will installing the APP SDK hurt anything or prevent this new applications from working correctly? as ask b/c the machine i plan to try this on already crunches Milkyway@Home on an AMD/ATI GPU (HD 6950 2GB to be exact), and i'm not entirely sure that the APP SDK isn't required for MW@H...
____________

will installing the APP SDK hurt anything or prevent this new applications from working correctly? as ask b/c the machine i plan to try this on already crunches Milkyway@Home on an AMD/ATI GPU (HD 6950 2GB to be exact), and I'm not entirely sure that the APP SDK isn't required for MW@H...

The SDK and the driver installer may both install (different!) versions of a certain 32-bit lib (at least under Linux) at the same time and this might cause the E@H app to fail while trying to even find the GPU, so very early in the execution. Since your machines are Windows only, I'd suggest to give the E@H app a try even with the SDK installed. If it fails just after start with something like this in its log:

> [15:49:08][7370][ERROR] Failed to get OpenCL platform/device info from BOINC (error: -1)!
> [15:49:08][7370][ERROR] Demodulation failed (error: -1)!

then you might encounter this particular problem.

But I'm rather optimistic this is not going to happen under Windows.

CU
HB

Is there a way to use this with an intel GPU like sandy or ivy bridge?

My understanding is that nowadays OpenCL runtime support is included in the default 'Catalyst software suite'. The current description on the download pages is 'OpenCL(tm) Driver', though in the past it has been called 'APP' (accelerated parallel processing) or even SDK.

There should never be any need for an end-user to download the quite separate Software Development Kit just to run a pre-compiled OpenCL application either here or at MilkyWay.

@Sunny129: could you PM me a link to the source of your suggestion that MilkyWay requires the full development kit, rather than just the driver/runtime version, please? This terminology confusion keeps cropping up at various message boards, and I'd like to get to the bottom of it.

Great news! Thanks Oliver & HB :-)

BOINC 7.0.27 seems to return that my Radeon HD 7970 doesn't support OpenCL. I have Catalyst 12.4 and my other OpenCL codes run OK.

http://einstein.phys.uwm.edu/show_host_detail.php?hostid=5359846
http://einstein.phys.uwm.edu/host_sched_logs/5359/5359846

Thoughts?
Gaurav
____________

My understanding is that nowadays OpenCL runtime support is included in the default 'Catalyst software suite'. The current description on the download pages is 'OpenCL(tm) Driver', though in the past it has been called 'APP' (accelerated parallel processing) or even SDK.

There should never be any need for an end-user to download the quite separate Software Development Kit just to run a pre-compiled OpenCL application either here or at MilkyWay.

@Sunny129: could you PM me a link to the source of your suggestion that MilkyWay requires the full development kit, rather than just the driver/runtime version, please? This terminology confusion keeps cropping up at various message boards, and I'd like to get to the bottom of it.

i found the info in the All work Units giving "Computational Error" thread on the Milkyway@Home forums. if you scroll down to message 53419 (posted by user Assimilator1), you'll see that he was getting WU errors. Matt Arsenault, one of the head project developers, somehow noticed that Assimilator1's Catalyst 12.1 driver installation was missing the OpenCL.dll, and suggested that he reinstall the drivers, paying special attention not to uncheck the APP SDK option. sure enough, that's what Assimilator1 had done on the first try, thinking that he didn't need the entire developer kit...and sure enough, reinstalling the drivers without excluding the SDK solved his problem.

Yes, this machine was on Albert too:
http://albert.phys.uwm.edu/show_host_detail.php?hostid=2803

Yes, both 32 & 64 bit OpenCL are in /usr/lib and /usr/lib32 (I just checked).

The difference is really the boinc 7.0.27 (vs 7.0.26 on Albert) and also the fact that I downgraded to Ubuntu 10 (from 11) because I was having some unrelated problems.

OK. I will try 7.0.26 and see what happens.
____________

OK. BTW, its not working on Albert anymore either. Same issue. So, its likely something with my system .. let me dig deeper.

http://albert.phys.uwm.edu/host_sched_logs/2/2955
____________

OK, not a geek here. I am running ver 7.0.25. E@home and Seti@home. Do I need to upgrade to 7.0.27. If so, where do I find it?

OK, not a geek here. I am running ver 7.0.25. E@home and Seti@home. Do I need to upgrade to 7.0.27. If so, where do I find it?

May be it is wrong, but i dont see opencl.dll in windows\system32 after installing Catalyst 12.4 from amd site.
no opencl.dll - no open cl :)
No capable GPU - from Boinc.

Hi all,

I have a HP Pavilion dv7 laptop which incorporates a Radeon HD 6490M graphics card.

I have checked through HP Support for my driver version and it shows 8.830.6.3000 which is of course is well below the 12.4 you suggest.

I have been to the AMD site to update but their program states that i should use my manufacturer version which is what is currently installed.

Does this eliminate my use of graphics processing for E@H ?

Can anyone give me assistance in getting this to work?

I have emailed HP for advice also as I wish to push some processing from CPU to GPU to help keep my laptop cool.

As i understand there is no open cl driver in catalyst 12.2-12.4. Only in 12.1.
Simply run gpu-z and check about Opencl tab.
If no Open cl - wait for new drivers from hp or install SDK.

I'm looking forward to the OSX Mountain Lion version.

Greetings Oliver

Congratulations on this major achievement, I can now GPU crunch for Einstein. Have successfully crunched the following workunits

http://einstein.phys.uwm.edu/workunit.php?wuid=123076333
http://einstein.phys.uwm.edu/workunit.php?wuid=123232447

Using the following system setup

Intel Quad 8400, Win7 x64, Boinc 7.0.27 Catalyst 12.4. HD 5850 GPU Card

I am running 3 CPU YOYO workunits and 1 free core for two GPU tasks of Einstein, as they are using 0.5 cpu each.

Regards
Tazzduke

i'm assuming that reducing "CPU consumption per GPU task" is one of the developers' goals in the attempt to get this app on par w/ the CUDA app? IIRC, BRP4 CUDA tasks only use 0.2 CPUs each...
____________

I am running an ATI Radeon HD 3400 Series. Will this work?

It requires OpenCL 1.1, which was introduced with HD 5000 series for Radeons, so no.
____________
Soli Deo Gloria

Excellent news. My 7870 on my new system and my 5770 on my old system are currently crunching away. I haven't reduced core usage, but then for heat/noise reasons I don't let any projects use more than about 40% total CPU (doing that leaves all my fans turning at minimal RPMs which is good). We'll see if I have any issues once a unit is completed but so far so good. It doesn't seem to be quite as efficient as the crunching my work laptop with an NVidia 4200 is doing, but that's at first blush and as you said this is an initial release.

I'm just happy that my GPU can help with the work.

May be it is wrong, but i dont see opencl.dll in windows\system32 after installing Catalyst 12.4 from amd site.
no opencl.dll - no open cl :)
No capable GPU - from Boinc.

Oliver,

I'm experiencing the same problem I had over @Albert:
Some artifacts appear on screen during the initial start-up phase.
The WU continue crunching, finished & reported.

But, the WU is flagged as validate error.

I had reported the same over @Albert (May-1st), but so far no response.

This mobile workstation crunched OpenCL WUs for SET AP/MB, DRTG, POEM++, PrimGrid and WCG (hcc-beta) all without any problems.

It had been crunching x3 POEM WUs at a time over past 2+ months without a single invalid WU. Host details @POEM.

Please, can you or anyone help me?

Thanks.


____________
terencewee*
Sicituradastra.

Thanks Oliver.

11.11 is the latest WHQL driver from AMD for FirePro m7820.

I'm downloading 12.4a mobility and will mod, install and report back whether the problem persists.


____________
terencewee*
Sicituradastra.

If someone manages to get everything working smoothly on 64-bit Ubuntu 10 (10.04 or 10.10), please do let me know. I think I'm having trouble because I want to stick to these. I have tried a number of things since yesterday and I can't seem to get BOINC 7.0.27 to return that I have a working OpenCL platform / driver.

Oliver, HB .. what distribution / version of Linux have you folks been using / testing on?

Thanks,
Gaurav
____________

Yes. clinfo works and my own OpenCL codes work fine. The same machine was doing fine with Ubuntu 11.04 for several months on Albert. I downgraded to 10.10 and now I can't get it to work!
____________

find /usr -name "libOpenCL*" | xargs file
/usr/lib32/libOpenCL.so.1: ELF 32-bit LSB shared object, Intel 80386, version 1 (SYSV), dynamically linked, not stripped
/usr/lib/libOpenCL.so.1: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, not stripped


ldd einsteinbinary_BRP4_1.24_x86_64-pc-linux-gnu-eah__atiOpenCL
linux-gate.so.1 => (0xf778a000)
libOpenCL.so.1 => /usr/lib32/libOpenCL.so.1 (0xf7768000)
libpthread.so.0 => /lib32/libpthread.so.0 (0xf774e000)
libm.so.6 => /lib32/libm.so.6 (0xf7728000)
libstdc++.so.6 => /usr/lib32/libstdc++.so.6 (0xf763d000)
libc.so.6 => /lib32/libc.so.6 (0xf74e2000)
/lib/ld-linux.so.2 (0xf778b000)
libdl.so.2 => /lib32/libdl.so.2 (0xf74de000)
libgcc_s.so.1 => /usr/lib32/libgcc_s.so.1 (0xf74c1000)

This looks good. The issue is really with BOINC that it refuses to report that I have an OpenCL supported device/driver.

____________

HB .. looks like I already have those 32-bit libs:

ldd einsteinbinary_BRP4_1.00_graphics_i686-pc-linux-gnu
linux-gate.so.1 => (0xf771b000)
libpthread.so.0 => /lib32/libpthread.so.0 (0xf76e6000)
libm.so.6 => /lib32/libm.so.6 (0xf76bf000)
libdl.so.2 => /lib32/libdl.so.2 (0xf76bb000)
libX11.so.6 => /usr/lib32/libX11.so.6 (0xf759e000)
libXext.so.6 => /usr/lib32/libXext.so.6 (0xf758e000)
libGL.so.1 => /usr/lib32/mesa/libGL.so.1 (0xf753b000)
libGLU.so.1 => /usr/lib32/libGLU.so.1 (0xf74cb000)
libc.so.6 => /lib32/libc.so.6 (0xf736f000)
/lib/ld-linux.so.2 (0xf771c000)
libxcb.so.1 => /usr/lib32/libxcb.so.1 (0xf7355000)
libXdamage.so.1 => /usr/lib32/libXdamage.so.1 (0xf7351000)
libXfixes.so.3 => /usr/lib32/libXfixes.so.3 (0xf734b000)
libXxf86vm.so.1 => /usr/lib32/libXxf86vm.so.1 (0xf7345000)
libdrm.so.2 => /lib32/libdrm.so.2 (0xf733a000)
libstdc++.so.6 => /usr/lib32/libstdc++.so.6 (0xf724f000)
libgcc_s.so.1 => /usr/lib32/libgcc_s.so.1 (0xf7233000)
libXau.so.6 => /usr/lib32/libXau.so.6 (0xf722f000)
libXdmcp.so.6 => /usr/lib32/libXdmcp.so.6 (0xf7229000)
librt.so.1 => /lib32/librt.so.1 (0xf721f000)

My own OpenCL apps .. I compiled and tested both the 32-bit & 64-bit versions yesterday. They work fine.
____________

Yes, please take a look now. I have switched back to 7.0.27 and also included the cc_config.xml

<cc_config>
<log_flags>
<coproc_debug>1</coproc_debug>
</log_flags>
</cc_config>

http://einstein.phys.uwm.edu/host_sched_logs/5363/5363793
____________

Oliver,

Installed 12.4a without requiring mod, recognized as Mobility 5800 series.

Upon WU startup encountered driver restarts at ~0.038%.
GPU-Z 0.6.2 shows a quick uptick of dedicated memory usage then flat-lined (even though the WU ran for another 10secs before suspension).

Updating to latest driver doesn't work.
Any other ideas?

EDIT: POEM++ OpenCL WUs are now running ~15mins faster! :)

EDIT2: Scratch that - it's ~29mins (!!!) faster.

Present clinfo:

Number of platforms: 1
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 1.2 AMD-APP (923.1)
Platform Name: AMD Accelerated Parallel Processing
Platform Vendor: Advanced Micro Devices, Inc.
Platform Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices cl_khr_d3d10_sharing


Platform Name: AMD Accelerated Parallel Processing
Number of devices: 2
Device Type: CL_DEVICE_TYPE_GPU
Device ID: 4098
Board name: ATI Mobility Radeon HD 5800 Series
Max compute units: 10
Max work items dimensions: 3
Max work items[0]: 256
Max work items[1]: 256
Max work items[2]: 256
Max work group size: 256
Preferred vector width char: 16
Preferred vector width short: 8
Preferred vector width int: 4
Preferred vector width long: 2
Preferred vector width float: 4
Preferred vector width double: 0
Native vector width char: 16
Native vector width short: 8
Native vector width int: 4
Native vector width long: 2
Native vector width float: 4
Native vector width double: 0
Max clock frequency: 700Mhz
Address bits: 32
Max memory allocation: 536870912
Image support: Yes
Max number of images read arguments: 128
Max number of images write arguments: 8
Max image 2D width: 8192
Max image 2D height: 8192
Max image 3D width: 2048
Max image 3D height: 2048
Max image 3D depth: 2048
Max samplers within kernel: 16
Max size of kernel argument: 1024
Alignment (bits) of base address: 2048
Minimum alignment (bytes) for any datatype: 128
Single precision floating point capability
Denorms: No
Quiet NaNs: Yes
Round to nearest even: Yes
Round to zero: Yes
Round to +ve and infinity: Yes
IEEE754-2008 fused multiply-add: Yes
Cache type: None
Cache line size: 0
Cache size: 0
Global memory size: 1073741824
Constant buffer size: 65536
Max number of constant args: 8
Local memory type: Scratchpad
Local memory size: 32768
Kernel Preferred work group size multiple: 64
Error correction support: 0
Unified memory for Host and Device: 0
Profiling timer resolution: 1
Device endianess: Little
Available: Yes
Compiler available: Yes
Execution capabilities:
Execute OpenCL kernels: Yes
Execute native function: No
Queue properties:
Out-of-Order: No
Profiling : Yes
Platform ID: 000007FEE9D22A08
Name: Juniper
Vendor: Advanced Micro Devices, Inc.
Device OpenCL C version: OpenCL C 1.2
Driver version: CAL 1.4.1720 (VM)
Profile: FULL_PROFILE
Version: OpenCL 1.2 AMD-APP (923.1)
Extensions: cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_atomic_counters_32 cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_popcnt cl_khr_d3d10_sharing
____________
terencewee*
Sicituradastra.

Huh .. seems to say no OpenCL libs found???

microway@ubuntu:~/Gaurav/EaH$ ./boinccmd --get_messages
1: 16-May-2012 11:38:30 (low) [] Starting BOINC client version 7.0.27 for x86_64-pc-linux-gnu
2: 16-May-2012 11:38:30 (low) [] log flags: file_xfer, sched_ops, task, coproc_debug
3: 16-May-2012 11:38:30 (low) [] Libraries: libcurl/7.21.0 OpenSSL/1.0.0e zlib/1.2.3.4 libidn/1.18
4: 16-May-2012 11:38:30 (low) [] Data directory: /home/microway/Gaurav/EaH
5: 16-May-2012 11:38:30 (low) [] Processor: 16 GenuineIntel Intel(R) Xeon(R) CPU E5-2660 0 @ 2.20GHz [Family 6 Model 45 Stepping 7]
6: 16-May-2012 11:38:30 (low) [] Processor: 20.00 MB cache
7: 16-May-2012 11:38:30 (low) [] 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 ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm dca sse4_1 sse4_2 x2apic popcnt aes xsave avx lahf_lm ida arat epb pln pts dts tpr_shadow vnmi flexpriority ept vpid
8: 16-May-2012 11:38:30 (low) [] OS: Linux: 2.6.35-32-generic
9: 16-May-2012 11:38:30 (low) [] Memory: 31.47 GB physical, 18.88 GB virtual
10: 16-May-2012 11:38:30 (low) [] Disk: 439.86 GB total, 404.98 GB free
11: 16-May-2012 11:38:30 (low) [] Local time is UTC -4 hours
12: 16-May-2012 11:38:30 (low) [] ATI GPU 0: AMD Radeon HD 79x0 series (Tahiti) (CAL version 1.4.1720, 3072MB, 2919MB available, 50 GFLOPS peak)
13: 16-May-2012 11:38:30 (low) [] No NVIDIA library found
14: 16-May-2012 11:38:30 (low) [] No OpenCL library found
15: 16-May-2012 11:38:30 (low) [Einstein@Home] URL http://einstein.phys.uwm.edu/; Computer ID 5363793; resource share 100
16: 16-May-2012 11:38:30 (low) [Einstein@Home] General prefs: from Einstein@Home (last modified 13-Apr-2012 19:31:22)
17: 16-May-2012 11:38:30 (low) [Einstein@Home] Computer location: home
18: 16-May-2012 11:38:30 (low) [Einstein@Home] General prefs: no separate prefs for home; using your defaults
19: 16-May-2012 11:38:30 (low) [] Preferences:
20: 16-May-2012 11:38:30 (low) [] max memory usage when active: 32228.60MB
21: 16-May-2012 11:38:30 (low) [] max memory usage when idle: 32228.60MB
22: 16-May-2012 11:38:30 (low) [] max disk usage: 10.00GB
23: 16-May-2012 11:38:30 (low) [] max CPUs used: 1
24: 16-May-2012 11:38:30 (low) [] (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
25: 16-May-2012 11:38:30 (low) [] Not using a proxy
26: 16-May-2012 11:38:30 (low) [Einstein@Home] Restarting task h1_0367.90_S6GC1__1818_S6LV1B_1 using einstein_S6LV1 version 113 (X64) in slot 0
27: 16-May-2012 11:38:30 (low) [Einstein@Home] Sending scheduler request: To fetch work.
28: 16-May-2012 11:38:30 (low) [Einstein@Home] Requesting new tasks for ATI
29: 16-May-2012 11:38:33 (low) [Einstein@Home] Scheduler request completed: got 0 new tasks
30: 16-May-2012 11:38:33 (low) [Einstein@Home] No work sent
31: 16-May-2012 11:38:33 (low) [Einstein@Home] see scheduler log messages on http://einstein.phys.uwm.edu//host_sched_logs/5363/5363793


____________

OK, but the link doesn't work

thanks

The BOINC server is currently down and will be for a while.

BM

It is a short at a substation causing a power outage to the hill area where the Space Sciences lab is located. Traffic signals are out on the hill, all the homes in that area have no power.

That means that the whole of BOINC and SETI will be offline until power has been restored and the lab equipment (including servers) has been checked over.

Ah .. got it to work! BOINC looks for "libOpenCL.so" and I only had "libOpenCL.so.1" for some reason. Some simple soft links solved that problem. :-)

Thanks everyone. I'm happy now.
____________

I started running GPU WU's yesterday, two of these have been granted credit so far.

The granted credit seems very low in comparison to other GPU projects.

http://einstein.phys.uwm.edu/result.php?resultid=288577533 was granted 500 credits for approx 4661 seconds of GPU (9200 per day).

http://einstein.phys.uwm.edu/result.php?resultid=288548856 was also granted 500 credits for 8654 seconds (5000 per day).

On MilkyWay, this system achieves between 80,000 and 140,000 credits per day (avg ~110).

On Collatz, this system achieved between 120,000 and 140,000 credits per day.
____________

MW and Collatz are both KNOWN for giving out WAY too many credits.

Same as PrimeGrid too.

Credits are useless anyways IMHO, but yes Einstein does seem to be on the low end for GPU credit.

GPUgrid gives bonus points if returned quickly, which I think is smart, but if you only care about points, than Einstein is not your place.

In terms of how long PV can take, currently with one GPU, 680, I would on average have about 120 in PV, since I've added my 670 last night, this number is now at 160, and expect it to go towards 220-240.

This is using GPU utilization of .33 on both cards. Would be nice, not necessary, to not have these wait so long for validation. Again, don't really care about points, but I just hate seeing so many PV's personally.

Happy crunching, and congratulations on getting OpenCl out, and so far I haven't read any serious issues on these forums. Good job

Great work on developing the OpenCL applications and bringing the applications to production. Congratulations.

In terms of how long PV can take, currently with one GPU, 680, I would on average have about 120 in PV, since I've added my 670 last night, this number is now at 160, and expect it to go towards 220-240.

I'll post in the crunching thread, don't want to hijack

Hi do somone have app_info for only Amd/Ati GPU or make one for my?

Is anyone else having problems getting to berkley.edu(BOINC CLIENT) besides myself?
____________

Is anyone else having problems getting to berkley.edu(BOINC CLIENT) besides myself?

Everyone has. See Richard's post earlier in this thread, and mine for our local copies of 7.0.28.

BM

I need app_info.xml for my Gpu only runs in 56%

I need app_info.xml for my Gpu only runs in 56%

Tack you Sunny129,I have missed that.

About OpenCL :)
All was very easy.
So i use WinXP and openCL driver fron Catalyst XP.
And OpenCL included only in 12.1
In 12.2-12.4 there is no Opencl drivers.
In Catalyst for Vista/7 opencl included in Catalyst 12.4.
Why so? I dont know.

Hello all,

first: great work done, just installed the new boinc client version, and my gpu does its jobs. But one question remains:

Is there an option to stop only BOINC's GPU utilization during exclusive application execution? I don't want to stop the calculations on the CPUs, but GPU-apps should stop if a game or something like that is started.

Thx, mahdia
____________

Is there an option to stop only BOINC's GPU utilization during exclusive application execution? I don't want to stop the calculations on the CPUs, but GPU-apps should stop if a game or something like that is started.

Is there an option to stop only BOINC's GPU utilization during exclusive application execution? I don't want to stop the calculations on the CPUs, but GPU-apps should stop if a game or something like that is started.

Sure, BOINC's activity menu let's you suspend it completely, or just the GPU tasks. You may have to switch to Advanced View. If you're on Windows, you can also use the BOINC icon in the system tray (right click on it)...

Hope this helps,

Oliver

Hi everyone,

We're proud to announce the official release of our first application for AMD/ATI graphics cards (GPU) and accelerated processing units (APU). Please use this thread to discuss this release.

Minimum Requirements:

• Windows or Linux
  
• BOINC Client 7.0.27 (download)
  
• ATI GPU or APU (OpenCL 1.1 compliant, equivalent to Radeon HD 5xxx)
  
• 512 MB video memory
  
• Catalyst Driver 12.x (don't install the APP SDK!)
  

Notes:

• Please be aware that all previous tests we ran could only do a limited amount of testing given the vast amount of different hard- and software run by our volunteers. There might still be smaller issues, in particular with validation. Please bear with us, we'll do our best to improve the application over the next weeks and months.
  
• The same applies to application performance. This is our first official release. It might not be on par with our CUDA application just yet. We still have ideas on how to improve application performance and we're going to introduce them in due course.
  
• Tip: performance might be improved if you set BOINC not to use all your CPU cores (e.g. all but one).
  
• Support for Apple Mac OS X is currently targeted for OS X Mountain Lion (10.8)
  

Known Issues:

• When running this application (all 32-bit) on 64-bit systems you might encounter the following error message (error number 255/-1):
  [ERROR] Failed to get OpenCL platform/device info from BOINC (error: -1)
  If that happens, please download the latest AMD/ATI Catalyst driver, reinstall it and reboot your computer.
  

Happy crunching!

Oliver

I am running two machines one with Windows 64 bit and the other with Ubuntu 12.04. after upgrading to Ubuntu my boinc just quit working. A version of boinc came with Ubuntu version 7.0.24 I want to install ver 7.0.27 and not sure how to do it. I just started using Ubuntu and sill learning. Have not been able to remove the older version of Boinc and installing the new version. Can anyone assist me?
____________

However, large parts of the CPU efficiency are contributed by the GPU driver and the way it yields CPU resources to the OS (other processes) while doing GPU computations. Bottom line, you should probably never compare CUDA/NVIDIA to OpenCL/ATI directly. The same applies to comparing actual performance since NVIDIA and ATI GPUs have significantly different architectures. Another reason is that according our experiences so far, ATI GPUs cover a much wider range in relative performance than NVIDIA GPUs. So your mileage may vary significantly compared to other volunteers, depending on the actual ATI GPU model you use.

Yes, instead of using <exclusive_app> use <exclusive_gpu_app> instead, see the following for details:

Claggy

Please, can anyone explain what means
"the "dangerous" option of 0.5 2tasks at once"
and
"app_info.xml file - a factor of 1 runs 1 task, a factor of 0.5 runs 2 simultaneous tasks, a factor of 0.33 runs 3 simultaneous tasks, and so on and so forth..."
and what about CUDA OpenCL - i'm going to upgrade to GTX680 - will it be able to compute too? or will I have to use FireGL?
____________

Please, can anyone explain what means
"the "dangerous" option of 0.5 2tasks at once"
and
"app_info.xml file - a factor of 1 runs 1 task, a factor of 0.5 runs 2 simultaneous tasks, a factor of 0.33 runs 3 simultaneous tasks, and so on and so forth..."
and what about CUDA OpenCL - i'm going to upgrade to GTX680 - will it be able to compute too? or will I have to use FireGL?

<coproc>
<count>n</count>
</coproc>

Here is our local copy of BOINC 7.0.28 for Windows, built by Rom today:

* boinc_7.0.28_windows_x86_64.exe
* boinc_7.0.28_windows_intelx86.exe

BM

...where "n" is the GPU utilization factor. when set to a value of 1, the BRP4 GPU apps would run only 1 GPU task at a time. when set to 0.5, the apps would run 2 GPU tasks simultaneously. when set to 0.33, the apps would run 3 GPU tasks simultaneously, and so on and so forth...

...where "n" is the GPU utilization factor. when set to a value of 1, the BRP4 GPU apps would run only 1 GPU task at a time. when set to 0.5, the apps would run 2 GPU tasks simultaneously. when set to 0.33, the apps would run 3 GPU tasks simultaneously, and so on and so forth...

Thank you for a good explanation. Could u please speciafy a little:

I have a factor of 0,5 by default for Radeon 5750. Is this good or bad? I didn't change anything.

Can anyone tell me what is new in boinc 28 ?
I have 5850 1 GB and util. factor 0.5, but boin run only 1 opencl task.
Can someone help me ? :)

Can anyone tell me what is new in boinc 28 ?
I have 5850 1 GB and util. factor 0.5, but boin run only 1 opencl task.
Can someone help me ? :)

@rahid:
maybe you did not delete an old app_info.xml?
____________

i did it throw control panel boinc :)
may be some days to wait, as boinc sad :)
on gtx 2 jobs downloaded and work, will test some days later on radeon.

Hi there,
I have a PC with a Pentium 4 CPU 1500Mhz, 1.5 Ghz. 2.0GB of ram,
Windows XP SP3, ATI Radeon 9200 Series Graphics card. Can I upgrade Boinc and now be able to use my GPU ?

Thanks,

Ivor
____________

Einstein will not crunch two BRPs units at once. Albert has no problem with doing so...
Is Albert twice as smart as Einstein.
Einstein will only do one unit at a time on my hd5770 . I ve checked and double checked my preferences
Einstein preferences are the same as Albert as far as I can tell.In both cases I ve set the use .5
option. Einstein is using .5 CPUs + 1 ATI GPU per unit whereas Albert is using .879 CPUs + .5 ATI GPUs.
Is anyone else having this problem

Is there a difference between these work units such that the top one needs to hog all the GPU
but only uses .5 CPU

Einstein...
Binary Radio Pulsar Search (Arecibo) v1.24(atiOpenCL)

Albert...
Binary Radio Pulsar Search v1.24 (atiOpenCL)

So what does Albert have that Einstein don t (sorry about the lack of proper punctuation - keyboard problems)

Hi there,
I have a PC with a Pentium 4 CPU 1500Mhz, 1.5 Ghz. 2.0GB of ram,
Windows XP SP3, ATI Radeon 9200 Series Graphics card. Can I upgrade Boinc and now be able to use my GPU ?

Thanks,

Ivor

ATI Radeon 9200 Series Graphics card.

I have the 7.0.28(x64) version running on a Windows 7 Ultimate 64 bit machine and it seems to be working ok, except that I sometimes get messages like below:

5/20/2012 9:48:28 PM | Einstein@Home | Task h1_0435.95_S6GC1__1142_S6LV1B_0 exited with zero status but no 'finished' file
5/20/2012 9:48:28 PM | Einstein@Home | If this happens repeatedly you may need to reset the project.



I have tried resetting the project but it still happens. I've had like 6 instances of this in 4 days so it's not a complete show stopper, but if anyone has knowledge or ideas about what might be causing this I would appreciate the info.

Thanks

See this FAQ for possible solutions.
____________
Jord

-The BOINC FAQ Service
- BOINC 7.0 FAQ

I used to be an adventurer like you. Then I took an arrow in the knee...

Hi guys, its a double precision vga needed?

im owner of 6850 :)

thx

p.
____________

Hi guys, its a double precision vga needed?

im owner of 6850 :)

thx

p.

Evening,

At the risk of posting to the wrong thread, should I be seeing an improvement in computation using the ATI/OpenCL app?

My ATI WUs are turning in at about 12 hours on GPU; CPU app was around 5 hours per unit. Stats graph shows a linear progression exactly as before...

Platform details: BOINC 7.0.28(x64) on Intel i650, 8GB RAM, ATI5570, Win7-64, Catalyst 12.4

I can't compare directly, as the GPU app has been doing, e.g.

Binary Radio Pulsar Search (Arecibo) v1.24 (atiOpenCL)
Run time CPU time Claim Granted
43,973.45 8,347.55 65.43 500.00

and CPU, e.g.

Gravitational Wave S6 LineVeto search v1.13 (SSE2)
Run time CPU time Claim Granted
25,863.85 20,980.90 74.98 251.18

Granted CPU-time is lower for the GPU but does that look odd or just not optimal? The 650 is a 73W part and the ATI a 39W board. Am I just that underpowered on the GPU front?

http://ark.intel.com/products/43546/Intel-Core-i5-650-Processor-(4M-Cache-3_20-GHz)
http://www.amd.com/us/products/desktop/graphics/ati-radeon-hd-5000/hd-5570/Pages/hd-5570-overview.aspx#2

(GTX690 owners take note!)

I solved the problem, resetting Einstein worked.

Hello, friends!
Is there any way to specify which of 2 video cards installed in my system E@H should use for computing?

Why: I'm going to upgrade to GTX680 (GAINWARD PHANTOM) but it is silent when idle and very noisy uder load. SO I want my current 5750 with passive cooling installed as a second card in 4x PCIE slot do all the job. Is it possible?
____________

Hello, friends!
Is there any way to specify which of 2 video cards installed in my system E@H should use for computing?

Why: I'm going to upgrade to GTX680 (GAINWARD PHANTOM) but it is silent when idle and very noisy uder load. SO I want my current 5750 with passive cooling installed as a second card in 4x PCIE slot do all the job. Is it possible?

hmm

I have a HD 4850, which has a rather limited but still functional support for OpenCL. Im guessing that the OpenCL version you used for the application is upgraded so much so that the 4xxx series is rendered incompatible? If not, i would really like to contribute my 4850 to Einstein@Home.

And to think I bought the 4850 2 months before the 5xxx series came out >.<

Am I just that underpowered on the GPU front?

Yes, most likely. The ATI GPU 5xxx series varies quite widely in performance and it also depends on how much CPU power is left to "feed" the GPU. It might turn out that your specific setup is simply not powerful enough to produce a significant speed-up when using the GPU.

Sorry,
Oliver

Yes. It is :)
I have 5850+c2d e7600.
It's more better to run 1 gpu task+1 cpu task.
If i will run another cpu task, it's not enough cpu resources for ati :(.
Yes priority is higher in gpu task, but this dont work on ati :(
But work's on Cuda :)
And if i will run 2 gpu task ( 1gb video ), but pci-e 1.1(may be this is reason) , speed dramatically slow down :(
And about computation error on gpu.
I have such errors after standby/resume or hibernate.
On ATi and Nvidia :)
And now i simply dont use this :)
And exclude boinc work directiry from Kaspersky scan.

I have an intel i5 quadcore, 8GB ram, 64 bit windows 7 and the ATI Radeon HD 5570 card. One GPU works fine but I can't seem to get both Radeon GPUs running BOINC jobs. For example, with 2 processors reserved for BOINC I see 3 BOINC Einstein@home tasks running, 2 for the 2 dedicated processors and 0.5 CPU + 1 GPU.

Next, with 3 processors reserved for BOINC and I see 4 BOINC tasks: 3 on the 3 100% dedicated processors and again only one GPU running (plus 0.5 CPU). I'd like to see both GPUs running. I suspect my one GPU limitation is caused by my inability to upgrade the AMD Catalyst beyond version 11.5. I have downloaded and "successfully" installed the new version 12.4 Catalyst several times. But the version number seems stuck at v11.5 and when I check updates it recommends I upgrade to version 12.4. Anyone else stuck on only one GPU or have difficulty upgrading to Catalyst 12.4?

Yeah, my roommate and I have the same problem - after installing 12.4 it still indicates the last install version; 12.3 in my case, 12.1 for my roomie.

Looking online other people have been having the same problem, no solution yet.

I have downloaded and "successfully" installed the new version 12.4 Catalyst several times. But the version number seems stuck at v11.5 and when I check updates it recommends I upgrade to version 12.4. Anyone else stuck on only one GPU or have difficulty upgrading to Catalyst 12.4?

Thanks for the great feedback! I am going to assume my non-updating Catalyst version number (11.5) will be fixed in an update. I just tried setting the GPU utilization factor from 1.0 to 0.5 in the hope of getting 2 simultaneous BRP tasks running on 2 GPUs. No luck, still just one GPU. I increased the number of CPUs from 2 to 3 and instead of another BRP task it took another Gamma Ray Pulsar task. Maybe I should constrain the jobs to be mostly binary radio pulsars.

I do have a 3-day backlog of Einstein@home jobs so maybe I have to work my way through them first before getting dual simultaneous GPUs working on my Radeon HD 5570.

Once new gpu downloads occur, the utilization factor changes.

Once new gpu downloads occur, the utilization factor changes.

It s good idea to make for Ati new type of task 1 cpu+ 1 gpu :)

Minimum Requirements:
[list]
* ATI GPU or APU (OpenCL 1.1 compliant, equivalent to Radeon HD 5xxx)
* 512 MB video memory

Hi there, first time poster long time reader... :)

Using only 80% CPU power is good for me as my rig is OC'd and this helps keep my CPU temps down to a manageable range.

Computer: Win 7Pro X64, i5-2500K CPU @ 3.30GHz (OC 4.5GHz), AMD HD6850, 8GB 1600 RAM, BOINC 7.0.28

P.S. In case anyone is interested in performance numbers:

Run time(sec) 3,739.87
CPU time(sec) 530.31
Claimed credit 6.90
Granted credit 500.00

I have downloaded and "successfully" installed the new version 12.4 Catalyst several times. But the version number seems stuck at v11.5 and when I check updates it recommends I upgrade to version 12.4. Anyone else stuck on only one GPU or have difficulty upgrading to Catalyst 12.4?

i have the same "problem," only its not really a problem for me...i'm able to crunch the new BRP4 ATI tasks on my HD 6950 without issue. i've "successfully" updated to Catalyst 12.4, but my driver version still says 11.5. i'm wondering if the 12.4 Catalyst package is just a bunch of updated run times bundled with the old 11.5 driver files...hence why my driver info shows v11.5, even though the v12.4 download installed without issue for me, and the new E@H ATI GPU app works just fine for me...

I think I see the CPU doing less when it just feeds the GPU. Instead of Task MAnager showing almost 97-100% CPU utilization the core running the BRP4 GPU is at 85-94% utilization. And it's delivering faster results. A BRP4 task that takes 41:51:24 when crunched on one CPU core takes only 4:09:44 when half a CPU core is feeding the video card. It looks as if the GPU speeds number crunching by a factor of 10 even with a modest Radeon HD 5570 card video.

Right now my B0INC Manager shows one half CPU is feeding the one GPU. Other user comments in this thread strongly suggest we expect faster GPU performance when the GPU is tended by one dedicated CPU core. So I changed my GPU utilization factor back to the default (1.0). Once I work through my 3-day backlog of tasks I expect the BOINC server will send me BRP4 tasks that run one GPU tended by one CPU core.

I think my machine is starting to got slow after I updated it.

I am running on an AMD Phenomen II x6 3,20 Ghz and have a HD 5850 plus 8 Gb ram

Win 7 64bit

What is the optimal setup - how much to CPU etc?

Minimum Requirements:
[list]
* ATI GPU or APU (OpenCL 1.1 compliant, equivalent to Radeon HD 5xxx)
* 512 MB video memory

Mo 28 Mai 2012 20:36:27 CEST | | OpenCL: ATI GPU 0: Cedar (driver version CAL 1.4.1720, device version OpenCL 1.2 AMD-APP (923.1), 512MB, 367MB available)

Is this enough video memory or not?

512 is 512. Should work.

Minimum Requirements:
[list]
* ATI GPU or APU (OpenCL 1.1 compliant, equivalent to Radeon HD 5xxx)
* 512 MB video memory

Mo 28 Mai 2012 20:36:27 CEST | | OpenCL: ATI GPU 0: Cedar (driver version CAL 1.4.1720, device version OpenCL 1.2 AMD-APP (923.1), 512MB, 367MB available)

Is this enough video memory or not?

512 is 512. Should work.

After ~2sec I see this error message in the boinc-manager:

"Planungspause: Not enough free CPU/GPU memory available! Delaying next attempt for at least 15 minutes..."

What I can do?
kernel 3.2.10-15-desktop i686 GNU/Linux

Hm - our plan class quotes 360MB GPU memory. You probably have less than that available. Could you disable some 3D features of your desktop (and restart the X11 server, i.e. usually log out & in again)?

Can anything be done to reduce the CPU usage further and increase the GPU usage?
Would the latest APP SDK 2.7 and OpenCL 1.2 help this project in terms of performance?
Thanks.

Question for the people with a low GPU-performance. Do you have "On multiprocessors systems, use at most" set to 100%? Consider lowering it to, possibly, give GPU performance a boost.
____________

Hi guys, its a double precision vga needed?

im owner of 6850 :)

thx

p.

Hello there,

I'm successfully running the BRP tasks on a single 6550 (the integrated GPU in the A8 series of AMD processors). Everything works great, run times per work unit are about 4 hours with CPU usage throttled at 75%. It makes me all giddy just thinking about it.

My issue is that sometimes, not all the time but sometimes, the BRP tasks ignore computing preferences settings in the BIONIC manager. My computer is scheduled to run @home projects from midnight to 8 am, but i if i'm casually using the machine and won't need any real processing power, i'll change my settings to let tasks run while computer is in use. If i do start something that requires processing power and change my settings to keep tasks from running, the BRP tasks carry on like they were British. I can even change the preferences to use 0 cores, and 0 CPU time and the tasks are indefatigable in the face of adversity. The status says it's suspended because the computer is in use, but the cheeky devil just keeps crunching away.

Not really a serious problem for me, and it won't prevent me from using it, but it can be a bit annoying if i actually need my GPU. Any advice?

Best,
Jonathan

Hi guys, its a double precision vga needed?

im owner of 6850 :)

thx

p.

I got first results
- 7.0.28 + latest CATALYSTs

8h15m is my time :-/
- i got 500points

2nd task is in progres // 80% after 6,5h :-(

First numbers looked good, i got 25% in 20min - so i expected that whole task will finish in 80min, which means cca 9000points/24h. But later something went wrong probably and i got these times :(

Is somebody with 6850 out there? - if yes, post the results pls..

i dont know where the problem is.. :-/

Is somebody with 6850 out there? - if yes, post the results pls..

I have this rig: http://einstein.phys.uwm.edu/hosts_user.php?userid=594958
The preferences are use 5 processors, 80% of CPU time, I am new to crunching. I know enough to get myself into trouble and get myself out of it only from time to time ;)

My problem is with the time taken to process GPU tasks. Boinc reports a nominal task time of 2:04:05 for all downloaded GPU tasks. So far the first 2 tasks are at 10:57:00 for 80% completion, using 0.5 CPU & ati GPU (Device(0)and(1)).

Could having the GPU's in crossfire have an effect? If so does reverting back to single GPU's help? I chickened out of reverting in case the tasks became corrupted or such. As to the GPU's - before anyone points out the obvious - they aren't high end, 2 for the price of 1! (no free steak knives though). They work quite well for other things I've used them for! So as to configuring my option or preferences to speed things up...thoughts anyone?

Outside of Boinc the computer is used now mainly for publisher, word docs & spreadsheets - very little load really. My concern is the GPU tasks downloaded to crunch can't be completed by due date at this rate.

Thanks HBE, duly noted and I tweaked the preferences.

I took a gamble re Crossfire. Anyone please feel free to comment...or tell me to politely go soak my head because I "zen"-ed this solution :) So this is how it went:

Checked card & driver set-up. ATI Catalyst is 12.4. FYI, it seems actual driver version is 8.961.0.0. First suspended all work & exited Boinc every time I changed settings - don't want corruptions of data. (seemed like a good idea?)

1) As a 2x 5450 cards in Crossfire, I auto over-clocked via Vision Control Centre from stock 650Mhz core clock /800 Mhz memory to 840Mhz/950Mhz. ATI Work unit times blew out in the end to 36 hrs+ as well as other CPU based work unit times began to climb, with two peaking at 92 & 96 hrs. Huh?

2)Un-linked crossfire. Auto over-clocked Device 0, Device 1 ran at stock speed - was unable to access over-clocking for that card in Vision Control Centre. Bizarrely, ATI Device 1 (stock speed) began to run faster work unit times and began to catch the ATI Device 0 over-clocked card. (scratches head then meditates). However work unit prediction times for pending work dropped considerably to about 12hrs.

3) My Gigabyte MB has an on-board 4290 (forgot about that being there). So into the bios & enable on-board video. Boot up and the 4290 now runs the screen and the two 5450's are inaccessible via Vision Control Centre/AMD Overdrive etc (listed as disabled) but they are not. I'm figuring since Boinc "found" them before these changes at least Boinc knows where to access them for work under this "disabled" arrangement. They are happily crunching away. Work unit times currently down to under 8 hrs and falling. Boinc might lose a bit with memory & CPU power being diverted towards on-board graphics but I can't see a noticebale loading at this time. I now only use the machine for spreadsheets, word & publisher, so graphic performance is no biggie to me.

I don't know why fix attempt #3 appeared to work and #1 was soooo bad but if any tech-head wants free tuition in how to drive a Road-Train in Australian outback, I'll swap lessons for a explanation. Maybe Crossfire warrants a closer look under the hood?

Go Well.

P.S.: In general, I would also recommend not to use the "xxx % of CPU time" throttling setting in the preferences. For multi-core CPUs as yours, it's better to use the "xxx % of cores" setting instead which overall has quite the same effect.

First suspended all work & exited Boinc every time I changed settings - don't want corruptions of data. (seemed like a good idea?)

2)Un-linked crossfire. Auto over-clocked Device 0, Device 1 ran at stock speed - was unable to access over-clocking for that card in Vision Control Centre. Bizarrely, ATI Device 1 (stock speed) began to run faster work unit times and began to catch the ATI Device 0 over-clocked card. (scratches head then meditates). However work unit prediction times for pending work dropped considerably to about 12hrs.

one thing i have come to accept is that you can not over clock a gpu anything like as much for crunching as you can for game`s,
A pixel out of place in a game can be ignored,
A pixel`s worth of data out of place in a work unit is a terminal error
And the auto over clock app`s that come with a graphics card do not alow for this by a long way.
Does anyone disagree ?

..and in most cases BOINC only reports the GPU type (e.g. "Cedar", "Cypress", etc) and not the card model itself.

As far as I understand from Charlie, the OpenCL (RAM) detection under OS X is broken, so instead of using any OpenCL library, they use the OpenGL library and read the RAM size and model name from that.
____________
Jord

-The BOINC FAQ Service
- BOINC 7.0 FAQ

I used to be an adventurer like you. Then I took an arrow in the knee...

My BRP4 tasks take 4 hours 23 minutes when using 0.5 CPU + 1.0 ATI GPU (Radeon HD 5570), compared with almost 40 hours when run on one core of my i5 quad core. I rebooted an hour ago and the latest BOINC Manager estimate for BRP4 tasks is 3 hours 23 minutes. These are welcome improvements - but still a factor of 3 or 4 less than the one hour BRP4 tasks reported earlier in this thread. I would like to speed things up if I can.

My computing preferences are 75% of the processors and 100% of CPU time.

Maybe unused processes are slowing things down on the x64 Win7 machine with 8GB ram. Skype is there but idle, Windows TaskManager runs in background. Plus several gadgets: CPU usage, network usage, drive usage, GPU usage, BOINC monitor, weather. I am managing the computer remotely using TeamViewer and monitoring cpu temperatures with CoreTemp 0.99.7, they are running 60-68C. The antivirus / firewall is Comodo.

Is there a point to running only selected applications - limiting tasks for example to BRP4? I now accept all Einstein@home tasks and typically I have 4 jobs running in parallel: x1 BRP4 job, and three others: a mix of Gravitational Wave S6 and Gamma Ray pulsar jobs. Sometimes I see 2 BRP4 jobs at once, one 4 hour job in the ATI GPU and one 40 hour job in the CPU. I expected to see both ATI GPUs running BRP4s.

While this may be the PC's best practical performance or a limitation of the Radeon HD 5570 I really appreciate any suggestions.

My BRP4 tasks take 4 hours 23 minutes when using 0.5 CPU + 1.0 ATI GPU (Radeon HD 5570), compared with almost 40 hours when run on one core of my i5 quad core. I rebooted an hour ago and the latest BOINC Manager estimate for BRP4 tasks is 3 hours 23 minutes. These are welcome improvements - but still a factor of 3 or 4 less than the one hour BRP4 tasks reported earlier in this thread. I would like to speed things up if I can.

Is there a point to running only selected applications - limiting tasks for example to BRP4? I now accept all Einstein@home tasks and typically I have 4 jobs running in parallel: x1 BRP4 job, and three others: a mix of Gravitational Wave S6 and Gamma Ray pulsar jobs. Sometimes I see 2 BRP4 jobs at once, one 4 hour job in the ATI GPU and one 40 hour job in the CPU. I expected to see both ATI GPUs running BRP4s.

Thanks again Sunny129, your HD 6950 has a factor of 4 floating point advantage over my HD 5570:
HD 6950 2253 GFLOPS
HD 5570 520 GFLOPS
http://en.wikipedia.org/wiki/Comparison_of_AMD_graphics_processing_units

...so my 4 hour time for a GPU BRP4 seems in line with your 40-64 minutes.

I can't explain where 3 hours running times came from after reboot instead of the 4 hours 20 minutes or so stable estimates from the last few days. The timing continues to change slowly after reboot, now a few hours later the queued BRP4 tasks are estimated at 3:46:38 instead of closer to 3 hours. Meanwhile the BRP4 task currently running in the GPU is estimated to take 4 hours 13 minutes. Things seem to be running OK.


Today I tried setting up for all cores: BOINC uses 100% of the CPU/all 4 cores with 100% duty cycle. Result: I still get only one GPU + 0.5 CPU job plus 4 other (non-GPU) Einstein@home jobs. I know it adds up to 4.5 cores on my 4-core i5. While the arithmetic seems off the multi-tasking works fine and work is happening much faster, overall almost twice as fast considering RAC. I reset back to 3 dedicated cores thinking that 5 simultaneous jobs would cause too enough resource conflicts to marginally reduce output.


I took your suggestion to uncheck the "Run CPU versions of applications for which GPU versions are available" box, maybe performance will improve. In any case the 10:1 advantage using just one of my 2 GPUs is pretty satisfying, kudos to the BOINC team programmers. And thanks to you for answering my questions, I am quite happy with my performance improvement and now I am thinking about checking Craigslist for good used ATI graphics cards. Maybe an HD 6950 or two.

...where "n" is the GPU utilization factor. when set to a value of 1, the BRP4 GPU apps would run only 1 GPU task at a time. when set to 0.5, the apps would run 2 GPU tasks simultaneously. when set to 0.33, the apps would run 3 GPU tasks simultaneously, and so on and so forth...

Thank you for a good explanation. Could u please speciafy a little:

I have a factor of 0,5 by default for Radeon 5750. Is this good or bad? I didn't change anything.

hmm...the default GPU utilization factor for all of my various machines under my Einstein@Home web preferences was 1, so i kind of assumed that everyone's GPUs would default to running only 1 GPU task at a time. either way, you should be fine with a factor of 0.5 (running 2 GPU tasks simultaneously) provided your HD 5750 is a 1GB model. you see, they also made some HD 5750 with only 512MB of VRAM onboard. considering each BRP4 ATI task consumes approx. 355MB of VRAM, an HD 5750 w/ only 512MB of VRAM would only be able to handle 1 task at a time efficiently. if you tried to run 2 tasks simultaneously on a 512MB card, you'd be over-utilizing the VRAM. while i'm fairly confident that it would work and not cause compute errors, the VRAM bottlneck would probably significantly increase your GPU task run times and take a serious toll on your GPU's compute efficiency. long story short, if you have a 512MB HD 5750, change your GPU utilization factor to 1 (and run only 1 GPU tasks at a time). if you have a 1GB HD 5750, leave your GPU utilization factor at 0.5 (and run 2 GPU tasks simultaneously).

@sunny129

forgive me, for i was too lazy to go back through the entire thread looking for those documented 60-minute run times and the types of GPUs they ran on. but if i had to guess, i'd say your GPU has reached its limits.

i'm actually quite puzzled as to how a restart shaved an hour off your run times...

My 6850 crunches in 65 minutes but its a lot more powerful than a 5570 with more than double the shaders - http://www.gpureview.com/show_cards.php?card1=624&card2=636

Hello! Thank you for explanation. I have 1Gb model/Core i7-2600 (8 virtual cores), ratio 0,5. But nevertheless I have only 9 tasks running at a time. Why so? I thought there will be 10 tasks (8 cores and 2 GPUs). 2 of the tasks are marked "0,5 CPU + 0,5 GPU" - what's that supposed to mean?

The HD6850 has 12 SIMDs or Compute Units.
The HD5570 has 5 SIMDs or Compute Units.

This way we can calculate the strength of the GPU; we do that by dividing the amount of stream processors by the amount of compute units.
HD6850 has 12 Compute Units. 960 stream shaders / 12 Compute Units = 80 OpenCL Elements.
HD5570 has 5 Compute Units. 400 stream shaders / 5 Compute Units = 80 OpenCL Elements.

Which means that in pure OpenCL speed, the HD6850 and HD5570 are equal.

...but shouldn't the HD 5570 have similar run times too then?

...but shouldn't the HD 5570 have similar run times too then?

When you put it in your system, yes. ;-)

But now you have to put into the equation:
- How much memory does the card have?
- What memory speed does the GPU have?
- What core speed does the GPU have?
- What kind of PCIe slot is it in?
- What kind of CPU is next to it?
- What else is this CPU doing?
- Did the user allow for the extra core for the GPU only?
- Is anything (CPU or GPU) overclocked and by how much?
- Which drivers are we using?

etc.

Today I tried setting up for all cores: BOINC uses 100% of the CPU/all 4 cores with 100% duty cycle. Result: I still get only one GPU + 0.5 CPU job plus 4 other (non-GPU) Einstein@home jobs.

My 6850 crunches in 65 minutes but its a lot more powerful than a 5570 with more than double the shaders - http://www.gpureview.com/show_cards.php?card1=624&card2=636

Comparing shaders between models isn't useful when it comes to OpenCL, or even CAL. That's only useful when you compare between GPUs of the same make and generation. Only then is more better.

What OpenCL uses is Compute Units and OpenCL Processing Elements.
See Geeks3D for a complete write-up and while you're there, get CAPS GPU Viewer. That'll show you the 'power' of your GPU.

From the introduction to OpenCL (PDF) we know that when you know the amount of SIMD processors on the GPU, that you know the amount of Compute Units, as these are the same.

...but shouldn't the HD 5570 have similar run times too then?

When you put it in your system, yes. ;-)

But now you have to put into the equation:
- How much memory does the card have?
- What memory speed does the GPU have?
- What core speed does the GPU have?
- What kind of PCIe slot is it in?
- What kind of CPU is next to it?
- What else is this CPU doing?
- Did the user allow for the extra core for the GPU only?
- Is anything (CPU or GPU) overclocked and by how much?
- Which drivers are we using?

etc.

true...i didn't think about all that LOL.

I direct you over to the PCIe 2.0 vs 3.0 thread in Cruncher's Corner.

In short, it makes a HUGE difference. These tasks rely heavily on CPU, when running 3 tasks, one can near max PCIe 3.0 bandwidth.

4 would max it for sure. This is why I stated in the 690 thread that a 690 would do more harm than good. Would be able to run 6 tasks at a time, but would be limited greatly by bandwidth.

Having a good OC processor makes a big difference for this project.

BOINC: 7.0.28
ATI: 12.3 (apparently 12.4 might have some issues with BOINC) - APP SDK installed.
GFX Card: Radeon 5670, 1GB onboard RAM

Collatz Conjecture works fine on this and uses 100% of my GPU.

However when I suspend that, the GPU goes to E@H (good) which is only using between 25-35% of the GPU (bad). The command that initialised it:


06/06/2012 20:51:18 | Einstein@Home | Starting task p2030.20100903.G43.15+00.93.C.b5s0g0.00000_504_1 using einsteinbinary_BRP4 version 124 (atiOpenCL) in slot 14

Extrapolating from time taken and % complete, I'm estimating about 10.5hrs to complete this task (don't know if that's fast or slow, but it could be 4 times faster if it used all my GPU).


Any thoughts?

I do not have any experience yet with AMD OpenCL but with the CUDA app, the most optimal setup for HT enabled CPUs is to set the maximum CPU usage to 50% as to avoid HT performance penalties. Being that the OpenCL app is also CPU dependent and if your CPU has HT, then setting cores to 50% could be beneficial. This change may also improve performance for Bulldozer processors that have a shared FPU for every two integer cores but I have not tested that myself.

Thanks for the responses folks.
CPU is a Phenon II, X4 (no hyperthreading)

Ah. Setting max CPU CORE usage to 75% (I have four cores) gets GPU usage to 80-90%. Setting back to 0% (i.e., no value) results in GPU usage plummeting again.
Setting CPU usage to to 50% (not core usage, but CPU usage) results in about 60% of my CPU being used.
Even suspending CPU usage entirely (with the suspend option) results in only about 80-90% of the GPU being used by E@H.

The problem with these settings is that I then lose usage of some of my CPU.


What Collatz does is that it gives the CPU thread high priority for the GPU task to ensure it will always be fed. It also seems to only allocate 0.01% of a CPU. In reality it is using about 0.2% of my CPU, or <1% of a core.

E@H claims to use 0.5 cores on BOINC, apparently set to "below normal" priority (which is higher than the "idle" the other threads are using). In reality its using 6% of my CPU time or 25% of one core.

You may wish to consult with the collatz folks to discover how they get it to use 100% of the GPU without needing to "disable" BOINC's use of a core.

Hello! Thank you for explanation. I have 1Gb model/Core i7-2600 (8 virtual cores), ratio 0,5. But nevertheless I have only 9 tasks running at a time. Why so? I thought there will be 10 tasks (8 cores and 2 GPUs). 2 of the tasks are marked "0,5 CPU + 0,5 GPU" - what's that supposed to mean?

that actually makes perfect sense. think about it - the 7 CPU tasks that are running require one core each, leaving only one core free. the 2 GPU tasks that are running require only 0.5 CPU's each (for a total of 1 whole core). so 7 CPU tasks consume 7 CPU cores, while 2 GPU tasks consume the 8th and final CPU core, for a total of 9 tasks running ay any one time...

But I still don't understand how the number of CPU cores occupied by moderating GPU tasks depends on quantity of this GPU tasks run simulateonously. As far as I understood no matter how many tasks u run on GPU simulationously - only one CPU core will be busy moderating GPU tasks. Am i right? So if I run only one task at a time on GPU one of CPU cores will still be occupied by the GPU task, correct?

As far as I understood no matter how many tasks u run on GPU simulationously - only one CPU core will be busy moderating GPU tasks. Am i right? So if I run only one task at a time on GPU one of CPU cores will still be occupied by the GPU task, correct?

To further on Sunny's point.

With my 3.2Ghz CPU, E@H GPU uses about 1/4 of a core. That means it costs me one core to run 4 E@H GPU tasks.

Collatz uses <1% of a core, so can theoretically run >100 GPU tasks per core.

I have no interest in the "points" and believe E@H to be more a worthwhile project (physics > maths ;-) ), but for simple optimal use of my hardware, I'll go with Collatz until E@H GPU becomes more CPU friendly.


Edit: Rather than Task Manager, on windows use Process Explorer - http://technet.microsoft.com/en-gb/sysinternals/bb896653 - now by microsoft and it even tells you how much of your GPU a specific task is using so you don't need to stop other tasks.

To further on Sunny's point.

With my 3.2Ghz CPU, E@H GPU uses about 1/4 of a core. That means it costs me one core to run 4 E@H GPU tasks.

Collatz uses <1% of a core, so can theoretically run >100 GPU tasks per core.

I have no interest in the "points" and believe E@H to be more a worthwhile project (physics > maths ;-) ), but for simple optimal use of my hardware, I'll go with Collatz until E@H GPU becomes more CPU friendly.


Edit: Rather than Task Manager, on windows use Process Explorer - http://technet.microsoft.com/en-gb/sysinternals/bb896653 - now by microsoft and it even tells you how much of your GPU a specific task is using so you don't need to stop other tasks.

No double precision here I'm afraid (already investigated MW@H). I'm stuck with Collatz for now. But at least its doing /something/.

It's nice work on notebook this AMD llano 3610mx + seymour ati gpu.
4CPU+2GPU process run. Good work/
Sorry 4 my bad eng.

If you are experiencing the below error on Ubuntu 12.04 64bit with an AMD / ATI graphics card. I have found that the fix is rather simple. In '/etc/OpenCL/vendors/' create a file called 'amdocl32.icd', in that file enter the following 'libamdocl32.so' and save. The next time the client is started E@H can use the card for processing.

<core_client_version>7.0.28</core_client_version>
<![CDATA[
<message>
process exited with code 255 (0xff, -1)
</message>
<stderr_txt>
[16:06:07][3184][INFO ] Application startup - thank you for supporting Einstein@Home!
[16:06:07][3184][INFO ] Starting data processing...
[16:06:07][3184][ERROR] Failed to get OpenCL platform/device info from BOINC (error: -1)!
[16:06:07][3184][ERROR] Demodulation failed (error: -1)!
16:06:07 (3184): called boinc_finish

</stderr_txt>
]]>

I tried to run E@H on my new secondary rig (http://einstein.phys.uwm.edu/show_host_detail.php?hostid=5448695).

It always posts an error log message when it tries to get work for GPU. I have disabled CPU usage. I checked the log: http://einstein.phys.uwm.edu//host_sched_logs/5448/5448695

I have the latest Catalyst driver installed (12.4) and tried it. It failed, I read that the Catalyst APP SDK could be the reason and uninstalled it. Still, I get the error message you can see above.

What can I do to get it working?

Thanks in advance.
____________

I've run this project before but now I have issues. After 3 failed tasks I updated the driver to the latest version and detached/reattached to the project.
Here's the error message from the failed tasks.

application version Binary Radio Pulsar Search (Arecibo) v1.24 (atiOpenCL)
Stderr output

<core_client_version>7.0.28</core_client_version>
<![CDATA[
<message>
There was an error while deleting the color transform. (0x7e3) - exit code 2019 (0x7e3)
</message>
<stderr_txt>
Activated exception handling...
[12:25:14][3364][INFO ] Starting data processing...
[12:25:18][3364][INFO ] Using OpenCL platform provided by: Advanced Micro Devices, Inc.
[12:25:18][3364][INFO ] Using OpenCL device "Cypress" by: Advanced Micro Devices, Inc.
[12:25:19][3364][INFO ] Checkpoint file unavailable: status.cpt (No such file or directory).
------> Starting from scratch...
[12:25:19][3364][INFO ] Header contents:
------> Original WAPP file: ./b2030.20111203.G44.88-01.88.N.b3s0g0.00000_DM27.20
------> Sample time in microseconds: 65.4762
------> Observation time in seconds: 274.62705
------> Time stamp (MJD): 55898.792610759854
------> Number of samples/record: 0
------> Center freq in MHz: 1214.289551
------> Channel band in MHz: 0.33605957
------> Number of channels/record: 960
------> Nifs: 1
------> RA (J2000): 191958.885399
------> DEC (J2000): 94335.9505005
------> Galactic l: 0
------> Galactic b: 0
------> Name: G44.88-01.88.N
------> Lagformat: 0
------> Sum: 1
------> Level: 3
------> AZ at start: 0
------> ZA at start: 0
------> AST at start: 0
------> LST at start: 0
------> Project ID: --
------> Observers: --
------> File size (bytes): 0
------> Data size (bytes): 0
------> Number of samples: 4194304
------> Trial dispersion measure: 27.2 cm^-3 pc
------> Scale factor: 0.00661517
[12:25:24][3364][INFO ] Seed for random number generator is 1149519249.
[12:25:36][3364][INFO ] Derived global search parameters:
------> f_A probability = 0.08
------> single bin prob(P_noise > P_thr) = 1.32531e-008
------> thr1 = 18.139
------> thr2 = 21.241
------> thr4 = 26.2686
------> thr8 = 34.6478
------> thr16 = 48.9581
[12:25:42][3364][ERROR] Error during OpenCL kernel setup: HSFFB (error: -5)
[12:25:42][3364][ERROR] Demodulation failed (error: 2019)!
12:25:42 (3364): called boinc_finish

</stderr_txt>
]]>

Now that I've reattached all i get is.
6/16/2012 1:01:00 PM | Einstein@Home | update requested by user
6/16/2012 1:01:02 PM | Einstein@Home | Sending scheduler request: Requested by user.
6/16/2012 1:01:02 PM | Einstein@Home | Not reporting or requesting tasks
6/16/2012 1:01:04 PM | Einstein@Home | Scheduler request completed