be trying it out. There's no tag error warning as what we have for cc_config.xml... in a far future perhaps.:D
----------------------------------------Where did you put the file? Think the most efficient way, since this is at app short name level [excellent not having to mess with version numbers]. One single file in the main data_dir would be great, it's user created, so not sure it should be in the projects folder where WCG controls what goes in and out. edit: OK, insert the missing s, restarted client, fetched more CEP2, suspended all Ready to start with a max allowed of 1 and suspended a running task. Only 1 CEP2 would run, suspending more tasks caused cores to go idle. Unsuspending, immediately resumed the non-CEP2 tasks. Now determine the proper file placement and fix the wiki. [Edit 2 times, last edit by Former Member at Dec 9, 2012 11:06:32 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
So, this is an excellent feature addition. You can have CEP2 run back to back by tweaking the number allowed "in progress" [in device profile], versus number to run [in app_info.xml]. There will always be a few at the top of the FIFO queue waiting for other CEP2 to complete. Hands free maximizing the clean energy contribution, without making the client inefficient or user impaired.
|
||
|
|
|