SETI@home v8 beta to begin on Tuesday
Message boards :
News :
SETI@home v8 beta to begin on Tuesday
Message board moderation
Previous · 1 . . . 33 · 34 · 35 · 36 · 37 · 38 · 39 . . . 99 · Next
| Author | Message |
|---|---|
 Eric J Korpela Send message Joined: 15 Mar 05 Posts: 1547 Credit: 27,183,456 RAC: 0 
|
Finally got around to releasing a version compiled with -fno-stack-protector. We'll see it if works. 
ID: 56277 ·  |
Raistmer Send message Joined: 18 Aug 05 Posts: 2423 Credit: 15,878,738 RAC: 0 
|
Quite interesting situation on my another NV host: SETI@home v8 8.06 windows_intelx86 (opencl_nvidia_cc1_sah) Number of tasks completed 9 Max tasks per day 42 Number of tasks today 0 Consecutive valid tasks 9 Average processing rate 40.83 GFLOPS Average turnaround time 0.56 days SETI@home v8 8.01 windows_intelx86 (cuda23) Number of tasks completed 307 Max tasks per day 364 Number of tasks today 8 Consecutive valid tasks 125 Average processing rate 32.58 GFLOPS Average turnaround time 0.39 days SETI@home v8 8.01 windows_intelx86 (cuda32) Number of tasks completed 6 Max tasks per day 40 Number of tasks today 0 Consecutive valid tasks 7 Average processing rate 33.53 GFLOPS Average turnaround time 0.19 days Yes, finally cuda23 should be the best for GSO9600+GT9400... but interesting how BOINC could anticipate it with such statistics?... News about SETI opt app releases: https://twitter.com/Raistmer ID: 56300 · |
|
Send message Joined: 25 Jan 16 Posts: 5 Credit: 147,975 RAC: 0
|
I have an Apple iMac 11,1 (late 2009) with a ATI Radeon HD 4850. The first GPU work unit seemed to be processing fine, but when I quit and rebooted, it then error'd out, and every GPU work unit downloaded since has had an immediate computation error. For example: Wed Jan 27 04:41:38 2016 | SETI@home Beta Test | Started download of 16no11aa.29372.17678.10.44.194 Wed Jan 27 04:41:40 2016 | SETI@home Beta Test | Finished download of 16no11aa.29372.17678.10.44.194 Wed Jan 27 04:41:40 2016 | SETI@home Beta Test | [coproc] Assigning ATI instance 0 to 16no11aa.29372.17678.10.44.194_1 Wed Jan 27 04:41:40 2016 | SETI@home Beta Test | Starting task 16no11aa.29372.17678.10.44.194_1 Wed Jan 27 04:41:41 2016 | SETI@home Beta Test | Computation for task 16no11aa.29372.17678.10.44.194_1 finished Wed Jan 27 04:41:41 2016 | SETI@home Beta Test | Output file 16no11aa.29372.17678.10.44.194_1_r794814907_0 for task 16no11aa.29372.17678.10.44.194_1 absent Any GPU suggestions? It does run the CPU SETI@home V8 8.05 seems to work fine. ID: 56306 · |
 Send message Joined: 18 Jan 06 Posts: 1038 Credit: 18,734,730 RAC: 0 
|
I have an Apple iMac 11,1 (late 2009) with a ATI Radeon HD 4850. The first GPU work unit seemed to be processing fine, but when I quit and rebooted, it then error'd out, and every GPU work unit downloaded since has had an immediate computation error. For example: Next version of the GPU app will solve that error. (hopefully) _\|/_ U r s ID: 56309 · |
Shawn Rothermund Send message Joined: 11 Sep 13 Posts: 22 Credit: 457,141 RAC: 0
|
I was just wondering about beta tonight my daily driver is doing both beta and main and is getting cuda 50,42,and 32 from main and v8.01 from beta are these leftover from beta or is 8.01 going to be the main release for cuda. ME AND MY BOY LOOKING FOR E.T. 
ID: 56310 · |
Gary Charpentier Send message Joined: 9 Apr 07 Posts: 1701 Credit: 4,622,751 RAC: 0
|
Different computer same problem, interesting ... Started with: 25-Jan-2016 01:09:59 [SETI@home Beta Test] Computation for task 16no11aa.3901.12357.6.40.155_0 finished 25-Jan-2016 01:10:01 [SETI@home Beta Test] Started upload of 16no11aa.3901.12357.6.40.155_0_0 25-Jan-2016 01:11:18 [---] Project communication failed: attempting access to reference site 25-Jan-2016 01:11:18 [SETI@home Beta Test] Temporarily failed upload of 16no11aa.3901.12357.6.40.155_0_0: connect() failed 25-Jan-2016 01:11:18 [SETI@home Beta Test] Backing off 00:03:57 on upload of 16no11aa.3901.12357.6.40.155_0_0 25-Jan-2016 01:11:20 [---] Internet access OK - project servers may be temporarily down. And a couple more of the above, then 25-Jan-2016 08:06:42 [SETI@home Beta Test] Started upload of 16no11aa.3901.12357.6.40.155_0_0 25-Jan-2016 08:06:47 [SETI@home Beta Test] Error reported by file upload server: can't open file 25-Jan-2016 08:06:47 [SETI@home Beta Test] Temporarily failed upload of 16no11aa.3901.12357.6.40.155_0_0: transient upload error 25-Jan-2016 08:06:47 [SETI@home Beta Test] Backing off 00:11:00 on upload of 16no11aa.3901.12357.6.40.155_0_0 25-Jan-2016 08:18:20 [SETI@home Beta Test] Started upload of 16no11aa.3901.12357.6.40.155_0_0 25-Jan-2016 08:18:20 [SETI@home Beta Test] Started upload of 16no11aa.3901.12766.6.40.67_0_0 25-Jan-2016 08:18:20 [SETI@home Beta Test] Started upload of 16no11aa.3901.12766.6.40.74_2_0 25-Jan-2016 08:18:25 [SETI@home Beta Test] Error reported by file upload server: can't open file 25-Jan-2016 08:18:25 [SETI@home Beta Test] Temporarily failed upload of 16no11aa.3901.12357.6.40.155_0_0: transient upload error 25-Jan-2016 08:18:25 [SETI@home Beta Test] Backing off 00:25:05 on upload of 16no11aa.3901.12357.6.40.155_0_0 25-Jan-2016 08:18:32 [SETI@home Beta Test] Finished upload of 16no11aa.3901.12766.6.40.67_0_0 25-Jan-2016 08:18:32 [SETI@home Beta Test] Finished upload of 16no11aa.3901.12766.6.40.74_2_0 25-Jan-2016 08:20:15 [SETI@home Beta Test] Started upload of 16no11aa.3901.12357.6.40.254_1_0 25-Jan-2016 08:20:32 [SETI@home Beta Test] Finished upload of 16no11aa.3901.12357.6.40.254_1_0 A bunch of the above with "can't open file" while other work has no issue... Then we get 26-Jan-2016 23:48:17 [SETI@home Beta Test] Started upload of 16no11aa.3901.12357.6.40.155_0_0 26-Jan-2016 23:48:18 [SETI@home Beta Test] Finished upload of 16no11aa.3901.12357.6.40.155_0_0 So it appears as if the "can't open file" is not an echo from this local machine and is a server side error.  
ID: 56312 · |
|
Send message Joined: 31 Dec 14 Posts: 6 Credit: 244,223 RAC: 0 
|
yesterday I finally had some time to fix drivers & BOINC to work on Ubuntu, check info here. but unfortunately, there was no jobs on SETi@home or BETAs...can someone tell me if there are in plan to be made or just finished all BETAs, so SETi@home hasn't started or the 3rd possibility? ;)  non-profit org. Play4Life in Zagreb, Croatia, EU ID: 56313 · |
 Send message Joined: 14 Feb 13 Posts: 606 Credit: 588,843 RAC: 0 |
I was just wondering about beta tonight my daily driver is doing both beta and main and is getting cuda 50,42,and 32 from main and v8.01 from beta are these leftover from beta or is 8.01 going to be the main release for cuda. Version numbers were reset to 8.00 for the transfer to main. if you look at the app processing it (e.g. in stderr) you'll see that it's both x41zi. So the CPU beta 8.04 apps are main 8.00 apps and cuda beta 8.01 arrived at main as 8.00, too. A person who won't read has no advantage over one who can't read. (Mark Twain) ID: 56319 · |
|
Send message Joined: 14 Feb 13 Posts: 606 Credit: 588,843 RAC: 0 |
yesterday I finally had some time to fix drivers & BOINC to work on Ubuntu, check info here. I don't quite understand you, but it does happen that beta and/or main run out of work, usually due to server issues. In times of high demand (e.g. after maintenance) you may also experience task shortage - look at the Server status page, if the ready to send buffer is low, you'll struggle to get work. Else check the log to see if you are getting any error messages. If you need help sorting out why you are not getting work, please open a dedicated thread. You'll need to enable a few log flags (starting with work_fetch_debug) and post a log cycle. edit: also sched_op_debug will probably be needed. A person who won't read has no advantage over one who can't read. (Mark Twain) ID: 56320 · |
|
Send message Joined: 2 Jul 13 Posts: 505 Credit: 5,019,318 RAC: 0
|
I've been battling a couple of these on Main for a few days. After running them a couple of times, and even turning them into ghosts and then resurrecting, they refused to upload when run as a GPU task. The last time I resurrected them as CPU tasks and one Finally uploaded and finished. I can't remember the second one, but this was the one that caused most of the grief, http://setiathome.berkeley.edu/result.php?resultid=4677951222 Thankfully it appears as though it's finally gone. Hopefully the other one will finally finish, they were both giving the same error, "can't open file". Different computer same problem, interesting ... ID: 56325 · |
|
Eric J Korpela Send message Joined: 15 Mar 05 Posts: 1547 Credit: 27,183,456 RAC: 0
|
For some reason the file_upload_handler was creating files in read only mode. So an incomplete upload would cause this error on retries. I had to manually reset the permissions on the files in the upload directory in order to fix it.
ID: 56326 · |
|
Send message Joined: 3 Jan 07 Posts: 1451 Credit: 3,272,268 RAC: 0 
|
For some reason the file_upload_handler was creating files in read only mode. So an incomplete upload would cause this error on retries. I had to manually reset the permissions on the files in the upload directory in order to fix it. Ah - that's finally tickled a few brain cells into life. Have a look at ce00365f866e97ae48b1f41ba9608dff4c459ffe - and maybe have a chat with David about the implications for incomplete/resumed uploads. Reference also http://lists.ssl.berkeley.edu/pipermail/boinc_dev/2015-October/022035.html It looks as if Christian Beer (who originated the suggestion) hit similar problems at the very end of October (http://lists.ssl.berkeley.edu/pipermail/boinc_dev/2015-October/022055.html), but there the trail goes cold - I can't find any more email discussion or further (related) code commits. Are particular client versions implicated in the cases which have surfaced here? ID: 56330 · |
|
Gary Charpentier Send message Joined: 9 Apr 07 Posts: 1701 Credit: 4,622,751 RAC: 0
|
For some reason the file_upload_handler was creating files in read only mode. So an incomplete upload would cause this error on retries. I had to manually reset the permissions on the files in the upload directory in order to fix it. FYI, I have not seen the issue at Atlas@home, VirtualLHC@home, or Einstein@home and I know that I get frequent incomplete uploads on the first two. Might check which version of the server they are running and what the configuration switches they are using are. Perhaps it is a simple configuration issue.
ID: 56339 · |
|
Send message Joined: 2 Jul 13 Posts: 505 Credit: 5,019,318 RAC: 0
|
Just a comparison between the Mac OpenCL App and the Mac CUDA App on one Laptop. Here are the OpenCL results, http://setiathome.berkeley.edu/results.php?hostid=7366840&offset=360. Most of the nVidia tasks are receiving Inconclusive with a couple Invalids. Here are the CUDA results, http://setiathome.berkeley.edu/results.php?hostid=7366840&offset=240. Not only are the results over twice as fast, they are being validated on the first attempt. Quite a difference. That machine is using the CUDA65 App from here, http://www.arkayn.us/forum/index.php?topic=191.msg4411#msg4411 ID: 56346 · |
|
Raistmer Send message Joined: 18 Aug 05 Posts: 2423 Credit: 15,878,738 RAC: 0
|
Just a comparison between the Mac OpenCL App and the Mac CUDA App on one Laptop. Here are the OpenCL results, http://setiathome.berkeley.edu/results.php?hostid=7366840&offset=360. Most of the nVidia tasks are receiving Inconclusive with a couple Invalids. Here are the CUDA results, http://setiathome.berkeley.edu/results.php?hostid=7366840&offset=240. Not only are the results over twice as fast, they are being validated on the first attempt. Quite a difference. That machine is using the CUDA65 App from here, http://www.arkayn.us/forum/index.php?topic=191.msg4411#msg4411 Would be good to create similar table I did for Linux & Windows: driver compatibility sheet. Seems this one: 10.8.14 310.42.15f01 incompatible one. OS X uses quite different driver numeration scheme so separate investigation required. News about SETI opt app releases: https://twitter.com/Raistmer ID: 56349 · |
|
Send message Joined: 2 Jul 13 Posts: 505 Credit: 5,019,318 RAC: 0
|
Just a comparison between the Mac OpenCL App and the Mac CUDA App on one Laptop. Here are the OpenCL results, http://setiathome.berkeley.edu/results.php?hostid=7366840&offset=360. Most of the nVidia tasks are receiving Inconclusive with a couple Invalids. Here are the CUDA results, http://setiathome.berkeley.edu/results.php?hostid=7366840&offset=240. Not only are the results over twice as fast, they are being validated on the first attempt. Quite a difference. That machine is using the CUDA65 App from here, http://www.arkayn.us/forum/index.php?topic=191.msg4411#msg4411 The major difference is nVidia has separated the CUDA driver from the Video driver in OSX. The OSX Video driver is built into the OS and can change with each OS version. It appears the OpenCL driver is also bundled with the OS and can also change with system version. Looking at the optional nVidia Video WEB driver you can see it also changes with each OS version. Go here, http://www.nvidia.com/Download/index.aspx?lang=en-us, select GeForce/GeForce 600 series/GeForce GTX 680 and look at the offered Drivers for OSX Operating Systems. Then look at the current CUDA driver, http://www.nvidia.com/object/macosx-cuda-7.5.25-driver.html, Supported MAC OS X: 10.11.x, 10.10.x, 10.9.x. Quite different than a different driver for each OS point version. ID: 56350 · |
|
Raistmer Send message Joined: 18 Aug 05 Posts: 2423 Credit: 15,878,738 RAC: 0
|
Anyway, compatibility table required. Just add OS X version then along with driver number. Also, why OS X OpenCL NV performance vs its CUDA performance so different regarding other OSes would be good to establish (there is no 2-fold performance differencies under other OSes). News about SETI opt app releases: https://twitter.com/Raistmer ID: 56352 · |
|
Send message Joined: 2 Jul 13 Posts: 505 Credit: 5,019,318 RAC: 0
|
Anyway, compatibility table required. Just add OS X version then along with driver number. It seems it depends on the Device. In OSX 10.10/11 with a GTX 660 & GTX 750Ti the difference between OpenCL & CUDA is about a 40% speed gain with CUDA on the Lower Angle Range tasks, the VHAR tasks are about the same. Some Devices are just a miss as can be seen by the GT 650M having the same problem with the older OSX and Older OpenCL App, http://setiathome.berkeley.edu/result.php?resultid=4523689036; System: Darwin x86_64 Kernel: 15.0.0 Driver version: 10.6.47 310.42.05f01 OpenCL-kernels filename : MultiBeam_Kernels_r2936.cl ar=0.410203 Run time: 1 hours 17 min 54 sec validation inconclusive ID: 56353 · |
|
Send message Joined: 18 Jan 06 Posts: 1038 Credit: 18,734,730 RAC: 0
|
inconclusive is no "miss" its undecided between two or more hosts.Anyway, compatibility table required. Just add OS X version then along with driver number. Additional your exemplary OS X host with the GT650M ( a mobile device! ) has assembled some inconclusives with the CUDA app as well. http://setiathome.berkeley.edu/results.php?hostid=7366840&offset=0&show_names=0&state=3&appid=29 _\|/_ U r s ID: 56360 · |
|
Send message Joined: 2 Jul 13 Posts: 505 Credit: 5,019,318 RAC: 0
|
inconclusive is no "miss" its undecided between two or more hosts.Anyway, compatibility table required. Just add OS X version then along with driver number. This was the question I was responding to, "Also, why OS X OpenCL NV performance vs its CUDA performance so different regarding other OSes would be good to establish (there is no 2-fold performance differencies under other OSes)." The 'Miss' is that a good number of the Mobile GPUs are running around Half Speed from where they should be running. This is my Old GTS 250 running the CUDA42 App, http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=71141. The Much Newer X50 series GPUs should be running around the same, such as this 650M, http://setiathome.berkeley.edu/results.php?hostid=7366840&offset=240. Check the times here on Beta, they are about the same as the previously quoted GT 650M with the OpenCL App, much slower than my old GTS250. http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=58196&offset=20 http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=76929&offset=60 http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=76929&offset=60 Do you really think a fairly New GT 750M should be running that Slow? I'd call that a Miss. Here's one where the iGPU is running Faster than the NV750M, http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=69829&offset=60 ??? There are many others... ID: 56361 · |
©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.