Message boards :
Number crunching :
OpenCL onATI Radeon HD 4670 worked. Once.
Message board moderation
Author | Message |
---|---|
![]() Send message Joined: 25 Jun 99 Posts: 30 Credit: 16,549,509 RAC: 62 ![]() ![]() |
I hear there has been a problem for a long time with OpenCL on ATI GPUs for iMacs with the ATI Radeon HD 4670 for a long time. I had some hope this would be fixed with the SETI@home 7.08 version, but the problem persisted...at least for me. Until today. BOINC claimed it completed ONE of the 45 or so units it downloaded, even though there were some errors in the sterr.txt The rest dumped with Computation Errors right away. I wonder what was different with this ONE special unit that made it claim to work. It would sure be nice if they all did. |
![]() Send message Joined: 25 Jun 99 Posts: 30 Credit: 16,549,509 RAC: 62 ![]() ![]() |
Forgot to mention. I had chalked the success up to running the latest Mac OS X update, bringing this machine up to 10.11.1. This was the first unit run after the update, so I thought the update fixed something. Until all the rest of them dumped. |
![]() ![]() Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 ![]() ![]() |
Worth to look into stderr log that posted in task details on web site: WARNING: This application needs newer GPU, at least ATI Radeon HD 5000 needed, exiting ! App not supposed to run on your GPU at all. Wording "WARNING" along with "exiting" little deceiving. ERROR would be more appropriate in such case cause execution can not be continued. And the reason of error - GPU task issued to unsupported GPU. Error in plan_class that allowed such action. Pity that beta testing did not discover that missing in plan class. |
Urs Echternacht ![]() Send message Joined: 15 May 99 Posts: 692 Credit: 135,197,781 RAC: 211 ![]() ![]() |
There are three OpenCL ATI Mac OS X apps each coming with a plan_class : - opencl_ati5zc_mac (for ATI Radeon HD 5000 and newer GPUs) - opencl_ati5_mac (for ATI Radeon HD 5000 and newer GPUs) - opencl_ati_mac (for ATI Radeon HD 4000 and newer GPUs) The first two app versions throw the error-message that Raistmer found when run on ATI HD 4000, because there was no way to exclude the ATI HD 4000 generation of GPUs via a plan_class from getting these apps (until time of release). Therefor the app itself had to warn the user and exit with an error before doing anything. The last app version is the one with the problem. Here is a stderr with the error that is thrown, because of app failure : Stderr Ausgabe _\|/_ U r s |
![]() ![]() Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 ![]() ![]() |
The first two app versions throw the error-message that Raistmer found when run on ATI HD 4000, because there was no way to exclude the ATI HD 4000 generation of GPUs via a plan_class from getting these apps (until time of release). What prevents? |
Eric Korpela ![]() Send message Joined: 3 Apr 99 Posts: 1382 Credit: 54,506,847 RAC: 60 ![]() ![]() |
Looking at the host it doesn't appear that it can successfully run any OpenCL app. My understanding was that there is nothing about the ati5 apps that prevent them from running on pre-ati5 architectures. They are just written to execute slightly differently for better performance on different GPUs. I suspect that this is a problem with whatever AMD/ATI driver is being used. AMD/ATI has a habit of ignoring problems new drivers cause on older hardware. Since AMD/ATI removed CAL HW version from their drivers, there's no way in the current BOINC code to exclude a ATI GPU based upon a device hardware version on a Mac. On PCs with old drivers we can still use CAL version. But for Mac, the best we have now is OpenCL version, but AMD reserves the right to uninvite specific hardware and specific operating systems from the party with each new driver release. @SETIEric@qoto.org (Mastodon) ![]() |
![]() ![]() Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 ![]() ![]() |
I see. Pity that BOINC doesn't able to parse clInfo in right degree to see that HD4xxx hardware present. As one can see from app's own output it is doable. |
Eric Korpela ![]() Send message Joined: 3 Apr 99 Posts: 1382 Credit: 54,506,847 RAC: 60 ![]() ![]() |
I see. Pity that BOINC doesn't able to parse clInfo in right degree to see that HD4xxx hardware present. As one can see from app's own output it is doable. We don't really have the resources to chase what GPU manufacturers name their cards (ATI is out of meaningful 4 digit numbers) or whether their cards live up to the standards they claim to follow. A name is a marketing device, not an engineering standard. An HD4760 may have nothing to do with what an HD4762 or an HDR4760 are capable of. An HD4760 manufactured in 2016 could be entirely different from one made in 2014. I guess I'll look through and see whether any OpenCL 1.0 cards are able to run 7.08 on Macintosh. If the answer is no, I can bump the requirement to OpenCL 1.1. @SETIEric@qoto.org (Mastodon) ![]() |
![]() ![]() Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 ![]() ![]() |
I see. Pity that BOINC doesn't able to parse clInfo in right degree to see that HD4xxx hardware present. As one can see from app's own output it is doable. Agree.
Good plan. |
Urs Echternacht ![]() Send message Joined: 15 May 99 Posts: 692 Credit: 135,197,781 RAC: 211 ![]() ![]() |
I see. Pity that BOINC doesn't able to parse clInfo in right degree to see that HD4xxx hardware present. As one can see from app's own output it is doable. Would prefer to find out why one the ATI HD 4000 app works on Windows and Linux, but not on Mac OS X. Sadly no one has reacted to my call for help over the last two weeks, so no testing resources are available. _\|/_ U r s |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 ![]() ![]() |
If I thought it would help I'd try my PC4670 in my Mac. However, I don't see how a full sized PC card would give the same environment as the Mini that's in an iMac. Plus, I don't even know if the PC4670 would work with OpenCL in a Mac, I know it will give an image on the screen. Here's another OpenCL 1.0 Card on a Mac, http://setiathome.berkeley.edu/results.php?hostid=4698278 BTW, I think I just got Robbed by two of those notoriously unreliable Mac Intel GPUs. From my experience the Intel GPU AP App isn't to be trusted, My GPU meanwhile HAD a spotless record, http://setiathome.berkeley.edu/workunit.php?wuid=1921981215 I wuz Robbed... |
Urs Echternacht ![]() Send message Joined: 15 May 99 Posts: 692 Credit: 135,197,781 RAC: 211 ![]() ![]() |
If I thought it would help I'd try my PC4670 in my Mac. However, I don't see how a full sized PC card would give the same environment as the Mini that's in an iMac. Plus, I don't even know if the PC4670 would work with OpenCL in a Mac, I know it will give an image on the screen.Would be great if you could try to reproduce that error in a standalone run with -v 2 option set. That could bring us a little closer to where in code the error happens. Here's another OpenCL 1.0 Card on a Mac, http://setiathome.berkeley.edu/results.php?hostid=4698278, which finishes workunits sporadically successful : other problem ! BTW, I think I just got Robbed by two of those notoriously unreliable Mac Intel GPUs. From my experience the Intel GPU AP App isn't to be trusted, My GPU meanwhile HAD a spotless record, http://setiathome.berkeley.edu/workunit.php?wuid=1921981215Could we stay on topic please. As long as the validator validates the way it does we all get "robbed" from time to time. _\|/_ U r s |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 ![]() ![]() |
If I thought it would help I'd try my PC4670 in my Mac. However, I don't see how a full sized PC card would give the same environment as the Mini that's in an iMac. Plus, I don't even know if the PC4670 would work with OpenCL in a Mac, I know it will give an image on the screen.Would be great if you could try to reproduce that error in a standalone run with -v 2 option set. That could bring us a little closer to where in code the error happens. Well, in my book, the Hosts returning the Errors aren't really a problem. The Host returning 'Successful' tasks that are Invalid, but chosen as Valid results when they team up Are a problem. My new card had Never received an Invalid in over 1100 AP tasks, the other Two have Very recently received Invalids. Just look at the Host Details pages and note the Consecutive number and the Completed numbers. I'll look into the 4670 later today. |
![]() Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 ![]() |
We don't really have the resources to chase what GPU manufacturers name their cards But on the other hand, can't you just filter on "Radeon HD 4670"? The newer GPUs from AMD are no longer called "HD XXXX" but "Rn XXX", e.g. R5 270, R7 280, R9 290. No other GPU manufacturer calls his GPU Radeon, that's property from AMD/ATI. The BOINC client sends that whole name along with scheduler requests, in Jonathan's case AMD Radeon HD 4670 (256MB), so I don't see why you can't have the server tell that scheduler/work requests with that sequence should not be sent applications for the 5xxx style GPUs and better. As for why the 5xxx style applications don't work on these GPUs, isn't that because these GPUs can only run the Multibeam application at a maximum workgroup size 128, where the workgroup size for the 5xxx apps is 256 (or more)? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 ![]() ![]() |
If I thought it would help I'd try my PC4670 in my Mac. However, I don't see how a full sized PC card would give the same environment as the Mini that's in an iMac. Plus, I don't even know if the PC4670 would work with OpenCL in a Mac, I know it will give an image on the screen.Would be great if you could try to reproduce that error in a standalone run with -v 2 option set. That could bring us a little closer to where in code the error happens. I just happened to think...I don't have the App to test. doh. So, remember that App I posted on Beta a while back? Said it MIGHT work on the HD4 cards. Well, it appears to work on My HD4 card. At least on bench. BOINC says; Mon Nov 2 19:30:19 2015 | | Starting BOINC client version 7.2.33 for x86_64-apple-darwin 19:43:49 (687): Can't open init data file - running in standalone mode Not using mb_cmdline.txt-file, using commandline options. 19:43:49 (687): Can't open init data file - running in standalone mode WARNING: init_data.xml missing OpenCL platform detected: Apple WARNING: BOINC supplied wrong platform! Number of OpenCL devices found : 3 BOINC assigns slot on device #0. WARNING: BOINC failed to provide OpenCL device, using own enumeration abilities Build features: SETI7 Non-graphics OpenCL OCL_CHIRP3 FFTW SSE4.1 64bit System: Darwin x86_64 Kernel: 13.4.0 CPU : Intel(R) Xeon(R) CPU E5472 @ 3.00GHz GenuineIntel x86, Family 6 Model 23 Stepping 6 Features : FPU TSC PAE APIC MTRR MMX SSE SSE2 HT SSE3 SSSE3 SSE4.1 OpenCL-kernels filename : MultiBeam_Kernels_r2934.cl INFO: can't open binary kernel file: .//MultiBeam_Kernels_r2934.cl_RadeonHD4670.bin_V7_13.4.0_10, continue with recompile... Info : Building Program (binary, clBuildProgram):main kernels: OK code 0 INFO: binary kernel file created WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_524288_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_8_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_16_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_32_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_64_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_128_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_256_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_512_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_1024_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_2048_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_4096_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_8192_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_16384_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_32768_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_65536_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 WARNING: Used device has emulated local memory, WG restriction applied. User-defined WG size will be ignored if exceeds allowed value WARNING: Used device has low amount of local memory,local memory FFT size will be reduced. User-defined local FFT size will be ignored if exceeds allowed value WARNING: can't open binary kernel file for oclFFT plan: .//MB_clFFTplan_RadeonHD4670_131072_gr64_lr16_wg64_tw0_r2934.bin_13.4.0_10, continue with recompile... WARNING: patching required max_kernel_wg_size=32 ar=0.775000 NumCfft=1169 NumGauss=6087368 NumPulse=1198014592 NumTriplet=2302301056 Currently allocated 145 MB for GPU buffers In v_BaseLineSmooth: NumDataPoints=1048576, BoxCarLength=8192, NumPointsInChunk=32768 OS X optimized S@H v7 application (based on S@H Enhanced by Alex Kan) Version info: SSE4.1x (Intel, Core 2-optimized v8-nographics) V5.13 by Alex Kan SSE4.1x OS X 64bit Build 2934 , Ported by : Raistmer, JDWhale, Urs Echternacht OpenCL version by Raistmer, r2934 Number of OpenCL platforms: 1 OpenCL Platform Name: Apple Number of devices: 3 Max compute units: 8 Max work group size: 1024 Max clock frequency: 750Mhz Max memory allocation: 134217728 Cache type: None Cache line size: 0 Cache size: 0 Global memory size: 402653184 Constant buffer size: 65536 Max number of constant args: 8 Local memory type: Scratchpad Local memory size: 16384 Queue properties: Out-of-Order: No Name: Radeon HD 4670 Vendor: AMD Driver version: 1.0 Version: OpenCL 1.0 Extensions: cl_APPLE_SetMemObjectDestructor cl_APPLE_ContextLoggingFunctions cl_APPLE_clut cl_APPLE_query_kernel_names cl_APPLE_gl_sharing cl_khr_gl_event Max compute units: 14 Max work group size: 256 Max clock frequency: 775Mhz Max memory allocation: 268435456 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 Queue properties: Out-of-Order: No Name: ATI Radeon Barts PRO Prototype Vendor: AMD Driver version: 1.2 (Sep 11 2015 11:29:57) Version: OpenCL 1.2 Extensions: cl_APPLE_SetMemObjectDestructor cl_APPLE_ContextLoggingFunctions cl_APPLE_clut cl_APPLE_query_kernel_names cl_APPLE_gl_sharing cl_khr_gl_event 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_byte_addressable_store cl_khr_image2d_from_buffer cl_khr_depth_images Max compute units: 14 Max work group size: 256 Max clock frequency: 900Mhz Max memory allocation: 268435456 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 Queue properties: Out-of-Order: No Name: ATI Radeon Barts XT Prototype Vendor: AMD Driver version: 1.2 (Sep 11 2015 11:29:57) Version: OpenCL 1.2 Extensions: cl_APPLE_SetMemObjectDestructor cl_APPLE_ContextLoggingFunctions cl_APPLE_clut cl_APPLE_query_kernel_names cl_APPLE_gl_sharing cl_khr_gl_event 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_byte_addressable_store cl_khr_image2d_from_buffer cl_khr_depth_images Work Unit Info: ............... Credit multiplier is : 2.85 WU true angle range is : 0.775000 Used GPU device parameters are: Number of compute units: 8 Single buffer allocation size: 64MB Total device global memory: 384MB max WG size: 1024 local mem type: Emulated period_iterations_num=23 Pulse: peak=0.6058553, time=69.27, period=0.1079, d_freq=1418922119.13, score=1.047, chirp=0, fft_len=32 Triplet: peak=8.350917, time=80.39, period=0.5439, d_freq=1418924865.71, chirp=0, fft_len=128 Triplet: peak=8.563328, time=80.39, period=0.5439, d_freq=1418924865.71, chirp=0, fft_len=128 Autocorr: peak=12.93239, time=20.13, delay=4.969, d_freq=1418920271.34, chirp=-0.83183, fft_len=128k Spike: peak=19.36639, time=46.98, d_freq=1418916062.37, chirp=2.4955, fft_len=128k Spike: peak=19.67712, time=73.82, d_freq=1418918905.44, chirp=-4.8986, fft_len=128k Spike: peak=21.39584, time=33.55, d_freq=1418915733.83, chirp=-5.638, fft_len=128k Spike: peak=18.88078, time=20.13, d_freq=1418920463.53, chirp=-7.0244, fft_len=128k Autocorr: peak=13.35424, time=33.55, delay=2.1202, d_freq=1418920561, chirp=8.1335, fft_len=128k Spike: peak=18.92986, time=20.13, d_freq=1418917368.68, chirp=-9.335, fft_len=128k Spike: peak=19.28574, time=87.24, d_freq=1418920545.39, chirp=-11.276, fft_len=128k Spike: peak=19.21026, time=90.26, d_freq=1418919043.86, chirp=-11.461, fft_len=512 Gaussian: peak=2.355918, mean=0.6171291, ChiSq=0.99945, time=41.1, d_freq=1418924703.22, score=0.3251901, null_hyp=1.438255, chirp=-20.426, fft_len=16k Spike: peak=19.76917, time=23.07, d_freq=1418916784.01, chirp=-20.796, fft_len=8k Spike: peak=19.70372, time=90.26, d_freq=1418919039.42, chirp=-22.922, fft_len=512 Pulse: peak=8.095232, time=89.99, period=4.535, d_freq=1418919045.43, score=1.089, chirp=-22.922, fft_len=512 Gaussian: peak=2.42509, mean=0.612058, ChiSq=0.9409937, time=9.227, d_freq=1418920421.68, score=0.1304379, null_hyp=1.421672, chirp=-22.922, fft_len=16k Gaussian: peak=2.620608, mean=0.6370828, ChiSq=1.001976, time=32.72, d_freq=1418919848.14, score=0.2222714, null_hyp=1.430435, chirp=-25.417, fft_len=16k Gaussian: peak=2.55242, mean=0.6347769, ChiSq=1.01491, time=36.07, d_freq=1418919762.86, score=0.4444447, null_hyp=1.455097, chirp=-25.417, fft_len=16k Spike: peak=20.13485, time=90.26, d_freq=1418919042.56, chirp=34.382, fft_len=512 Gaussian: peak=2.551749, mean=0.6185498, ChiSq=0.9597445, time=39.43, d_freq=1418919421.81, score=0.2516403, null_hyp=1.432027, chirp=-36.878, fft_len=16k Gaussian: peak=2.514693, mean=0.620205, ChiSq=0.946231, time=41.1, d_freq=1418919359.94, score=0.2128706, null_hyp=1.428727, chirp=-36.878, fft_len=16k Gaussian: peak=2.580093, mean=0.6034984, ChiSq=0.9280221, time=24.33, d_freq=1418915417.99, score=0.2594237, null_hyp=1.432688, chirp=40.852, fft_len=16k Spike: peak=18.54538, time=95.63, d_freq=1418917049.85, chirp=-41.407, fft_len=32k Pulse: peak=2.358688, time=76.4, period=0.7864, d_freq=1418920638.41, score=1.132, chirp=-45.843, fft_len=256 Best spike: peak=21.39584, time=33.55, d_freq=1418915733.83, chirp=-5.638, fft_len=128k Best autocorr: peak=13.35424, time=33.55, delay=2.1202, d_freq=1418920561, chirp=8.1335, fft_len=128k Best gaussian: peak=2.55242, mean=0.6347769, ChiSq=1.01491, time=36.07, d_freq=1418919762.86, score=0.4444447, null_hyp=1.455097, chirp=-25.417, fft_len=16k Best pulse: peak=2.358688, time=76.4, period=0.7864, d_freq=1418920638.41, score=1.132, chirp=-45.843, fft_len=256 Best triplet: peak=8.563328, time=80.39, period=0.5439, d_freq=1418924865.71, chirp=0, fft_len=128 Flopcounter: 147250443239.658813 Spike count: 11 Autocorr count: 2 Pulse count: 3 Triplet count: 2 Gaussian count: 7 Time cpu in use since last restart: 62.1 seconds GPU device sync requested... ...GPU device synched 19:47:04 (687): called boinc_finish(0) That's even the Correct results. I wonder how long it would take on a normal shorty... |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 ![]() ![]() |
If I thought it would help I'd try my PC4670 in my Mac. However, I don't see how a full sized PC card would give the same environment as the Mini that's in an iMac. Plus, I don't even know if the PC4670 would work with OpenCL in a Mac, I know it will give an image on the screen.Would be great if you could try to reproduce that error in a standalone run with -v 2 option set. That could bring us a little closer to where in code the error happens. OK, WTH is this? http://setiathome.berkeley.edu/workunit.php?wuid=1922047397 How can you get an Invalid on a Blanking too much RFI? percent blanked: 100.00 I think I see a pattern here. I was sent a Resend, the other Host completed the task before I did. So, I GET AN INVALID! For Successfully completing the Resend that was sent Me? I remember this from a long time ago, Why has it resurfaced? Not happy... Oh, the 4670 is running a shorty offline. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 ![]() ![]() |
It finally finished the 'Shorty'. 2 hours 34 minutes...for a shorty. The numbers look good, it's this task finished by my 6850 in 8 minutes 38 seconds, http://setiathome.berkeley.edu/result.php?resultid=4492715295 WU true angle range is : 1.267785 Used GPU device parameters are: Number of compute units: 14 Single buffer allocation size: 128MB Total device global memory: 1024MB max WG size: 256 local mem type: Real LotOfMem path: yes period_iterations_num=64 Triplet: peak=9.509696, time=24.18, period=1.058, d_freq=1418791046.14, chirp=0, fft_len=64 Autocorr: peak=19.88901, time=60.4, delay=5.2296, d_freq=1418789216.35, chirp=2.5473, fft_len=128k Triplet: peak=9.397565, time=24.18, period=1.058, d_freq=1418791038.6, chirp=5.9994, fft_len=64 Triplet: peak=9.497669, time=24.18, period=1.058, d_freq=1418791031.05, chirp=11.999, fft_len=64 Autocorr: peak=18.36794, time=73.82, delay=2.7014, d_freq=1418790007.88, chirp=12.807, fft_len=128k Triplet: peak=9.767583, time=24.18, period=1.058, d_freq=1418791023.51, chirp=17.998, fft_len=64 Triplet: peak=9.760864, time=24.18, period=1.058, d_freq=1418791015.96, chirp=23.998, fft_len=64 Triplet: peak=9.475115, time=24.18, period=1.058, d_freq=1418791008.41, chirp=29.997, fft_len=64 Triplet: peak=9.041977, time=24.18, period=1.058, d_freq=1418791000.87, chirp=35.996, fft_len=64 Triplet: peak=8.912941, time=56.54, period=1.245, d_freq=1418793450.44, chirp=40.496, fft_len=256 Best spike: peak=23.79977, time=60.4, d_freq=1418789340.64, chirp=-25.154, fft_len=128k Best autocorr: peak=19.88901, time=60.4, delay=5.2296, d_freq=1418789216.35, chirp=2.5473, fft_len=128k Best gaussian: peak=0, mean=0, ChiSq=0, time=-2.122e+11, d_freq=0, score=-12, null_hyp=0, chirp=0, fft_len=0 Best pulse: peak=2.466364, time=80.44, period=0.2277, d_freq=1418786422.46, score=0.9858, chirp=-47.995, fft_len=32 Best triplet: peak=9.767583, time=24.18, period=1.058, d_freq=1418791023.51, chirp=17.998, fft_len=64 Flopcounter: 11907137883730.005859 Spike count: 0 Autocorr count: 2 Pulse count: 0 Triplet count: 8 Gaussian count: 0 Time cpu in use since last restart: 148.1 seconds GPU device sync requested... ...GPU device synched 20:31:37 (1062): called boinc_finish(0) The numbers look close enough to me, 20:22:02 (1044): Can't open init data file - running in standalone mode ... WU true angle range is : 1.267785 Used GPU device parameters are: Number of compute units: 8 Single buffer allocation size: 64MB Total device global memory: 384MB max WG size: 1024 local mem type: Emulated period_iterations_num=23 Triplet: peak=9.509691, time=24.18, period=1.058, d_freq=1418791046.14, chirp=0, fft_len=64 Autocorr: peak=19.88901, time=60.4, delay=5.2296, d_freq=1418789216.35, chirp=2.5473, fft_len=128k Triplet: peak=9.397559, time=24.18, period=1.058, d_freq=1418791038.6, chirp=5.9994, fft_len=64 Triplet: peak=9.497672, time=24.18, period=1.058, d_freq=1418791031.05, chirp=11.999, fft_len=64 Autocorr: peak=18.36793, time=73.82, delay=2.7014, d_freq=1418790007.88, chirp=12.807, fft_len=128k Triplet: peak=9.767584, time=24.18, period=1.058, d_freq=1418791023.51, chirp=17.998, fft_len=64 Triplet: peak=9.760864, time=24.18, period=1.058, d_freq=1418791015.96, chirp=23.998, fft_len=64 Triplet: peak=9.475114, time=24.18, period=1.058, d_freq=1418791008.41, chirp=29.997, fft_len=64 Triplet: peak=9.041972, time=24.18, period=1.058, d_freq=1418791000.87, chirp=35.996, fft_len=64 Triplet: peak=8.912944, time=56.54, period=1.245, d_freq=1418793450.44, chirp=40.496, fft_len=256 Best spike: peak=23.79977, time=60.4, d_freq=1418789340.64, chirp=-25.154, fft_len=128k Best autocorr: peak=19.88901, time=60.4, delay=5.2296, d_freq=1418789216.35, chirp=2.5473, fft_len=128k Best gaussian: peak=0, mean=0, ChiSq=0, time=-2.122e+11, d_freq=0, score=-12, null_hyp=0, chirp=0, fft_len=0 Best pulse: peak=2.466364, time=80.44, period=0.2277, d_freq=1418786422.46, score=0.9858, chirp=-47.995, fft_len=32 Best triplet: peak=9.767584, time=24.18, period=1.058, d_freq=1418791023.51, chirp=17.998, fft_len=64 Flopcounter: 14724690759208.898438 Spike count: 0 Autocorr count: 2 Pulse count: 0 Triplet count: 8 Gaussian count: 0 Time cpu in use since last restart: 2867.2 seconds GPU device sync requested... ...GPU device synched 22:56:57 (1044): called boinc_finish(0) If anyone wants to test it, it's here; http://www.arkayn.us/forum/index.php?topic=130.msg4364#msg4364 Have Fun. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 ![]() ![]() |
So, it appears the App works fine the First time it's run on the Radeon 4670, however if you restart the App or run it the second time it Errors out. Seems the problem is the Kernel file it leaves after running. The file MultiBeam_Kernels_r2934.cl_RadeonHD4670.bin_V7_13.4.0_10 is the Only kernel file it leaves behind, if you Remove that file before starting the App, it appears to run normally. If the file is present during launch you receive the Error; ... Currently allocated 185 MB for GPU buffers SIGSEGV: segmentation violation Crashed executable name: MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin Machine type Intel 80486 (64-bit executable) System version: Macintosh OS 10.9.5 build 13F1134 Tue Nov 3 02:00:58 2015 -- Warning: /usr/bin/atos is moving and will be removed from a future OS X release. It is now available in the Xcode developer tools to be invoked via: `xcrun atos` To silence this warning, pass the '-d' command-line flag to this tool. -- atos cannot load symbols for the file MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin for architecture x86_64. 0 MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin 0x00000001090f2bfb SIGPIPE: write on a pipe with no reader 1 MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin 0x00000001090e3ed6 SIGPIPE: write on a pipe with no reader 2 libsystem_platform.dylib 0x00007fff951755aa SIGPIPE: write on a pipe with no reader 3 libsystem_c.dylib 0x00007fff9462988c SIGPIPE: write on a pipe with no reader 4 CoreFoundation 0x00007fff8d8d5710 SIGPIPE: write on a pipe with no reader 5 CoreFoundation 0x00007fff8d8d538e SIGPIPE: write on a pipe with no reader 6 ATIRadeonX2000GLDriver 0x000000010a834687 SIGPIPE: write on a pipe with no reader 7 OpenCL 0x00007fff93088f8a SIGPIPE: write on a pipe with no reader 8 OpenCL 0x00007fff93075e29 SIGPIPE: write on a pipe with no reader 9 MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin 0x0000000108f35e2b SIGPIPE: write on a pipe with no reader 10 MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin 0x0000000108f37808 SIGPIPE: write on a pipe with no reader 11 MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin 0x0000000108f7dc62 SIGPIPE: write on a pipe with no reader 12 MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin 0x0000000108f83e02 SIGPIPE: write on a pipe with no reader 13 MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin 0x0000000108f337d0 SIGPIPE: write on a pipe with no reader 14 MBv7_7.08r2934_ati_sse41_x86_64-apple-darwin 0x0000000108d8e744 Thread 0 crashed with X86 Thread State (64-bit): rax: 0x0100001f rbx: 0x00000003 rcx: 0x7fff56e6f5a8 rdx: 0x00000028 rdi: 0x7fff56e6f610 rsi: 0x00000003 rbp: 0x7fff56e6f5f0 rsp: 0x7fff56e6f5a8 r8: 0x00000c0b r9: 0x00000000 r10: 0x000003b0 r11: 0x00000206 r12: 0x000003b0 r13: 0x00000028 r14: 0x7fff56e6f610 r15: 0x00000c0b rip: 0x7fff8b068a1a rfl: 0x00000206 ... It doesn't leave the other Kernel files after running, it shouldn't leave this one either. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 ![]() ![]() |
It took a little longer running under BOINC. My guess is the 4670 doesn't like the oclFFT Plan, back when first trying the new APv7 App it gave much longer times using the oclFFT settings. But it did finish, http://setiathome.berkeley.edu/result.php?resultid=4492978697 Run time: 3 hours 19 min 24 sec CPU time: 46 min 37 sec Validate state: Valid |
Urs Echternacht ![]() Send message Joined: 15 May 99 Posts: 692 Credit: 135,197,781 RAC: 211 ![]() ![]() |
I just happened to think...I don't have the App to test. doh. TBar, check your client_state.xml for the download url. At Beta it was http://boinc2.ssl.berkeley.edu/beta/download/setiathome_7.08_x86_64-apple-darwin__opencl_ati_mac (or similar => r2936) Please use -v 2 commandline option. _\|/_ U r s |
©2023 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.