Setting up Linux to crunch CUDA90 and above for Windows users

Author	Message
TBar Volunteer tester Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 6,279	Message 1950989 - Posted: 21 Aug 2018, 0:07:14 UTC - in response to Message 1950977. Last modified: 21 Aug 2018, 0:08:10 UTC So he installed the Same System again? Expecting Different Results this time? Hopefully it will work out.... Which System/Kernel is he having trouble with? I'll mark that down for future reference. Stephen, you appear to be a little behind. 0.96 had problems from Day One with the Arecibo tasks. Finally it was put to rest after a horde of Shorties. See, V0.96 Ate My Shorties...Just Like TBar Said it Would. It was replaced by the Bug Fix V0.97 on Saturday. Also on Saturday We had a large number of Volunteer Alpha Tests step up to test the Bug Fix Release...probably didn't know they were Alpha testers. Fortunately, I haven't seen any reports claiming "V0.97 Ate My 'Puter", so, we are Much further along with testing than normal for Two Days. I will probably post the 0.97 version for Ubuntu 14.04 soon, along with the Pascal version of 0.97 which seems to need 16.04. So far, the only troubling thing I've seen is 0.97 sometimes reports One less Pulse than the other App. Troubling 'cause I don't see any reason for it unlike the difference in Pulses seen with zi3v. zi3v would sometimes report an additional Pulse, but, it was a Pulse with a Score of Exactly One which the other Apps weren't reporting. ID: 1950989 · Reply Quote

Keith Myers Volunteer tester Send message Joined: 29 Apr 01 Posts: 11744 Credit: 1,160,866,277 RAC: 4,249	Message 1950977 - Posted: 20 Aug 2018, 23:20:30 UTC - in response to Message 1950972. He was able to install the 396.51 drivers after a complete distribution re-install. So whatever was hanging up the driver installation got sorted out on the new installation of the OS. Didn't help that earlier this morning the ppa server and ppa key server were unavailable for an hour. Back up again thankfully. So he got another host up on the 0.97 special app. Looks like he determined that the casita router or switch was buckling under the traffic load, and could only support two systems, because the systems that couldn't connect to the internet in the casita work fine from the house. I told him to get a better unmanaged switch for the casita. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) ID: 1950977 · Reply Quote

Stephen "Heretic" Volunteer tester Send message Joined: 20 Sep 12 Posts: 5384 Credit: 192,787,363 RAC: 1,426	Message 1950974 - Posted: 20 Aug 2018, 22:41:39 UTC - in response to Message 1950935. [quote]@ TBar . . I believe you said you have a version of 0.96 that was compiled on a machine with a compatible software environment to work on this rig? If so is there a link to download it? Stephen ? This machine is running Ubuntu 14.04.1 running the software receive around Noon on Saturday, https://setiathome.berkeley.edu/show_host_detail.php?hostid=6906726 This is a recent result on an Arecibo VLAR, https://setiathome.berkeley.edu/result.php?resultid=6906865500 From Noon on Saturday to Noon on Monday is TWO Days, do you think it is Safe to release Software after TWO days of Testing? I suppose you can argue it's more Safe than releasing it with Zero Days testing, as was done Saturday, but, I don't know of anyone else that releases Software with ZERO Days of testing. Most of the time the testing is measured in Weeks. So, do you think Two days is enough?[/quote] . . Hi TBar, . . I was actually asking about the 0.96 version which I believe you have been running for a while now, sorry about the confusion. As to testing time? No, empirically an app should be tested for more than a couple of days on a single machine before being released to the general public. But if you are asking would I take the gamble and try it? Of course I would :) OK I will have to wait a while longer, but your previous message about getting CUDA92 compatible drivers loaded had me drooling ... Stephen :( ID: 1950974 · Reply Quote

Stephen "Heretic" Volunteer tester Send message Joined: 20 Sep 12 Posts: 5384 Credit: 192,787,363 RAC: 1,426	Message 1950972 - Posted: 20 Aug 2018, 22:27:31 UTC - in response to Message 1950921. Just tried to help Zalster install the 396.51 drivers via the .run file in root recovery terminal. Met with disaster. Run file installation detected errors and was given permission to fix them and it did and then just threw out a bug error and corrupted the file system. System won't load anymore. Will have to reinstall complete again. Only reason this was attempted is that the system complained about installing the 396.51 drivers from the Software Updater. . . That's bad news, it explains why his numbers have dropped off a bit though ... :( Stephen :( ID: 1950972 · Reply Quote

TBar Volunteer tester Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 6,279	Message 1950935 - Posted: 20 Aug 2018, 18:52:23 UTC - in response to Message 1950902. @ TBar . . I believe you said you have a version of 0.96 that was compiled on a machine with a compatible software environment to work on this rig? If so is there a link to download it? Stephen ? This machine is running Ubuntu 14.04.1 running the software receive around Noon on Saturday, https://setiathome.berkeley.edu/show_host_detail.php?hostid=6906726 This is a recent result on an Arecibo VLAR, https://setiathome.berkeley.edu/result.php?resultid=6906865500 From Noon on Saturday to Noon on Monday is TWO Days, do you think it is Safe to release Software after TWO days of Testing? I suppose you can argue it's more Safe than releasing it with Zero Days testing, as was done Saturday, but, I don't know of anyone else that releases Software with ZERO Days of testing. Most of the time the testing is measured in Weeks. So, do you think Two days is enough? ID: 1950935 · Reply Quote

TBar Volunteer tester Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 6,279	Message 1950925 - Posted: 20 Aug 2018, 17:26:48 UTC - in response to Message 1950921. Last modified: 20 Aug 2018, 18:16:05 UTC It sounds as though whatever System was installed simply doesn't like that machine. I do hope you try a Different System this time. There are a large number of choices, you don't have to keep installing the same one expecting different results. There are even point updates that lock the Kernel at a certain level so it never changes beyond that level. All you need to know is which one works best on that hardware. I have the 14.04.1 version installed, the kernel Never changes from 3.13.xx. You could start until you find one that works. I see four different ones here, start with 16.04.1, then maybe 16.04.2, 16.04.3....16.04.4, http://old-releases.ubuntu.com/releases/xenial/ Or maybe start with 16.04.4, if no problems stay there. If problems, try 16.04.3.... ID: 1950925 · Reply Quote

Keith Myers Volunteer tester Send message Joined: 29 Apr 01 Posts: 11744 Credit: 1,160,866,277 RAC: 4,249	Message 1950921 - Posted: 20 Aug 2018, 16:33:22 UTC Just tried to help Zalster install the 396.51 drivers via the .run file in root recovery terminal. Met with disaster. Run file installation detected errors and was given permission to fix them and it did and then just threw out a bug error and corrupted the file system. System won't load anymore. Will have to reinstall complete again. Only reason this was attempted is that the system complained about installing the 396.51 drivers from the Software Updater. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) ID: 1950921 · Reply Quote

Stephen "Heretic" Volunteer tester Send message Joined: 20 Sep 12 Posts: 5384 Credit: 192,787,363 RAC: 1,426	Message 1950908 - Posted: 20 Aug 2018, 14:46:54 UTC - in response to Message 1950889. Load balancing on the cpu/gpu. When I run each of two video cards (gtx 750Ti's) with one core / gpu I get a cpu usage rate of 80~%. When I back it off to say 0.33 of a cpu (I am using app_config.xml for this) per gpu, then it pegs at 100%. I have read someplace that using a core / card under this application is 3%-10% faster. Any guidance? Tom . . Hi Tom, . . When you set the CPU usage in BOINC manager it can have a strange behaviour if you use the wrong values in app_info (or app_config). If you set it to use say 89% of cpu cores in manager, leaving one to support the GPU, but then in app_info tell it to use 1 CPU core per GPU task it might only run crunching on 6 cores instead of 7, but if you set the value in app_config to 0.99 it will happily use all 7 as you intended. If running multiple GPUs and you set it to 75% in manager but set it over 0.5 in app_info it will only use 5 not 6, while setting it to 0.49 or less it will again happily use 6 as intended. I take it you are running CUDA 90 with the default of BS on? Then setting manager to 0.89 and app_info to 0.49 or less will see crunching on 7 cores and the other supporting one task on each GPU. You can expect to find the CPU fully utilised or even perhaps a bit overcommitted in this state. If you run with bs off (using the -nobs option in the commandline section of app_info) then you should heed TBars warnings about severe over commit on CPUs supporting the task on each GPU. It would be wise to leave one core for each GPU plus one for safety to avoid this problem. Using -nobs will gain a noticeable but not huge increase in performance. It can be worth it if your CPU cores are not fast processors when crunching. Stephen . ID: 1950908 · Reply Quote

Stephen "Heretic" Volunteer tester Send message Joined: 20 Sep 12 Posts: 5384 Credit: 192,787,363 RAC: 1,426	Message 1950902 - Posted: 20 Aug 2018, 14:15:58 UTC - in response to Message 1950842. @ TBar . . I believe you said you have a version of 0.96 that was compiled on a machine with a compatible software environment to work on this rig? If so is there a link to download it? Stephen ? ID: 1950902 · Reply Quote

Tom M Volunteer tester Send message Joined: 28 Nov 02 Posts: 4936 Credit: 276,046,078 RAC: 1,048	Message 1950889 - Posted: 20 Aug 2018, 12:47:21 UTC Load balancing on the cpu/gpu. When I run each of two video cards (gtx 750Ti's) with one core / gpu I get a cpu usage rate of 80~%. When I back it off to say 0.33 of a cpu (I am using app_config.xml for this) per gpu, then it pegs at 100%. I have read someplace that using a core / card under this application is 3%-10% faster. Any guidance? Tom "I owe", "I owe", "Its off to work I go" (from a bumper sticker on a smallish Mercedes Benz) (on the back of a Semi Tractor) "If you can read this bumper sticker, I've LOST MY TRAILER!" ID: 1950889 · Reply Quote

Stephen "Heretic" Volunteer tester Send message Joined: 20 Sep 12 Posts: 5384 Credit: 192,787,363 RAC: 1,426	Message 1950879 - Posted: 20 Aug 2018, 11:56:21 UTC - in response to Message 1950833. I am pleased to announce that my I7-3770 machine (#ID: 8564832 is the Linux version) has successfully undergone dual boot upgrade and is happily munching along at the moment. I will have to boot back into Windows to get the rest of its tasks processed out. The "au" tasks make it harder to see that the Gtx 750Ti is running a bunch faster but it is clear that the cpu is running its tasks faster than another box I have which doesn't have the AVX instruction set. I am very pleased, even if I am running a version that is 2-3 releases behind. :) Tom . . I am running zi3v CUDA80 and pleased with the results. Still I would like to get 0.97 running :) . . MORE POWER ! ugh! ugh! ugh! (I watched Home Improvements too much) Stephen :) ID: 1950879 · Reply Quote

Stephen "Heretic" Volunteer tester Send message Joined: 20 Sep 12 Posts: 5384 Credit: 192,787,363 RAC: 1,426	Message 1950878 - Posted: 20 Aug 2018, 11:53:02 UTC @ Raistmer . . Have you succeeded with that portable flashdrive based Linux setup yet? . .I am very distracted trying to sort out my setup woes ATM, I hope you are up and running the way you want very soon. Stephen :) ID: 1950878 · Reply Quote

TBar Volunteer tester Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 6,279	Message 1950874 - Posted: 20 Aug 2018, 10:27:45 UTC - in response to Message 1950869. Last modified: 20 Aug 2018, 10:51:32 UTC Looks like I have some computers to update before the fun starts ... Either that or you can wait for me to post the App compiled in Ubuntu 14.04.1 like I've been doing. The only reason those machines are working now is I compiled all the earlier Apps on 14.04.1, otherwise, you would have had the same problem from the get-go. I've tested the two Petri Apps on my 1050 against the CUDA 9.1 App and the difference on the BLC tasks is around 1 or 2%. So, if you have a Pascal card in 16.04 it will be about 1.5% faster than a Maxwell in Ubuntu 14.04.1. I think most people with the older systems can live with only being 48.5% faster rather than 50% faster than zi3v. Here is the last run after yet another useless tweak. The times are pretty much consistent using different settings; Current WU: blc01_2bit_guppi_58137_29542_HIP45689_0020.26400.818.21.44.80.vlar.wu ---------------------------------------------------------------- Running default app with command :... setiathome_x41p_V0.97_x86_64-pc-linux-gnu_cuda91 -nobs -device 0 gCudaDevProps.multiProcessorCount = 5 Work data buffer for fft results size = 320864256 MallocHost G=67108864 T=33554432 P=18874368 (16) MallocHost tmp_PoTP=16777216 MallocHost tmp_PoTP2=16777216 MallocHost tmp_PoTT=16777216 MallocHost tmp_PoTG=4194304 MallocHost best_PoTP=16777216 MallocHost bestPoTG=4194304 Allocating tmp data buf for unroll 5 MallocHost tmp_smallPoT=524288 MallocHost PowerSpectrumSumMax=1572864 CUDA stream priority range: low 0 and high: -1 GPSF 58.426357 58 94.732101 Gauss: start 58 stop 6 len -52 Sigma > GaussTOffsetStop: 58 > 6 AcIn 16779264 AcOut 33558528 Mallocing blockSums 24576 bytes before async chirp after fft plans bB.....bB............bB............................................................................bB..................................bB..................................................FFtLen : spike gauss autocorr triplet pulse 1: 0 0 0 0 0 2: 0 0 0 0 0 4: 0 0 0 0 0 8: 9 0 0 9 0 16: 19 0 0 19 0 32: 39 0 0 39 39 64: 77 0 0 77 77 128: 153 0 0 153 153 256: 307 0 0 307 307 512: 613 0 0 613 613 1024: 1225 0 0 1225 1225 2048: 2449 0 0 2449 2449 4096: 4897 0 0 4897 4897 8192: 9793 0 0 9793 9793 16384: 2463 0 0 2463 0 32768: 9849 0 0 9849 0 65536: 11817 0 0 11817 0 131072: 47271 0 47272 0 0 Best scores written Out file closed Cuda free done Cuda device reset done Elapsed Time: ....................... 193 seconds ---------------------------------------------------------------- Running app with command : .......... setiV0.97.linux_x64_10x0 -nobs -device 0 gCudaDevProps.multiProcessorCount = 5 Work data buffer for fft results size = 320864256 MallocHost G=67108864 T=33554432 P=18874368 (16) MallocHost tmp_PoTP=16777216 MallocHost tmp_PoTP2=16777216 MallocHost tmp_PoTT=16777216 MallocHost tmp_PoTG=4194304 MallocHost best_PoTP=16777216 MallocHost bestPoTG=4194304 Allocating tmp data buf for unroll 5 MallocHost tmp_smallPoT=524288 MallocHost PowerSpectrumSumMax=1572864 CUDA stream priority range: low 0 and high: -1 GPSF 58.426357 58 94.732101 Gauss: start 58 stop 6 len -52 Sigma > GaussTOffsetStop: 58 > 6 AcIn 16779264 AcOut 33558528 Mallocing blockSums 24576 bytes before async chirp after fft plans bB.....bB............bB............................................................................bB..................................bB..................................................FFtLen : spike gauss autocorr triplet pulse 1: 0 0 0 0 0 2: 0 0 0 0 0 4: 0 0 0 0 0 8: 9 0 0 9 0 16: 19 0 0 19 0 32: 39 0 0 39 39 64: 77 0 0 77 77 128: 153 0 0 153 153 256: 307 0 0 307 307 512: 613 0 0 613 613 1024: 1225 0 0 1225 1225 2048: 2449 0 0 2449 2449 4096: 4897 0 0 4897 4897 8192: 9793 0 0 9793 9793 16384: 2463 0 0 2463 0 32768: 9849 0 0 9849 0 65536: 11817 0 0 11817 0 131072: 47271 0 47272 0 0 Best scores written Out file closed Cuda free done Cuda device reset done Elapsed Time : ...................... 189 seconds Speed compared to default : ......... 102 % ----------------- Comparing results Result : Strongly similar, Q= 100.0% ---------------------------------------------------------------- Running app with command : .......... setiV0.97.multi_sm.linux_X86_64_cuda92 -nobs -device 0 gCudaDevProps.multiProcessorCount = 5 Work data buffer for fft results size = 320864256 MallocHost G=67108864 T=33554432 P=18874368 (16) MallocHost tmp_PoTP=16777216 MallocHost tmp_PoTP2=16777216 MallocHost tmp_PoTT=16777216 MallocHost tmp_PoTG=4194304 MallocHost best_PoTP=16777216 MallocHost bestPoTG=4194304 Allocating tmp data buf for unroll 5 MallocHost tmp_smallPoT=524288 MallocHost PowerSpectrumSumMax=1572864 CUDA stream priority range: low 0 and high: -1 GPSF 58.426357 58 94.732101 Gauss: start 58 stop 6 len -52 Sigma > GaussTOffsetStop: 58 > 6 AcIn 16779264 AcOut 33558528 Mallocing blockSums 24576 bytes before async chirp after fft plans bB.....bB............bB............................................................................bB..................................bB..................................................FFtLen : spike gauss autocorr triplet pulse 1: 0 0 0 0 0 2: 0 0 0 0 0 4: 0 0 0 0 0 8: 9 0 0 9 0 16: 19 0 0 19 0 32: 39 0 0 39 39 64: 77 0 0 77 77 128: 153 0 0 153 153 256: 307 0 0 307 307 512: 613 0 0 613 613 1024: 1225 0 0 1225 1225 2048: 2449 0 0 2449 2449 4096: 4897 0 0 4897 4897 8192: 9793 0 0 9793 9793 16384: 2463 0 0 2463 0 32768: 9849 0 0 9849 0 65536: 11817 0 0 11817 0 131072: 47271 0 47272 0 0 Best scores written Out file closed Cuda free done Cuda device reset done Elapsed Time : ...................... 189 seconds Speed compared to default : ......... 102 % ----------------- Comparing results Result : Strongly similar, Q= 100.0% ---------------------------------------------------------------- Done with blc01_2bit_guppi_58137_29542_HIP45689_0020.26400.818.21.44.80.vlar.wu ==================================================================== Current WU: blc16_2bit_guppi_58185_76028_Dw1_off_0033.2471.1636.22.45.95.vlar.wu ---------------------------------------------------------------- Running default app with command :... setiathome_x41p_V0.97_x86_64-pc-linux-gnu_cuda91 -nobs -device 0 gCudaDevProps.multiProcessorCount = 5 Work data buffer for fft results size = 320864256 MallocHost G=67108864 T=33554432 P=18874368 (16) MallocHost tmp_PoTP=16777216 MallocHost tmp_PoTP2=16777216 MallocHost tmp_PoTT=16777216 MallocHost tmp_PoTG=4194304 MallocHost best_PoTP=16777216 MallocHost bestPoTG=4194304 Allocating tmp data buf for unroll 5 MallocHost tmp_smallPoT=524288 MallocHost PowerSpectrumSumMax=1572864 CUDA stream priority range: low 0 and high: -1 GPSF 112.274147 112 182.123337 Gauss: start 112 stop -48 len -160 Sigma > GaussTOffsetStop: 112 > -48 AcIn 16779264 AcOut 33558528 Mallocing blockSums 24576 bytes before async chirp after fft plans bBbBbBbB.bB...bB...........bB...bBbB..bB......................................................................................................................................bB...................bBP.................FFtLen : spike gauss autocorr triplet pulse 1: 0 0 0 0 0 2: 0 0 0 0 0 4: 0 0 0 0 0 8: 13 0 0 13 0 16: 25 0 0 25 0 32: 51 0 0 51 51 64: 101 0 0 101 101 128: 203 0 0 203 203 256: 407 0 0 407 407 512: 813 0 0 813 813 1024: 1627 0 0 1627 1627 2048: 3255 0 0 3255 3255 4096: 6511 0 0 6511 6511 8192: 13021 0 0 13021 13021 16384: 2463 0 0 2463 0 32768: 9849 0 0 9849 0 65536: 11817 0 0 11817 0 131072: 47271 0 47272 0 0 Best scores written Out file closed Cuda free done Cuda device reset done Elapsed Time: ....................... 228 seconds ---------------------------------------------------------------- Running app with command : .......... setiV0.97.linux_x64_10x0 -nobs -device 0 gCudaDevProps.multiProcessorCount = 5 Work data buffer for fft results size = 320864256 MallocHost G=67108864 T=33554432 P=18874368 (16) MallocHost tmp_PoTP=16777216 MallocHost tmp_PoTP2=16777216 MallocHost tmp_PoTT=16777216 MallocHost tmp_PoTG=4194304 MallocHost best_PoTP=16777216 MallocHost bestPoTG=4194304 Allocating tmp data buf for unroll 5 MallocHost tmp_smallPoT=524288 MallocHost PowerSpectrumSumMax=1572864 CUDA stream priority range: low 0 and high: -1 GPSF 112.274147 112 182.123337 Gauss: start 112 stop -48 len -160 Sigma > GaussTOffsetStop: 112 > -48 AcIn 16779264 AcOut 33558528 Mallocing blockSums 24576 bytes before async chirp after fft plans bBbBbBbB.bB...bB...........bB...bBbB..bB......................................................................................................................................bB...................bBP.................FFtLen : spike gauss autocorr triplet pulse 1: 0 0 0 0 0 2: 0 0 0 0 0 4: 0 0 0 0 0 8: 13 0 0 13 0 16: 25 0 0 25 0 32: 51 0 0 51 51 64: 101 0 0 101 101 128: 203 0 0 203 203 256: 407 0 0 407 407 512: 813 0 0 813 813 1024: 1627 0 0 1627 1627 2048: 3255 0 0 3255 3255 4096: 6511 0 0 6511 6511 8192: 13021 0 0 13021 13021 16384: 2463 0 0 2463 0 32768: 9849 0 0 9849 0 65536: 11817 0 0 11817 0 131072: 47271 0 47272 0 0 Best scores written Out file closed Cuda free done Cuda device reset done Elapsed Time : ...................... 225 seconds Speed compared to default : ......... 101 % ----------------- Comparing results Result : Strongly similar, Q= 100.0% ---------------------------------------------------------------- Running app with command : .......... setiV0.97.multi_sm.linux_X86_64_cuda92 -nobs -device 0 gCudaDevProps.multiProcessorCount = 5 Work data buffer for fft results size = 320864256 MallocHost G=67108864 T=33554432 P=18874368 (16) MallocHost tmp_PoTP=16777216 MallocHost tmp_PoTP2=16777216 MallocHost tmp_PoTT=16777216 MallocHost tmp_PoTG=4194304 MallocHost best_PoTP=16777216 MallocHost bestPoTG=4194304 Allocating tmp data buf for unroll 5 MallocHost tmp_smallPoT=524288 MallocHost PowerSpectrumSumMax=1572864 CUDA stream priority range: low 0 and high: -1 GPSF 112.274147 112 182.123337 Gauss: start 112 stop -48 len -160 Sigma > GaussTOffsetStop: 112 > -48 AcIn 16779264 AcOut 33558528 Mallocing blockSums 24576 bytes before async chirp after fft plans bBbBbBbB.bB...bB...........bB...bBbB..bB......................................................................................................................................bB...................bBP.................FFtLen : spike gauss autocorr triplet pulse 1: 0 0 0 0 0 2: 0 0 0 0 0 4: 0 0 0 0 0 8: 13 0 0 13 0 16: 25 0 0 25 0 32: 51 0 0 51 51 64: 101 0 0 101 101 128: 203 0 0 203 203 256: 407 0 0 407 407 512: 813 0 0 813 813 1024: 1627 0 0 1627 1627 2048: 3255 0 0 3255 3255 4096: 6511 0 0 6511 6511 8192: 13021 0 0 13021 13021 16384: 2463 0 0 2463 0 32768: 9849 0 0 9849 0 65536: 11817 0 0 11817 0 131072: 47271 0 47272 0 0 Best scores written Out file closed Cuda free done Cuda device reset done Elapsed Time : ...................... 225 seconds Speed compared to default : ......... 101 % ----------------- Comparing results Result : Strongly similar, Q= 100.0% ---------------------------------------------------------------- Done with blc16_2bit_guppi_58185_76028_Dw1_off_0033.2471.1636.22.45.95.vlar.wu ==================================================================== Current WU: blc3_2bit_guppi_57432_28897_HIP57494_OFF_0014.14006.416.18.27.18.vlar.wu ---------------------------------------------------------------- Running default app with command :... setiathome_x41p_V0.97_x86_64-pc-linux-gnu_cuda91 -nobs -device 0 gCudaDevProps.multiProcessorCount = 5 Work data buffer for fft results size = 320864256 MallocHost G=67108864 T=33554432 P=18874368 (16) MallocHost tmp_PoTP=16777216 MallocHost tmp_PoTP2=16777216 MallocHost tmp_PoTT=16777216 MallocHost tmp_PoTG=4194304 MallocHost best_PoTP=16777216 MallocHost bestPoTG=4194304 Allocating tmp data buf for unroll 5 MallocHost tmp_smallPoT=524288 MallocHost PowerSpectrumSumMax=1572864 CUDA stream priority range: low 0 and high: -1 GPSF 255.441376 255 413.903625 Gauss: start 255 stop -191 len -446 Sigma > GaussTOffsetStop: 255 > -191 AcIn 16779264 AcOut 33558528 Mallocing blockSums 24576 bytes before async chirp after fft plans bBbB....bB...bB.......................bB...bB...................................................................................bBP................................bBP.pP..pP.....pP....pP.pP...........pP..pP............pP.........pP.........pP Elapsed Time: ....................... 272 seconds ---------------------------------------------------------------- Running app with command : .......... setiV0.97.linux_x64_10x0 -nobs -device 0 gCudaDevProps.multiProcessorCount = 5 Work data buffer for fft results size = 320864256 MallocHost G=67108864 T=33554432 P=18874368 (16) MallocHost tmp_PoTP=16777216 MallocHost tmp_PoTP2=16777216 MallocHost tmp_PoTT=16777216 MallocHost tmp_PoTG=4194304 MallocHost best_PoTP=16777216 MallocHost bestPoTG=4194304 Allocating tmp data buf for unroll 5 MallocHost tmp_smallPoT=524288 MallocHost PowerSpectrumSumMax=1572864 CUDA stream priority range: low 0 and high: -1 GPSF 255.441376 255 413.903625 Gauss: start 255 stop -191 len -446 Sigma > GaussTOffsetStop: 255 > -191 AcIn 16779264 AcOut 33558528 Mallocing blockSums 24576 bytes before async chirp after fft plans bBbB....bB...bB.......................bB...bB...................................................................................bBP................................bBP.pP..pP.....pP....pP.pP...........pP..pP............pP.........pP.........pP Elapsed Time : ...................... 268 seconds Speed compared to default : ......... 101 % ----------------- Comparing results Result : Strongly similar, Q= 100.0% ---------------------------------------------------------------- Running app with command : .......... setiV0.97.multi_sm.linux_X86_64_cuda92 -nobs -device 0 gCudaDevProps.multiProcessorCount = 5 Work data buffer for fft results size = 320864256 MallocHost G=67108864 T=33554432 P=18874368 (16) MallocHost tmp_PoTP=16777216 MallocHost tmp_PoTP2=16777216 MallocHost tmp_PoTT=16777216 MallocHost tmp_PoTG=4194304 MallocHost best_PoTP=16777216 MallocHost bestPoTG=4194304 Allocating tmp data buf for unroll 5 MallocHost tmp_smallPoT=524288 MallocHost PowerSpectrumSumMax=1572864 CUDA stream priority range: low 0 and high: -1 GPSF 255.441376 255 413.903625 Gauss: start 255 stop -191 len -446 Sigma > GaussTOffsetStop: 255 > -191 AcIn 16779264 AcOut 33558528 Mallocing blockSums 24576 bytes before async chirp after fft plans bBbB....bB...bB.......................bB...bB...................................................................................bBP................................bBP.pP..pP.....pP....pP.pP...........pP..pP............pP.........pP.........pP Elapsed Time : ...................... 267 seconds Speed compared to default : ......... 101 % ----------------- Comparing results Result : Strongly similar, Q= 100.0% ---------------------------------------------------------------- Done with blc3_2bit_guppi_57432_28897_HIP57494_OFF_0014.14006.416.18.27.18.vlar.wu ID: 1950874 · Reply Quote

Raistmer Volunteer developer Volunteer tester Send message Joined: 16 Jun 01 Posts: 6242 Credit: 106,370,077 RAC: 275	Message 1950872 - Posted: 20 Aug 2018, 9:46:03 UTC Even didn't finish logon process yet but already got " Failed to update" message. Looks like even didn't changed OS at all, same nasty things as in Windows here. Anyway, I don't need to like Ubuntu, just use it on purpose... And purpose was to create Portable Linux bootable flash to test Linux-based software. Will check if it can boot on PC with 1050Ti being created on netbook... SETI apps news We're not gonna fight them. We're gonna transcend them. ID: 1950872 · Reply Quote

Brent Norman Volunteer tester Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 1,893	Message 1950869 - Posted: 20 Aug 2018, 8:44:53 UTC - in response to Message 1950849. Yea, I'm not sure where Petri came along with gcc 5.2.1, I can only get mine up to 4.8.5 ... If that is even the problem. I'm having no luck with Ubuntu 14 or Mint 17. Uggg that is most of my computers. Mint 18 with the muti app, seem to be fine so far on my i7-960 with 980/780Ti/750Ti Arecibo shorties on the 980 @ 58s, and 780Ti @ 50s ... impressive. Looks like I have some computers to update before the fun starts ... ID: 1950869 · Reply Quote

TBar Volunteer tester Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 6,279	Message 1950849 - Posted: 20 Aug 2018, 3:43:27 UTC - in response to Message 1950847. I'm not even sure what the problem is. However it does appear your GCC is from 15.10; gcc (Ubuntu 5.2.1-22ubuntu2) 5.2.1 20151010 Again, I have no idea if that's a problem. Now that I think about it, I wonder what happened to that problem with the AKv8 folder and GCC 5.2+....it just faded away. ID: 1950849 · Reply Quote

petri33 Volunteer tester Send message Joined: 6 Jun 02 Posts: 1668 Credit: 623,086,772 RAC: 354	Message 1950847 - Posted: 20 Aug 2018, 3:23:03 UTC I must have updated my gnu compiler at some time. Would it be of any help if i built the app with -static-libstdc++ ? To overcome Heisenbergs: "You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones ID: 1950847 · Reply Quote

TBar Volunteer tester Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 6,279	Message 1950842 - Posted: 20 Aug 2018, 2:57:27 UTC - in response to Message 1950834. Well, 18.04 is only 6 months old, and I usually wait until Systems are 1 year old before moving to them. I haven't heard of any nasty problems with 18.04, so, it might be OK. I'm going to wait a little longer though. I really don't understand why the Apps aren't working with the Kernel 4.4 systems, the way it usually works is you need equal to or higher than the system that compiled the App. The only thing that looks a little strange is Petri's GCC version seems a little high. From trying to compile Apps in the AKv8 folder we found that GCC 4.9 in Ubuntu 15.04 worked, but GCC 5.2 in 15.10 had problems. So, it would appear GCC 5.21 is a little higher than you would expect for Petri's Ubuntu 14.04.4. Or, maybe they just updated it since the last time I looked. The version of GCC in Ubuntu 14.04.1 that I used says 4.8.4. Anyway, the App I compiled is working in 14.04.1 with Kernel 3.13. ID: 1950842 · Reply Quote

Keith Myers Volunteer tester Send message Joined: 29 Apr 01 Posts: 11744 Credit: 1,160,866,277 RAC: 4,249	Message 1950836 - Posted: 20 Aug 2018, 2:04:48 UTC - in response to Message 1950834. Either 16.04 or 18.04. 18.04 has a future till 2024. There are a lot of nice new features in 18.04. I've been happy with it. Always a bit jarring to pull up a chair to one of my first systems with 16.04 still on it. I've become much more comfortable and used to 18.04 by now. In fact, my first Linux system just popped up a upgrade notice saying click here for the latest and greatest 18.04. I put it off till the end of the contest. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) ID: 1950836 · Reply Quote

Stephen "Heretic" Volunteer tester Send message Joined: 20 Sep 12 Posts: 5384 Credit: 192,787,363 RAC: 1,426	Message 1950834 - Posted: 20 Aug 2018, 1:40:23 UTC - in response to Message 1950819. This system is running release 131 which is pretty much up to date. This system is the one on which I foolishly installed the repository version of BOINC. Yes, both those systems are showing Kernel 4.4. They Should work. There might be something else going on there. Did you try the Other App? The one that is listed in the app_info that Keith and myself used? You should give it a try. I've told you before, it should be simple to change the Repository BOINC over to Berkeley BOINC. The boinc-client folder is basically the Berkeley BOINC without the BOINC Apps installed. The biggest problem is you need to use a clean install of Ubuntu so you don't have any configuration files from the Repository BOINC in your system. All you really need to save your HOST ID is the client_state.xml file, the rest of the folder is gravy. You just need to ZIP the boinc-client folder and save it for the New System before you install the New System. . . It all seems to come down to the c++ library libstdc++6. That library is installed on this host but those two extensions are missing which 0.97 needs, GLIBCXX_3.4.20 and 21. I did a re-install of the library with Synaptic but those later extensions are still MIA. . . I have some errands and then I'll get lunch and come back for another try. Which version of Ubuntu OS should I go for ... 18.04? Stephen ? ID: 1950834 · Reply Quote

©2020 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.