GPU on Binary Radio Pulsar Search (Perseus) not working
17 Jul 2013 12:08:07 UTC
Topic 197065
(moderation:
Hello,
I am wondering that my computer is not useing with the GPU insinde. My PC is working on a Perseus Arm Survey file around two hours for 1 % so that means 200 hours for 100%. I think this should go faster. The GPU Load on "TechPowerUp GPU-Z0.7.2" is Zero = 0%
BOINC Client 7.0.64 (x64)
OS: Windows 7 Pro SP1
Graphic Card: AMD Radaeon HD7500/7600M Series (latest Driver installed)
CPU: Intel i5-3320M CPU @2,6 GHz
HP Laptop
I allow useing 100% of the CPUs and useing the GPU in the BOINC Manger.
Please give me a tip to get the GPU running.
Philipp
PS: GPU Load on SETI avg. 97%
GPU on Binary Radio Pulsar Search (Perseus) not working
)
Philip,
there is more than one possibility why this does not work.
Can you pls post the first 20 - 25 lines of the messages here?
If I am right your i5 processor has also a built in gpu (HD4000). Sometimes the intel installer routines do a bad job causing confusion with the open-cl drivers.
In short: for AMD you need the CCC, which includes the open-cl package, for Intel you need a correct driver and sometimes one needs to remove the wrong drivers first.
Alex
Hi Alex, Here is the
)
Hi Alex,
Here is the data:
17.07.2013 18:08:44 | | Starting BOINC client version 7.0.64 for windows_x86_64
17.07.2013 18:08:44 | | log flags: file_xfer, sched_ops, task
17.07.2013 18:08:44 | | Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6
17.07.2013 18:08:44 | | Data directory: C:\ProgramData\BOINC
17.07.2013 18:08:44 | | Running under account Philipp
17.07.2013 18:08:44 | | Processor: 4 GenuineIntel Intel(R) Core(TM) i5-3320M CPU @ 2.60GHz [Family 6 Model 58 Stepping 9]
17.07.2013 18:08:44 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 popcnt aes syscall nx lm vmx smx tm2 pbe
17.07.2013 18:08:44 | | OS: Microsoft Windows 7: Professional x64 Edition, Service Pack 1, (06.01.7601.00)
17.07.2013 18:08:44 | | Memory: 3.94 GB physical, 7.87 GB virtual
17.07.2013 18:08:44 | | Disk: 195.21 GB total, 75.35 GB free
17.07.2013 18:08:44 | | Local time is UTC +2 hours
17.07.2013 18:08:44 | | CAL: ATI GPU 0: AMD Radeon HD 6570/6670/7570/7670 series (Turks) (CAL version 1.4.1741, 1024MB, 991MB available, 960 GFLOPS peak)
17.07.2013 18:08:44 | | OpenCL: AMD/ATI GPU 0: AMD Radeon HD 6570/6670/7570/7670 series (Turks) (driver version 1124.2 (VM), device version OpenCL 1.2 AMD-APP (1124.2), 1024MB, 991MB available, 960 GFLOPS peak)
17.07.2013 18:08:44 | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 7964573; resource share 100
17.07.2013 18:08:44 | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7045286; resource share 100
17.07.2013 18:08:44 | World Community Grid | URL http://www.worldcommunitygrid.org/; Computer ID 2368639; resource share 100
17.07.2013 18:08:44 | Einstein@Home | General prefs: from Einstein@Home (last modified 17-Jul-2013 12:29:49)
17.07.2013 18:08:44 | Einstein@Home | Computer location: work
17.07.2013 18:08:44 | Einstein@Home | General prefs: no separate prefs for work; using your defaults
17.07.2013 18:08:44 | | Reading preferences override file
17.07.2013 18:08:44 | | Preferences:
17.07.2013 18:08:44 | | max memory usage when active: 2014.86MB
17.07.2013 18:08:44 | | max memory usage when idle: 3626.75MB
17.07.2013 18:08:44 | | max disk usage: 15.00GB
17.07.2013 18:08:44 | | max download rate: 1024000 bytes/sec
17.07.2013 18:08:44 | | max upload rate: 1024000 bytes/sec
17.07.2013 18:08:44 | | (to change preferences, visit a project web site or select Preferences in the Manager)
17.07.2013 18:08:44 | | Not using a proxy
17.07.2013 18:09:19 | | Suspending computation - initial delay
thanks in advance
philipp
Try this
)
Try this thread:
http://einsteinathome.org/node/196369
RE: I allow useing 100% of
)
Are you saying that you use the 4 cores of the CPU for other intensive tasks? The OpenCL application here at Einstein likes to have one CPU core for itself to play with. If you did not allow that, you get times like you see.
For test sake, try to set BOINC to use 3 CPU cores (On multiprocessors, use 75% of processors), that will leave the last core free for anything else plus the GPU calculations.
Hello Ageless, I´ve tried
)
Hello Ageless,
I´ve tried your recommendations. Now the GPU is working on avg. 80%. Now I got the situation that my CPUs are not running on 100% anymore. Now they work on avg. 80% also.
If it is possible to run the CPUs on 100% and the GPU also (even on 80%) this would be great.
I will see if there is a solution on the link of Alex.
thanks
philipp
RE: I will see if there
)
Philipp,
since it is working now the link I posted will not help you; this was a matter of incompatibility with the intel open-cl drivers.
What you should do next ist do some testing with the number of tasks running on your gpu. You can earn many more credits this way. I use to run 3 tasks on AMD and 2 tasks on Intel.
This is my i3 (2 cores / 4 threads) using all threads to feed the gpu's.
https://dl.dropboxusercontent.com/u/50246791/einstein%20mix2.PNG
Did you already check your processor; does it have a built-in gpu?
Alex
Hello Alex, When I can
)
Hello Alex,
When I can believe wikipedia my i5-3320M CPU got a GPU Type "HD4000" with 16 extension units...
Thanks in advance and greetings from Austria
Philipp
This could give you a big
)
This could give you a big boost in crunching BRP-Arecibo wu's.
http://albert.phys.uwm.edu/forum_thread.php?id=8954&nowrap=true#112691
This is how you could install the correct intel open-cl driver.
In adition you might need an cc_config.xml in your boinc data directory to enable all gpu's
1
in addition to run more than one wu on the intel you need an app_config.xml in the einstein project directory.
einsteinbinary_BRP4
.5
.5
This occupies one CPU but might finish 2 wu's in 25 - 30 min.
But make shure to have sufficient cooling, an optimized setting can easily blow your system! Watch the temps, you can use gadgets from addgadgets.com There are also tools like gpu-z and cpu-z available.
Once you have seen the maximum performance of your system you need to adjust the settings for stable operation within thermal limits. THIS is a MUST on laptops!
Good luck,
Alex
RE: If it is possible to
)
No, not without the GPU application slowing down again. The 3 leftover cores are still running at 100%, in Windows showing at a maximum of 25% per application). If you use an external program to check load and temperatures, such as Core Temp, then it'll show that the load is divided between the 4 cores. But that's normal. It's because you didn't disable the core in hardware, but in BOINC. Your OS will still see the CPU as having 4 cores.
hello Alex, I´ve tried to
)
hello Alex,
I´ve tried to install the driver from the link you sended in the post above. unfortunately i get the error message: "Dieses System erfüllt nicht die Mindestanforderungen für die Installation der Software". OS is Win7 pro.
I´ve download the file "Win64_15319" and "Win64_15313" *.exe or *.zip.
philipp