Message boards :
Number crunching :
The Highest Ranked SETI AMD Host is a MAC: Time for a STOCK MAC APP?
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 . . . 21 · Next
Author | Message |
---|---|
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
I had to give up on Beta. The machine had gone over 6 hours without any work. According to the Details page, it had completed over 400 MB ATI tasks without any trouble, so, I had to go where the tasks were. Seems it may have been a little chilly last night, or it's just that the Shorties run at a higher GPU load than the longer tasks. Right now it's running 2 at a time and the IO sensor reads 99F. Also, the CPU load is low when running Shorties and rises when running the longer tasks. It's now running here; SETI@home v7 tasks for computer 6796479 I suppose this will work until the APs arrive....hopefully tomorrow ;-) |
Chris Adamek Send message Joined: 15 May 99 Posts: 251 Credit: 434,772,072 RAC: 236 |
Yeah I've had not any troubles either. Nice to not have idle GPU's on the Macs finally! I do wish there was a decent GPU load app for the Mac like there are on other platforms. Oh well. Thanks, Chris |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
I do wish there was a decent GPU load app for the Mac like there are on other platforms. Oh well. Cause same codebase used for Mac builds I would compare defines listed in stderr for OS X and Windows versions first to find the reason of slowdown. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
OK, the only Windows on my Mac is the decrepit XP. I do have a recent Ubuntu though with Cat 14.9 drivers, Computer 6906726. I just ran a few with the only setting -sbs 256. I'm only getting 2 estimates in Linux, 7.5 minutes and ~20 minutes. The 20 minute ones in Linux actually finish in 20 minutes or less. The 22 minutes ones in OSX, with the same setting, are taking over 30 minutes with high CPU times. According to ATI Config in Linux the GPU load is in the Mid-90% range. Any suggestions for OSX? |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
|
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
That was even less helpful than the first time. Perhaps you could give some hint as to which defines I'm looking for? |
Chris Adamek Send message Joined: 15 May 99 Posts: 251 Credit: 434,772,072 RAC: 236 |
I don't know if it is any real indication, but under application details the Mac openCL app gets between 35-40 GFLops, which is scarcely faster than the CPU scores. That's said they run much faster than the CPU by several times depending on the GPU type (my ati is about double the speed of the iGPU which is about double the speed of the CPU) that said, the fans don't even come on on either machine, which they do with astropulse. Purely qualitative without any means of measuring GPU load... Probably not overly helpful but that is what I've noticed. Chris |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
ERROR: clFFT_CreatePlan[7] failed: -46 And that concludes this round of testing. That setting Works in Windows... So far my results are, the best cmdline settings to use is NO settings. That gives the best times on VHARs, but doesn't help the longer tasks that take close to twice as long as on other platforms, and use close to 50% CPU time. So, to run the longer tasks closer to other platforms' times you need to run 2 tasks at a time. That still doesn't help the 50% CPU time, and 50% CPU time on 6 tasks will just about Kill a Quad core processor. About all you can do is hope you don't run 6 long tasks at the same time... When are the AstroPulses going to be released? Soon? Not quick enough I'm afraid. yawn. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Build features: SETI7 Non-graphics OpenCL USE_OPENCL_HD5xxx OCL_CHIRP3 ASYNC_SPIKE FFTW SSE3 64bit |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
I woke up to an Error this morning; http://setiathome.berkeley.edu/result.php?resultid=3844537330 <core_client_version>7.4.26</core_client_version> <![CDATA[ <message> process exited with code 193 (0xc1, -63) </message> <stderr_txt> OpenCL platform detected: Apple Number of OpenCL devices found : 3 BOINC assigns slot on device #0. ... Work Unit Info: ............... Credit multiplier is : 2.85 WU true angle range is : 0.566235 Used GPU device parameters are: Number of compute units: 14 Single buffer allocation size: 64MB Total device global memory: 1024MB max WG size: 256 local mem type: Real period_iterations_num=23 Spike: peak=24.22948, time=73.82, d_freq=1421171185.69, chirp=1.7311, fft_len=128k Spike: peak=24.40215, time=73.82, d_freq=1421171185.68, chirp=1.7321, fft_len=128k Spike: peak=24.1709, time=46.98, d_freq=1421174156.7, chirp=15.855, fft_len=128k Spike: peak=24.05127, time=46.98, d_freq=1421174156.7, chirp=15.859, fft_len=128k Triplet: peak=10.21797, time=24.1, period=2.753, d_freq=1421170712.04, chirp=43.543, fft_len=256 Triplet: peak=9.188973, time=24.1, period=2.753, d_freq=1421170706.17, chirp=44.882, fft_len=256 Triplet: peak=10.6643, time=24.1, period=2.753, d_freq=1421170714.26, chirp=45.218, fft_len=256 SIGABRT: abort called Crashed executable name: seti_boinc-hd5xxx built using BOINC library version 7.5.0 Machine type Intel 80486 (64-bit executable) System version: Macintosh OS 10.10.1 build 14B25 Mon Dec 8 06:54:28 2014 atos cannot load symbols for the file seti_boinc-hd5xxx for architecture x86_64. 0 seti_boinc-hd5xxx 0x000000010427fe1b 1 seti_boinc-hd5xxx 0x000000010426cd41 2 libsystem_platform.dylib 0x00007fff9a35df1a 3 ??? 0x00007fd2326066d8 4 libsystem_c.dylib 0x00007fff93191b73 5 libsystem_pthread.dylib 0x00007fff9bb9bcfe 6 libsystem_pthread.dylib 0x00007fff9bb9bdb2 7 libsystem_pthread.dylib 0x00007fff9bb96c99 8 OpenCL 0x00007fff9633afa4 9 OpenCL 0x00007fff9633ae40 10 seti_boinc-hd5xxx 0x0000000104085d66 11 ??? 0x0000000000000003 Thread 0 crashed with X86 Thread State (64-bit): rax: 0x0100001f rbx: 0x00000000 rcx: 0x7fff5bbcc288 rdx: 0x00000028 rdi: 0x7fff5bbcc2f0 rsi: 0x00000003 rbp: 0x7fff5bbcc2d0 rsp: 0x7fff5bbcc288 r8: 0x00000a0f r9: 0x00000000 r10: 0x000003b0 r11: 0x00000206 r12: 0x000003b0 r13: 0x00000028 r14: 0x7fff5bbcc2f0 r15: 0x00000a0f rip: 0x7fff8d26852e rfl: 0x00000206 .... The other host found; Spike count: 4 Autocorr count: 0 Pulse count: 0 Triplet count: 3 Gaussian count: 0 So it looks like it crashed near the finish. !!!!!!!!!!!!!! Whoa, My AP Pendings are validating and my RAC is rising. It Lives!!!!! |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Seems as though one of the early ones has received one of those Instant Invalids. As soon as the first Wingperson reported, Mine was deemed Invalid without a Third opinion. http://setiathome.berkeley.edu/workunit.php?wuid=1643201203 Task Computer Sent Time reported Status Run time CPU time Credit Application 3841159156 6796479 6 Dec 2014, 21:15:36 UTC 7 Dec 2014, 5:32:09 UTC Completed, marked as invalid 602.49 110.73 0.00 SETI@home v7 Anonymous platform (ATI GPU) 3841159157 6922409 6 Dec 2014, 21:15:38 UTC 12 Dec 2014, 19:05:32 UTC Completed, validation inconclusive 5,227.19 5,206.68 pending SETI@home v7 v7.00 3853155918 --- --- --- Unsent Both Hosts found the Same Results... Currently the machine is fighting a Horde of the Low Angle Range tasks that are taking Forever, http://setiathome.berkeley.edu/results.php?hostid=6796479&offset=0&show_names=0&state=0&appid=11 If the APs ever get fixed, I'm not sure I will be able to resist the Abort button. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
So, I loaded the New Build, and set it to run one of the Low Angle Range tasks here, http://setiathome.berkeley.edu/workunit.php?wuid=1648426203. Looks like it will be just as slow as the First build on the LARs. The Output is normal so far; OpenCL platform detected: Apple Number of OpenCL devices found : 3 BOINC assigns slot on device #0. Info: BOINC provided OpenCL device ID used Build features: SETI7 Non-graphics OpenCL USE_OPENCL_HD5xxx OCL_CHIRP3 ASYNC_SPIKE FFTW SSE3 64bit System: Darwin x86_64 Kernel: 14.0.0 CPU : Intel(R) Xeon(R) CPU E5462 @ 2.80GHz GenuineIntel x86, Family 6 Model 23 Stepping 6 Features : FPU VME DE PSE TSC MSR PAE MCE CX8 APIC SEP MTRR PGE MCA CMOV PAT PSE36 CLFSH DS ACPI MMX FXSR SSE SSE2 SS HTT TM PBE SSE3 DTES64 MON DSCPL VMX EST TM2 SSSE3 CX16 TPR PDCM SSE4.1 OpenCL-kernels filename : MultiBeam_Kernels_r2760.cl INFO: can't open binary kernel file: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MultiBeam_Kernels_r2760.clHD5_ATIRadeonBartsPROPrototype.bin_V7_14.0.0_12Sep28201422, continue with recompile... Info : Building Program (binary, clBuildProgram):main kernels: OK code 0 INFO: binary kernel file created WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_524288_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_8_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_16_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_32_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_64_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_128_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_256_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_512_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_1024_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_2048_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_4096_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_8192_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_16384_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_32768_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_65536_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... WARNING: can't open binary kernel file for oclFFT plan: /Volumes/Mov1/BOINC/Maverick/BOINC Data/projects/setiathome.berkeley.edu/MB_clFFTplan_ATIRadeonBartsPROPrototype_131072_gr64_lr16_wg256_tw0_r2760.bin_14.0.0_12Sep28201422, continue with recompile... ar=0.125303 NumCfft=146007 NumGauss=0 NumPulse=251836619413 NumTriplet=251942388363 Currently allocated 121 MB for GPU buffers In v_BaseLineSmooth: NumDataPoints=1048576, BoxCarLength=8192, NumPointsInChunk=32768 OS X optimized S@H Enhanced application by Alex Kan OpenCL version by Raistmer, r2760 AMD HD5 version by Raistmer Number of OpenCL platforms: 1 OpenCL Platform Name: Apple .......... Work Unit Info: ............... Credit multiplier is : 2.85 WU true angle range is : 0.125303 Used GPU device parameters are: Number of compute units: 14 Single buffer allocation size: 64MB Total device global memory: 1024MB max WG size: 256 local mem type: Real period_iterations_num=23 Autocorr: peak=17.88812, time=100.7, delay=5.9049, d_freq=1421152181.68, chirp=-1.6101, fft_len=128k It should be finished in an Hour...or 2 :-( I set the other LARs to run on the CPU, which shouldn't be much longer than on the GPU running 2 at a time... |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Another Error this morning. http://setiathome.berkeley.edu/result.php?resultid=3885235660 <core_client_version>7.4.36</core_client_version> <![CDATA[ <message> process exited with code 193 (0xc1, -63) </message> <stderr_txt> OpenCL platform detected: Apple Number of OpenCL devices found : 3 BOINC assigns slot on device #2. ... Triplet: peak=11.08351, time=24.82, period=4.837, d_freq=1419545249.34, chirp=-1.5611, fft_len=128 Triplet: peak=11.72965, time=24.82, period=4.837, d_freq=1419545236.43, chirp=-2.0814, fft_len=128 Spike: peak=25.04794, time=46.98, d_freq=1419538526.53, chirp=3.0547, fft_len=128k Spike: peak=24.56862, time=46.98, d_freq=1419538526.53, chirp=3.0593, fft_len=128k Spike: peak=24.62643, time=46.98, d_freq=1419538526.54, chirp=3.0612, fft_len=128k Triplet: peak=11.04704, time=24.82, period=4.837, d_freq=1419545248.15, chirp=-4.6832, fft_len=128 Autocorr: peak=18.24141, time=46.98, delay=0.48763, d_freq=1419542029.22, chirp=21.577, fft_len=128k Autocorr: peak=18.06157, time=46.98, delay=0.48763, d_freq=1419542029.31, chirp=21.579, fft_len=128k SIGABRT: abort called Crashed executable name: seti_boinc-hd5xxx built using BOINC library version 7.5.0 Machine type Intel 80486 (64-bit executable) System version: Macintosh OS 10.10.1 build 14B25 Fri Jan 2 07:02:25 2015 atos cannot load symbols for the file seti_boinc-hd5xxx for architecture x86_64. 0 seti_boinc-hd5xxx 0x0000000105aafe1b 1 seti_boinc-hd5xxx 0x0000000105a9cd41 2 libsystem_platform.dylib 0x00007fff8efd4f1a 3 ??? 0x00007fca34100590 4 libsystem_c.dylib 0x00007fff8e08cb73 5 libsystem_pthread.dylib 0x00007fff9126ccfe 6 libsystem_pthread.dylib 0x00007fff9126cdb2 7 libsystem_pthread.dylib 0x00007fff91267c99 8 OpenCL 0x00007fff976790b3 9 OpenCL 0x00007fff97678c0b 10 OpenCL 0x00007fff97679746 11 libdispatch.dylib 0x00007fff94e05c13 12 libdispatch.dylib 0x00007fff94e09365 13 libdispatch.dylib 0x00007fff94e0aecc 14 libdispatch.dylib 0x00007fff94e086b7 15 libdispatch.dylib 0x00007fff94e16fe4 16 libsystem_pthread.dylib 0x00007fff912696cb 17 libsystem_pthread.dylib 0x00007fff912674a1 ... The other host found; Spike count: 8 Autocorr count: 2 Pulse count: 0 Triplet count: 3 Gaussian count: 2 I think I've received one of these with every version of BOINC I've tried except 7.2.33. Back to 7.2.33 I suppose as it appears the BOINC version has absolutely nothing to do about Not receiving GPU APs. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Still nothing new on the development front, http://lunatics.kwsn.net/10-alternate-hardware-platforms/osx-multibeam-opencl-question.msg57873.html#msg57873 I have now run upwards of 7000 ATI tasks with the Mac MB App and haven't suffered a Single Invalid task. I finally did receive one of the Errors with BOINC 7.2.33. So for now, it appears to be quite a usable App. It still gives much lower credit than the AstroPulse Apps, but, it's about the same as with other platforms using ATI cards, maybe a little better. The best performance is when running 2 MB tasks at a time and avoiding the Lower Angle Range tasks. It runs "Shorties" very well. http://setiathome.berkeley.edu/results.php?hostid=6796479&offset=20&show_names=0&state=0&appid=11 |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
Think they will ever adapt it for the NV cards in the apples? Got 2 MacBook Pro where the GPUs are sitting idle :( |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Hmmmm.... The first link in my last post will take you to the post with the Download link. That package contains the Apps; seti_boinc-hd5xxx seti_boinc-intel seti_boinc-nv So, the nVidia version has been out for a while now. I haven't seen any reports of it being used though. |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
I'll try tomorrow. Football today and rest |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
Installed both the NV and intel just in case. Neither has had any downloads. I've checked and made sure I've got "use Intel" and "Nvidia" check in my preferences as well as setiv7 as well as the astropulse clicked. Only thing running on the GPU is the astropulses. Took forever to get them but nothing else has downloaded. The CPU is full of work units. I'll leave it and see if anything happens over the next few days |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Check also that MultiBeam work allowed, not only GPU types. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Installed both the NV and intel just in case. If you just want to test the App and see if it works you can easily set a CPU MB to run on the other device. All you have to do is change the <results> entry in the client_state.xml file to match your app_info entry for the other device. In my case I would Suspend the task, Stop BOINC and change the Suspended CPU MB task; <result> <name>16jl12aa.20539.22158.438086664206.12.198_0</name> <final_cpu_time>0.000000</final_cpu_time> <final_elapsed_time>0.000000</final_elapsed_time> <exit_status>0</exit_status> <state>2</state> <platform>x86_64-apple-darwin</platform> <version_num>705</version_num> <suspended_via_gui/> <wu_name>16jl12aa.20539.22158.438086664206.12.198</wu_name> <report_deadline>1426122956.000000</report_deadline> <received_time>1421422702.375307</received_time> <file_ref> <file_name>16jl12aa.20539.22158.438086664206.12.198_0_0</file_name> <open_name>result.sah</open_name> </file_ref> </result> To <result> <name>16jl12aa.20539.22158.438086664206.12.198_0</name> <final_cpu_time>0.000000</final_cpu_time> <final_elapsed_time>0.000000</final_elapsed_time> <exit_status>0</exit_status> <state>2</state> <platform>x86_64-apple-darwin</platform> <version_num>703</version_num> <plan_class>opencl_ati5_mac</plan_class> <suspended_via_gui/> <wu_name>16jl12aa.20539.22158.438086664206.12.198</wu_name> <report_deadline>1426122956.000000</report_deadline> <received_time>1421422702.375307</received_time> <file_ref> <file_name>16jl12aa.20539.22158.438086664206.12.198_0_0</file_name> <open_name>result.sah</open_name> </file_ref> </result> Of course Your app_info would have different entries than mine. Provided everything else is set correctly, when you start BOINC you will see the task listed as a MB GPU task. Cross your fingers and Resume the task... Good Luck |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.