Message boards :
Number crunching :
GUPPI Rescheduler for Linux and Windows - Move GUPPI work to CPU and non-GUPPI to GPU
Message board moderation
Previous · 1 . . . 23 · 24 · 25 · 26 · 27 · 28 · 29 . . . 37 · Next
Author | Message |
---|---|
528 Send message Joined: 23 Nov 99 Posts: 89 Credit: 338,541,253 RAC: 367 |
I've been using Qopt manually, both in unattended(-u -l) and manual mode with no problem when I run them from my desktop. I finally got around to setting up a Task Manager task and have an issue. When I run Qopt.exe with the parameters -u -l, my graphics card is not recognized on the client restart. I have to shut down the client and restart it to see the graphics card. Any TM gurus with some thoughts? A Living example of Artificial Intelligence. |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1853 Credit: 268,616,081 RAC: 1,349 |
I've been using Qopt manually, both in unattended(-u -l) and manual mode with no problem when I run them from my desktop. I finally got around to setting up a Task Manager task and have an issue. When I run Qopt.exe with the parameters -u -l, my graphics card is not recognized on the client restart. I have to shut down the client and restart it to see the graphics card. Any TM gurus with some thoughts?Just to be clear, you are talking about Task Scheduler? Just to check, did you read the notes in the read_me about task scheduler setups? In case not, or for anyone else: ========== SCHEDULING ---------- QOpt has been tested in conjunction with Windows Task Scheduler. In order for QOpt to operate successfully in this environment, the following steps are recommended: 1. Enter Task Scheduler: Control Panel --> Administrative Tools --> Task Scheduler 2. Create Task, with following conditions and parameters: --> General --> When running, use following user account: --> (normal account name with privileges to access QOpt location on network) --> Run only when user is logged on (mandatory) --> Triggers --> As desired. Example: Start at 12:02:00 AM, repeat every 4 hours --> Actions --> Start a program. Example: C:\Scheduling\QOpt.exe --> Specify command line options. Example: -u -l --> Specify directory to run it in (where QOPt is located) 3. Save task and enable, then Run to test schedule and correct if needed. If Qopt can be run fom a command line with options and succeeds, but fails under Task Scheduler, it is most likely an error with access rights based on setup step 2. above that must be corrected What you're describing sounds like some of the weirdness I was seeing when I didn't have "Run only when user is logged on" selected. |
528 Send message Joined: 23 Nov 99 Posts: 89 Credit: 338,541,253 RAC: 367 |
What you're describing sounds like some of the weirdness I was seeing when I didn't have "Run only when user is logged on" selected. That fixed it for me! Thanks! A Living example of Artificial Intelligence. |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1853 Credit: 268,616,081 RAC: 1,349 |
What you're describing sounds like some of the weirdness I was seeing when I didn't have "Run only when user is logged on" selected. Great! |
Stephen "Heretic" Send message Joined: 20 Sep 12 Posts: 5557 Credit: 192,787,363 RAC: 628 |
. . @ BILBG . . Hey there, . . I finally got up the nerve to try your method for recovering ghosted tasks. I was pleasantly surprised that I didn't stuff it up and I successfully recovered 21 "Lost tasks". The only disappointment was that even though I cleared my GPU cache and there were almost 250 ghosts in my queues, it was only 21 tasks. I guess I will have to repeat the procedure several more times to get back to accurate queue numbers. Stephen . |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
. . @ BILBG I don't remember posting "method for recovering ghosted tasks", this should be someone else ;) Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
Stephen "Heretic" Send message Joined: 20 Sep 12 Posts: 5557 Credit: 192,787,363 RAC: 628 |
. . @ BILBG . . Oops! . . Sorry about that, in my mind it was you, but thanks for the other advice you have provided anyway :) Stephen PS: . . I did a little back tracking and it was Jeff Buck to whom I owe the thank you for this one :) . . |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Just a quick note. Finally got Kaspersky Tech Support to fix the false positive on QOpt.7z and QOpt.zip on Jim's FTP server files. Only took 2 weeks of daily email exchanges. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1853 Credit: 268,616,081 RAC: 1,349 |
Just a quick note. Finally got Kaspersky Tech Support to fix the false positive on QOpt.7z and QOpt.zip on Jim's FTP server files. Only took 2 weeks of daily email exchanges. Hey Keith, thanks for the update! Good grins, but I'm afraid ultimately useless. Chances are, next time I recompile they'll be back. Just nature of the beast, I'm afraid. Case in point, Windoze Defender (sic) has been blessing both the x32 and x64 versions of 1.02k, and then today just decided to wipe them off the drive as malware. Go figure .... Good news is, I don't have any open bugs, and haven't got any plans to attack any of the items I have on the wish list, so there's a pretty good chance I won't be recompiling any time soon. Now, of course, if someone else comes up with something they can't live without, that's in my power to address, I'll sure look at it. But I don't know what it would be ... |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
History of QOpt_1_02k.7z scans: https://www.virustotal.com/en/file/6ed1837aacae8610217a7dfcebda4d32c341aca892a1df11ad8cf2af3ad34b48/analysis/1477127000/ https://www.virustotal.com/en/file/6ed1837aacae8610217a7dfcebda4d32c341aca892a1df11ad8cf2af3ad34b48/analysis/1477646040/ https://www.virustotal.com/en/file/6ed1837aacae8610217a7dfcebda4d32c341aca892a1df11ad8cf2af3ad34b48/analysis/1477646049/ Note: How to get the info about old scans: - the last part of the link 1477646049/ is time in seconds - UNIX TimeStamp "(based on seconds since standard epoch of 1/1/1970)" http://www.onlineconversion.com/unix_time.htm For the above links it does not need to be exact - you can see that to go to the previous scan 1477646040/ (which is 3-4 days ago) I "deleted" only 9 seconds from the last scan. To go 5-6 days back I used Windows Calculator to subtract 86400 (24*3600) seconds 5-6 times: 1477646049 - 86400 [Enter] [Enter] [Enter] ... Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
Case in point, Windoze Defender (sic) has been blessing both the x32 and x64 versions of 1.02k, and then today just decided to wipe them off the drive as malware. Go figure .... So "Add an exclusion to Windows Defender" don't help?: https://support.microsoft.com/en-us/instantanswers/64495205-6ddb-4da1-8534-1aeaf64c0af8/add-an-exclusion-to-windows-defender http://answers.microsoft.com/en-us/windows/forum/windows_8-security/how-to-stop-windows-defender-from-deleting-a/f4e3641a-5598-495d-b2a5-bd6ead226071 And you are right that "ask me first what to do" "is not an option in Windows Defender on Windows 10" and "Ah that sucks" http://answers.microsoft.com/en-us/protect/forum/protect_defender-protect_start/how-do-i-set-windows-defender-in-windows-10-to-ask/2187e072-6bed-4d2d-bc9e-12b734148332 There is a hope for some hidden Settings if you want to search: https://www.google.com/#q=windows+defender+registry+key  - ALF - "Find out what you don't do well ..... then don't do it!" :)  |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1853 Credit: 268,616,081 RAC: 1,349 |
Case in point, Windoze Defender (sic) has been blessing both the x32 and x64 versions of 1.02k, and then today just decided to wipe them off the drive as malware. Go figure .... Problem is that, as far as I can tell, "real time protection" doesn't respect the exclusion list. First time I did a compile and had a false positive, probably a month ago, I added exclusions for the specific file, the working directories and finally the entire drive. Thing is, once the definitions get updated and it does a detect, you have to do a restore and allow on something that should never have been touched. But when the definitions then update again, its possible but not definite, that you'll get hit again. Also, if you update the file, the exclusions are still listed but no longer valid until re-added, even if the update is the identical copy of the file that was being replaced due to deletion. All of this would probably not be an issue if I was keeping a copy on each machine. I think that using the network drive, the issues with UNC drive definitions, and having 5 machines using it across the network, contribute to this. But since that's the functionality I tout for this, it's incumbent on me to "eat my own dog food" as it were. |
Stubbles Send message Joined: 29 Nov 99 Posts: 358 Credit: 5,909,255 RAC: 0 |
Hey fellow Optimizers! Who's doing what with the Andromeda Galaxy tasks (aka Messier031)? From the few I've processed on my 2 rigs (cuz I still keep a cache of close to 1000 tasks for my tests and script devs), they seem to be even better on the CPU than regular Guppis (blc...guppi...vlars), and just as bad on my nVidia GPUs (2*750Ti & 1 GTX1060). On my 2 Xeon W3550s, the MESSIERs are all running under 2.5hrs while the regular guppis can sometimes take up to 3hrs. I haven't processed enough for good solid stats yet though. Has anyone processed at least 100 on CPU cores to confirm my preliminary observation? Cheers, RobG :-) |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Haven't processed anywhere near 100 on CPU. Mine are processing just like a normal BLC VLAR on the CPU mostly. Maybe 30 minutes faster, so instead of 2 hours I've seen a lot process in 90 minutes or so. They process about the same as a GUPPI on the GPU, around 8 minutes or so. Two up on the GPU's. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
They process about the same as a GUPPI on the GPU, around 8 minutes or so. Two up on the GPU's. I run 1 GPU WU at a time, and the current non VLAR Guppies are processing a couple of minutes faster (no less than 2, sometimes almost 4) than the VLAR ones do. Grant Darwin NT |
Stubbles Send message Joined: 29 Nov 99 Posts: 358 Credit: 5,909,255 RAC: 0 |
They process about the same as a GUPPI on the GPU, around 8 minutes or so. Two up on the GPU's. Hey Grant, Glad to see you've become an optimiser :-D ...even though you're still reporting guppis on GPU :-p R ;-) |
Stubbles Send message Joined: 29 Nov 99 Posts: 358 Credit: 5,909,255 RAC: 0 |
Haven't processed anywhere near 100 on CPU. Mine are processing just like a normal BLC VLAR on the CPU mostly. Maybe 30 minutes faster, so instead of 2 hours I've seen a lot process in 90 minutes or so. They process about the same as a GUPPI on the GPU, around 8 minutes or so. Two up on the GPU's. I just find that a sample of less than 50 tasks on a device (CPU or GPU) is usually not enough, since from experience, whenever I reported AVGs of a sample of 30-50 tasks, it was often not very close to the longterm AVG. Maybe a sample of 100 tasks is too much but 50 is definitely not enough for a new set of tasks. If anyone is running BoincTasks, you should be able to easily import your History.cvs into a spreadsheet. Then, it's just a matter of using the Filter to select only the MESSIER031s on CPU or GPU. |
Stephen "Heretic" Send message Joined: 20 Sep 12 Posts: 5557 Credit: 192,787,363 RAC: 628 |
Haven't processed anywhere near 100 on CPU. Mine are processing just like a normal BLC VLAR on the CPU mostly. Maybe 30 minutes faster, so instead of 2 hours I've seen a lot process in 90 minutes or so. They process about the same as a GUPPI on the GPU, around 8 minutes or so. Two up on the GPU's. . . Hi Stubs, . . My rigs were bunkered over the weekend (I had to go to a wedding) so I only started getting these Blc6's today. But of the handful that have so far run on my GTX950 they seem to have runtimes in proportion with Arecibo tasks of a similar AR value (given that Jeff's statement is accurate for the majority of them). While NARAs (normal AR Arecibo tasks) take 30 to 42 Mins (the longer times being on tasks with an AR in the 30's) these have taken times like 44 to 47 Mins which is sort of within the normal variation. Unlike most Green Bank VLARs which take way over 60 mins on that GPU. . . Typical NARA (0.42) . . 35 to 36 Mins . . Typical Blc3 (0.01?). . 62 to 68 Mins . . These Blc6 (stated as 0.30) . . 44 to 48 Mins . . But early days yet, I have a couple of dozen queued on the 950 and I will monitor the results. Stephen . |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
My BOINCTasks history file gets trimmed to 2 days automatically so I ended up with only 81 CPU tasks to average. Sample for GPU is larger at 275 tasks. 81 CPU Messier031 Tasks = 86.94 minutes average 275 GPU Messier031 Tasks = 10.14 minutes average Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Stephen "Heretic" Send message Joined: 20 Sep 12 Posts: 5557 Credit: 192,787,363 RAC: 628 |
My BOINCTasks history file gets trimmed to 2 days automatically so I ended up with only 81 CPU tasks to average. Sample for GPU is larger at 275 tasks. . . Ok, so how does that compare to other Green Bank tasks or Arecibo for that matter? Stephen . |
©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.