Posts by Harri Liljeroos


log in
1) Message boards : Number crunching : Open beta for OpenCL AstroPulse for Intel GPUs started (Message 1380218)
Posted 446 days ago by Harri Liljeroos
I've managed to get first one crunched. Here's the stderr output:

<core_client_version>7.0.64</core_client_version>
<![CDATA[
<stderr_txt>
Priority of worker thread raised successfully
Priority of process adjusted successfully, below normal priority class used
GPU type not found in init_data.xml
OpenCL platform detected: NVIDIA Corporation
OpenCL platform detected: Intel(R) Corporation
WARNING: BOINC supplied wrong platform!
BOINC assigns device 0
WARNING: BOINC failed to provide OpenCL device, using own enumeration abilities
Used GPU device parameters are:
Number of compute units: 16
Single buffer allocation size: 256MB
max WG size: 512

Build features: Non-graphics OpenCL USE_OPENCL_INTEL OCL_ZERO_COPY COMBINED_DECHIRP_KERNEL FFTW USE_INCREASED_PRECISION USE_SSE2 x86
CPUID: Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz

Cache: L1=64K L2=256K

CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3
AstroPulse v.6
Non-graphics FFTW USE_CONVERSION_OPT
Windows x86 rev 1764, V6 match, by Raistmer with support of Lunatics.kwsn.net team. SSE2

OpenCL version by Raistmer

oclFFT fix for ATI GPUs by Urs Echternacht
ffa threshold mods by Joe Segur
SSE3 dechirping by JDWhale
Combined dechirp kernel by Frizz
Number of OpenCL platforms: 2


OpenCL Platform Name: NVIDIA CUDA
Number of devices: 1
Max compute units: 4
Max work group size: 1024
Max clock frequency: 928Mhz
Max memory allocation: 268353536
Cache type: Read/Write
Cache line size: 128
Cache size: 65536
Global memory size: 1073414144
Constant buffer size: 65536
Max number of constant args: 9
Local memory type: Scratchpad
Local memory size: 49152
Queue properties:
Out-of-Order: Yes
Name: GeForce GTX 650 Ti
Vendor: NVIDIA Corporation
Driver version: 314.22
Version: OpenCL 1.1 CUDA
Extensions: cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_d3d9_sharing cl_nv_d3d10_sharing cl_khr_d3d10_sharing cl_nv_d3d11_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll 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_fp64


OpenCL Platform Name: Intel(R) OpenCL
Number of devices: 1
Max compute units: 16
Max work group size: 512
Max clock frequency: 350Mhz
Max memory allocation: 325058560
Cache type: Read/Write
Cache line size: 64
Cache size: 2097152
Global memory size: 1300234240
Constant buffer size: 65536
Max number of constant args: 8
Local memory type: Scratchpad
Local memory size: 65536
Queue properties:
Out-of-Order: No
Name: Intel(R) HD Graphics 4000
Vendor: Intel(R) Corporation
Driver version: 9.18.10.3071
Version: OpenCL 1.2
Extensions: cl_khr_icd 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_gl_sharing cl_khr_d3d10_sharing cl_intel_dx9_media_sharing cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_event cl_khr_gl_msaa_sharing cl_khr_depth_images cl_khr_gl_depth_images cl_khr_dx9_media_sharing cl_khr_d3d11_sharing cl_khr_image2d_from_buffer


state.fold_buf_size_short=65536; state.fold_buf_size_long=262144
INFO: can't open binary kernel file: D:\ProgramData\BOINC/projects/setiathome.berkeley.edu\AstroPulse_Kernels_r1764.cl_IntelRHDGraphics4000.bin_V6, continue with recompile...
INFO: binary kernel file created
WARNING: can't open binary kernel file for oclFFT plan: D:\ProgramData\BOINC/projects/setiathome.berkeley.edu\AP_clFFTplan_IntelRHDGraphics4000_32768_r1764.bin, continue with recompile...

single pulses: 2
repetitive pulses: 1
percent blanked: 0.00
Single pulse: peak_power=88.65; dm=-2141; scale=5
Single pulse: peak_power=131.4; dm=14334; scale=6
Rep. pulse: peak_power=2566; dm=14928; fft_num=0; scale=4896; ffa_scale=4; period=-1.0722e-086

class T_remove_radar: total=3.74e+009, N=1, <>=3.74e+009, min=3.74e+009, max=3.74e+009
class T_main_loop_L1: total=5.78e+013, N=111, <>=5.20e+011, min=4.58e+011, max=6.07e+011
class T_FFT_forward: total=1.34e+011, N=909312, <>=1.47e+005, min=7.25e+004, max=6.80e+008
class T_remove_radar_randomize: total=5.41e+009, N=1817736, <>=2.98e+003, min=3.26e+002, max=3.19e+007
class T_build_chirp_table: total=0.00e+000, N=0, <>=0.00e+000, min=1.84e+019, max=0.00e+000
class T_DataWrite: total=0.00e+000, N=0, <>=0.00e+000, min=1.84e+019, max=0.00e+000
class T_DataWrite_ns: total=0, N=0, <>=0, min=0 max=0
class T_oclReadBuf: total=1.02e+008, N=909312, <>=1.12e+002, min=1.80e+001, max=1.45e+007
class T_ChirpWrite: total=0.00e+000, N=0, <>=0.00e+000, min=1.84e+019, max=0.00e+000
class T_ChirpWrite_ns: total=0, N=0, <>=0, min=0 max=0
class T_dechirp: total=5.34e+011, N=909312, <>=5.88e+005, min=2.58e+005, max=3.03e+008
class Dechirp_ns: total=0, N=0, <>=0, min=0 max=0
class Half_ns: total=0, N=0, <>=0, min=0 max=0
class T_PC_single_pulse_kernel_FFA_update: total=4.27e+013, N=909312, <>=4.70e+007, min=4.87e+006, max=6.99e+009
class PC_ns: total=0, N=0, <>=0, min=0 max=0
class T_oclReadBuf: total=1.02e+008, N=909312, <>=1.12e+002, min=1.80e+001, max=1.45e+007
class T_oclWriteBuf: total=0.00e+000, N=0, <>=0.00e+000, min=1.84e+019, max=0.00e+000
class T_FFT_inverse: total=3.63e+011, N=909312, <>=3.99e+005, min=1.61e+005, max=1.89e+008
class T_ffa: total=1.41e+013, N=1998, <>=7.08e+009, min=2.41e+009, max=5.24e+010
class T_GPU_buffer_read_backs: total=4, N=4, <>=1, min=1 max=1
OCL_ZERO_COPY USE_OPENCL USE_OPENCL_INTEL OPENCL_WRITE USE_INCREASED_PRECISION SMALL_CHIRP_TABLE COMBINED_DECHIRP_KERNEL
rev 1764
05:58:09 (896): called boinc_finish

</stderr_txt>
]]>


Run time 17,018.30; CPU time 13,856.16. It ran during the night so I wasn't monitoring the system. Anyway it was running 8 CPU tasks and 2 CUDA tasks at the same time, so no free CPU was available.

Now waiting for wingman to finish his.
2) Message boards : Number crunching : Open beta for OpenCL AstroPulse for Intel GPUs started (Message 1371703)
Posted 464 days ago by Harri Liljeroos
I think all this part in quota should be removed


Thanks, I may try that later. I'll wait and see if the scheduler fix has any affect for this. As we are not splitting any AstroPulse at the moment, it is hard to tell.
3) Message boards : Number crunching : Open beta for OpenCL AstroPulse for Intel GPUs started (Message 1371243)
Posted 465 days ago by Harri Liljeroos
try to decieve BOINC telling it that Intel GPU AP is CPU app.
SETI BOINC server has no intel GPU app so we need same trick as when there was no ATi support from BOINC..


How would I do that? By removing this section from the app_info.xml?

<avg_ncpus>0.04</avg_ncpus>
<max_ncpus>0.2</max_ncpus>
<plan_class>opencl_intel_100</plan_class>
<cmdline></cmdline>
<coproc>
<type>intel_gpu</type>
<count>1</count>
</coproc>


Or just by removing the <plan_class> section?
4) Message boards : Number crunching : Open beta for OpenCL AstroPulse for Intel GPUs started (Message 1371016)
Posted 465 days ago by Harri Liljeroos
I briefly looked at SongBird's app_info and noticed that you don't have AstroPulse application for your Cuda card. Your app_info seems to be similar than mine for the AstroPulse Intel OpenCL (mine is from aistub file) except for the last <app_version> section. I did add that part to my app_info.xml to see if that makes any difference. Well, it didn't. Unfortunately I can not get any AstroPulse WUs for the Intel HDGraphics. Maybe this is because I have the Cuda application for AstroPulse running as well? This all I get as a response for scheduler request:

SETI@home 23.5.2013 20:17:22 Sending scheduler request: To fetch work.
SETI@home 23.5.2013 20:17:22 Requesting new tasks for intel_gpu
SETI@home 23.5.2013 20:17:26 Scheduler request completed: got 0 new tasks
SETI@home 23.5.2013 20:17:26 No tasks sent
SETI@home 23.5.2013 20:17:26 No tasks are available for SETI@home Enhanced
SETI@home 23.5.2013 20:17:26 No tasks are available for Astropulse v505
SETI@home 23.5.2013 20:17:26 No tasks are available for AstroPulse v6
SETI@home 23.5.2013 20:17:26 Tasks for AMD/ATI GPU are available, but your preferences are set to not accept them
SETI@home 23.5.2013 20:17:26 Project has no tasks available

5) Questions and Answers : GPU applications : OpenCL on Intel HD Graphics 4000 not recognized (Message 1369915)
Posted 469 days ago by Harri Liljeroos
... but when I connected the monitor to it, it started working...

So Intel drivers have the same problem as Nvidia a few years ago (drivers at that time did not detect computing capabilities when no monitor is connected. So people invented 'dummy plug')


I installed the Intel application and Boinc is requesting work for it from Seti but haven't received any. Reply is "no tasks available", only for ATI/AMD GPU.

Ask this in "Open beta for OpenCL AstroPulse for Intel GPUs started" so people which use this app can give you advice what to do:
http://setiathome.berkeley.edu/forum_thread.php?id=70717

Astropulse tasks are not always available but since you have no problems to get them for CPU and NVIDIA GPU (and all the tasks are the same):
http://setiathome.berkeley.edu/results.php?hostid=6969332&offset=0&show_names=0&state=0&appid=12

... probably there is something 'special' to do (OpenCL apps for Intel GPUs is 'new uncharted territory' for BOINC)



Will post to the thread, others have had similar problems. Thanks for your advice.
6) Message boards : Number crunching : Open beta for OpenCL AstroPulse for Intel GPUs started (Message 1369914)
Posted 469 days ago by Harri Liljeroos
Hi,
I have set up my Intel HDGraphics 4000 for OpenCL and installed Raistmer's application. I noticed that I needed to plug in my monitor to that card to make Boinc or GPU-Z to recognize its OpenCL capability, installing the driver and SDK were not enough. Normally the monitor is hooked to Nvidia GTX 650 Ti.

Now Boinc is requesting work for intel_gpu but hasn't received any, the response is "no work avaliable for Seti Enhanced or AstroPulse 5.05 or v6". For CPU and CUDA I do get AstroPulse WUs and they run OK. This is in Windows 7 64 bit with Boinc 7.0.64.

Here's the host http://setiathome.berkeley.edu/show_host_detail.php?hostid=6969332
7) Questions and Answers : GPU applications : OpenCL on Intel HD Graphics 4000 not recognized (Message 1368905)
Posted 472 days ago by Harri Liljeroos
The 3D Caps Viewer didn't first find the Intel Graphics at all, but when I connected the monitor to it, it started working. Now also GPU-Z and Boinc says Intel has OpenCL. GPU-Z doesn't find Cuda on my Nvidia card anymore but Boinc sees it and crunches away with Nvidia.

I installed the Intel application and Boinc is requesting work for it from Seti but haven't received any. Reply is "no tasks available", only for ATI/AMD GPU.

Thank you for your help.
8) Questions and Answers : GPU applications : OpenCL on Intel HD Graphics 4000 not recognized (Message 1368798)
Posted 472 days ago by Harri Liljeroos
Hi,
I am trying to get OpenCL to work on my Intel GPU. The CPU is I7-3770 on a WIn 7 Enterprise 64 bit. So far I have done the following:
- enabled the internal graphics in the BIOS
- allocated to it 512 MB memory in the BIOS
- installed the latest driver for it (9.18.10.3071)
- installed the Intel® SDK for OpenCL* Applications 2013 (64 bit)

The computer also has a Nvidia GTX 650 Ti where the monitor is attached to. It runs the Lunatics CUDA application for MB and the OpenCL application for AP. GPU-Z sees both graphics cards but indicates for the Intel HD Graphics 4000 that no OpenCL is available (shows 0 MB memory for it though). Windows device manager indicates Intel HD Graphics working properly. Boinc is 64 bit version of 7.0.64 and does not find the Intel card.

I have studied the Intel SDK installation notes and checked:
- INTELOCLSDKROOT environment variable points to correct folder
- registry entries mentioned in installation notes are there with correct values

I have Raistmers AP application downloaded but not yet installed. Propably no point while OpenCL support is missing.

Can somebody help, what I am doing wrong or is there still something missing?
9) Message boards : Number crunching : BOINCstats partial graph display (Message 1308043)
Posted 650 days ago by Harri Liljeroos
I've experienced similar behaviour on my old Compaq/HP 8710w laptop with Win XP. It has a Nvidia Quadro FX 3600M GPU. Some of the graphs on BOINCstats would go black randomly even while scrolling up and down my stats page with FireFox. The symptoms got worse the longer the computer and FireFox were on. Reloading the page with CTRL+F5 might sometimes help or just blacken some other graph on the page. Restarting the computer usually helped for a while but eventually the problem reappeared. Adblock Plus is in use on that machine. IE did not have the that problem and I never used Chrome.

I tried different driver versions but that did not help. I did use that GPU for crunching for a while, but the high temperature did kill the original GPU, so now it crunches only with the CPU. I am currently using the driver version 176.87, which is actually the latest from HP for that machine, but newer drivers directly from Nvidia did not cure the problem.

I just viewed my own stats with that computer using FireFox and the graphs are showing correctly. This might have been the first I've looked at them with this computer since the layout was changed at BOINCstats. The computer has been on for about a week but it hasn't been used for anything else than crunching and I could not reproduce the problem. I'll leave the FireFox on and try again later to see if the problem still persists.
10) Message boards : Number crunching : Upload server - disabled ? (Message 1289963)
Posted 700 days ago by Harri Liljeroos
Something certainly "wasn't right" this time round. despite everyone trying to get uploads (and for that matter downloads) the thin blue line fell to the bottom and stuck there for most of Saturday and Sunday (UK times). When it picked up there was quite a surge in it, now settling down to its usual level - I reckon there was a visit from a tyre kicker sometime Sunday evening (UK time) along with the end of the Astropulse production.


If you look closely to the graph, you'll see that it has automatic scaling and the bottom is now 10 Mbit/s. So it was still receiving quite a lot of data even in worst situation.
11) Message boards : Technical News : Get Out of My House (Jan 18 2011) (Message 1068183)
Posted 1321 days ago by Harri Liljeroos
Hey Matt, any hints on how you are picking your thread titles these days?


The titles seem to be names of Kate Bush songs.
12) Message boards : Number crunching : Now disabled @ SETI@home: VLAR WUs not to GPUs (Message 1020162)
Posted 1496 days ago by Harri Liljeroos
@Sutaru

Thanks for pointing that extra "/" out, cut and paste fault. Unfortunately can't edit it anymore.

@Miep

Actually mine is the V12 and my wingman has the 6.09. And he got it first.
13) Message boards : Number crunching : Now disabled @ SETI@home: VLAR WUs not to GPUs (Message 1020094)
Posted 1497 days ago by Harri Liljeroos
I've got four of these too on my host. All .vlar_0 and .vlar_1. And I also have app_info for 6.08 and cuda23.

This one 01jn10ab.19594.7429.16.10.176.vlar was sent also to a standard 6.09 application.
14) Questions and Answers : GPU applications : Where to download standard cuda application and modify app_info (Message 969421)
Posted 1665 days ago by Harri Liljeroos
Thank you for your reply Sutaru,
I solved the issue by downloading the Lunatics Unified Installer 0.2 and took the Cuda 2.3 dll's from Einstein. The temperature is now controlled by the TThrottle and I set the max GPU temp to 81°C. For Einstein the throttling is activating only sometimes but for Seti it is almost constantly active and the application is throttled to 25...35 % to keep the temperature at 81. This of course slows down the calculation. I am wondering if it would make any difference if I'd use the 2.2 dll's. Probably there would be less throttling but calculation would take the same amount of time.

It is difficult to compare how the computer is now doing as there aren't many this kind of GPU's around (CPU T9500, GPU Quadro FX 3600M: peak GFLOPS 160)

I don't have a cooler for the laptop, but it is raised about 1 cm above the desk to let the air flow freely.
15) Questions and Answers : GPU applications : Where to download standard cuda application and modify app_info (Message 967695)
Posted 1672 days ago by Harri Liljeroos
Thanks for the info, I'll check that out.
16) Questions and Answers : GPU applications : Where to download standard cuda application and modify app_info (Message 967686)
Posted 1672 days ago by Harri Liljeroos
That's what I'll do, but first I will drain the cache (1+1 days) by setting NNT and let it finish all WU's. Maybe the temps would stay down if I could somehow get hold on the older cuda 2.2 version of the science application (6.08?).
17) Questions and Answers : GPU applications : Where to download standard cuda application and modify app_info (Message 967670)
Posted 1672 days ago by Harri Liljeroos
Hi,
I am trying to set up my laptop to use optimized CPU applications and use at the same time standard cuda application. The reason for this is that the laptop's gpu does not have a fan of it's own and is running too hot with optimized cupa application.

When using only Lunatics optimized CPU applications the GPU temps get around 75 degrees celsius. With optimized GPU application the temps get around 87 degrees. I am now using the GPU application for Einstein which does not use much of the GPU and the GPU temp is then about 81 degrees.

So I was wondering if using only the standard SETI cuda application I could keep the temp down and still use the GPU. So I'd like to know where to find the standard gpu applications and if someone had a ready formulated section for the app_info.xml using that application. I deleted from my current app_info the section for gpu to not use the optimized application.
18) Message boards : Number crunching : graphic card (Message 963453)
Posted 1690 days ago by Harri Liljeroos
Dang it...

I just got asked the question, and I can't remember the answer, or if there is one...

Is it possible to verify the Cuda dll version that you are currently running?


Using Process Explorer from www.sysinternals.com you can see what dll's are being used by different programs. You need to set the GPU processing always active so that you can select the running exe without the program exiting. Or at least you have to very quick. Maybe you can make the Process Explorer to write things to a log file, I haven't studied it so thorouhly.
19) Message boards : Number crunching : No CPU work, only CUDA, won't even request (Message 925779)
Posted 1845 days ago by Harri Liljeroos
Does the backoff work only halfway if it was only affecting the CPU requests and not the GPU requests? If a request to server is made anyway shouldn't it then ask work for both GPU and CPU at the same time if host is lacking work for CPU and GPU?
20) Message boards : Number crunching : No CPU work, only CUDA, won't even request (Message 925632)
Posted 1845 days ago by Harri Liljeroos
Thank you for your answers. Things have moved along and CPU work has now been downloaded and the host is cruching now.


Next 20

Copyright © 2014 University of California