Want to run Einstein opencl_intel Parkes Beta
29 Nov 2015 3:16:19 UTC
Topic 198328
(moderation:
Enabled iGD in BIOS on a i7 4790K (8 cores with HT)
currently config to 6 cores ( I run 3 parkes per Nvidia GPU)
Intel driver for HD Graphics 4600 is 10.18.15.4279
OS is Windows 10 Pro 64 bit
also - 2 discrete GPUs GTX 970 SC
Boinc is listing both intel gpu AND the Nvidias as GPU0
11/28/2015 6:31:20 PM | | Starting BOINC client version 7.6.9 for windows_x86_64
11/28/2015 6:31:20 PM | | log flags: file_xfer, sched_ops, task
11/28/2015 6:31:20 PM | | Libraries: libcurl/7.39.0 OpenSSL/1.0.2a zlib/1.2.8
11/28/2015 6:31:20 PM | | Data directory: C:\ProgramData\BOINC
11/28/2015 6:31:20 PM | | Running under account Jay
11/28/2015 6:31:25 PM | | CUDA: NVIDIA GPU 0: GeForce GTX 970 (driver version 358.50, CUDA version 7.5, compute capability 5.2, 4096MB, 3346MB available, 4257 GFLOPS peak)
11/28/2015 6:31:25 PM | | CUDA: NVIDIA GPU 1: GeForce GTX 970 (driver version 358.50, CUDA version 7.5, compute capability 5.2, 4096MB, 3346MB available, 4257 GFLOPS peak)
11/28/2015 6:31:25 PM | | OpenCL: NVIDIA GPU 0: GeForce GTX 970 (driver version 358.50, device version OpenCL 1.2 CUDA, 4096MB, 3346MB available, 4257 GFLOPS peak)
11/28/2015 6:31:25 PM | | OpenCL: NVIDIA GPU 1: GeForce GTX 970 (driver version 358.50, device version OpenCL 1.2 CUDA, 4096MB, 3346MB available, 4257 GFLOPS peak)
11/28/2015 6:31:25 PM | | OpenCL: Intel GPU 0: Intel(R) HD Graphics 4600 (driver version 10.18.15.4279, device version OpenCL 1.2, 1922MB, 1922MB available, 200 GFLOPS peak)
11/28/2015 6:31:25 PM | | OpenCL CPU: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz (OpenCL driver vendor: Intel(R) Corporation, driver version 5.2.0.10083, device version OpenCL 1.2 (Build 10083))
11/28/2015 6:31:25 PM | SETI@home | Found app_info.xml; using anonymous platform
11/28/2015 6:31:25 PM | | Host name: JBird
11/28/2015 6:31:25 PM | | Processor: 8 GenuineIntel Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz [Family 6 Model 60 Stepping 3]
11/28/2015 6:31:25 PM | | 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 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 pbe fsgsbase bmi1 smep bmi2
11/28/2015 6:31:25 PM | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.10240.00)
11/28/2015 6:31:25 PM | | Memory: 31.69 GB physical, 63.69 GB virtual
11/28/2015 6:31:25 PM | | Disk: 223.03 GB total, 124.29 GB free
11/28/2015 6:31:25 PM | | Local time is UTC -6 hours
11/28/2015 6:31:25 PM | Einstein@Home | Found app_config.xml
11/28/2015 6:31:25 PM | SETI@home | Found app_config.xml
11/28/2015 6:31:25 PM | | Config: run apps at regular priority
11/28/2015 6:31:25 PM | | Config: use all coprocessors
11/28/2015 6:31:25 PM | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 11913206; resource share 25
11/28/2015 6:31:25 PM | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 7589235; resource share 200
11/28/2015 6:31:25 PM | Einstein@Home | General prefs: from Einstein@Home (last modified 28-Nov-2015 12:36:52)
11/28/2015 6:31:25 PM | Einstein@Home | Computer location: home
11/28/2015 6:31:25 PM | | General prefs: using separate prefs for home
11/28/2015 6:31:25 PM | | Reading preferences override file
11/28/2015 6:31:25 PM | | Preferences:
11/28/2015 6:31:25 PM | | max memory usage when active: 32452.81MB
11/28/2015 6:31:25 PM | | max memory usage when idle: 32452.81MB
11/28/2015 6:31:25 PM | | max disk usage: 50.00GB
11/28/2015 6:31:25 PM | | max CPUs used: 6
current app_config is:
einsteinbinary_BRP6
0.33
0.2
Do I need a custom entry in app_config referencing intel GPU?
What other specific settings(prefs or computing) are needed
other than 100% cores?
Want to run Einstein opencl_intel Parkes Beta
)
Have you read the various threads about the potential issues when trying to run tasks on Intel GPUs? In case you haven't, I'll list them here. There may be other relevant messages in other threads as well.
The original announcement contains a lot of discussion. There was also a separate feedback thread and comments have continued there until quite recently. With the advent of Windows 10, there is now an active thread specific to that version. Hopefully, between the three, you should find everything you need to know to avoid problems ranging from getting tasks in the first place to avoiding invalid results due to bad driver versions. I don't run any Windows machines so I can't help you with driver related advice.
BOINC isn't really doing that. It's saying that you have two NVIDIA GPUs, GPU 0 and GPU 1 and that these are both CUDA capable and OpenCL capable. It's also saying you have an Intel GPU 0 - a HD 4600 which is OpenCL capable. NVIDIA GPU 0 is NOT the same as Intel GPU 0 because the manufacturer is different. BOINC is not confused :-).
No, unless you want to attempt running more than 1 task at a time. You'd be unwise to attempt that until you were certain you could get valid results running singly. You need to seek advice about potential issues if you try at any stage running more than 1 internal GPU task.
You need to allow Intel GPU work.
You may need to connect a monitor or dummy plug to the internal GPU (or extend the desktop).
You need to check and confirm (from the links above) that you are using a known good driver version.
You need to check if there might be adverse consequences for the discrete GPU tasks by running on the internal GPU. I suspect there might be.
You should carefully read the Windows 10 thread to be forewarned about any issues.
You should ask specific questions there if anything is not clear or if something unexpected happens while trying to get things going.
Good luck with finding a satisfactory configuration.
Cheers,
Gary.
Thanks Gary I appreciate it
)
Thanks Gary I appreciate it
You're welcome! Let us all
)
You're welcome!
Let us all know how you get on. If you learn any tricks, make sure you write them up in the Windows 10 thread for the benefit of others. If you find things that don't work, write them up too.
Thanks.
Cheers,
Gary.
In the past I found that
)
In the past I found that running tasks on my 4790K's iGPU caused a big hit on my discreet GPU processing times so it will be interesting to see if/ how the new Win10 drivers play in your configuration.
Happy crunching,
Steve
--------------------------
- Crunch, Crunch, Crunch -
--------------------------
OK 2 questions: Does
)
OK 2 questions:
Does 10.18.15.4279 driver work?
and is this the right syntax to replace app_config or do I need app_config head and tail for it?
einsteinbinary_BRP6
BRP6-Beta-cuda55
0.2
0.33
einsteinbinary_BRP6
BRP6-Beta-opencl-intel_gpu
1.0
1.0