Message boards :
Number crunching :
Panic Mode On (101) Server Problems?
Message board moderation
Previous · 1 . . . 9 · 10 · 11 · 12 · 13 · 14 · 15 . . . 27 · Next
Author | Message |
---|---|
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14654 Credit: 200,643,578 RAC: 874 |
Keep arguing Semantics. We are watching you dance around an Obvious problem with the SETI Stats on a SETI webpage from a SETI Server. Yes, it is. And it can be even more interesting if you understand it. Although it isn't publicly advertised, exactly the same generic BOINC code runs at Milkyway: http://milkyway.cs.rpi.edu/milkyway/gpu_list.php They have Macs running NVidia cards, but not running ATI cards - exactly the opposite of the display here. So, who's running a sample NV card on Mac here, so we can look at the timings? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Here is One returning results every few minutes, http://setiathome.berkeley.edu/results.php?hostid=7644315&offset=120 There are many, Many Mac Laptops with nVidia cards returning many more than that one host. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
http://www.primegrid.com/gpu_list.php shows both Nvidia and AMD GPUs on Macs for a variety of applications. |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
They have Macs running NVidia cards, but not running ATI cards - exactly the opposite of the display here. So, who's running a sample NV card on Mac here, so we can look at the timings? Computer 7297852 NVIDIA Quadro K5000/GeForce GT 640 Computer 7599680 NVIDIA GeForce GTX 770 Computer 6863747 NVIDIA GeForce GTX 680MX Computer 7321524 NVIDIA GeForce GTX 680MX Computer 6883710 NVIDIA GeForce GTX 660 Ti Computer 7414871 NVIDIA GeForce GTX 680MX Computer 7453012 NVIDIA GeForce GTX 675MX Computer 7088836 NVIDIA GeForce GTX 680MX Computer 7163398 NVIDIA GeForce GTX 780M Computer 7719315 NVIDIA GeForce GTX 675MX Computer 7773885 NVIDIA GeForce GTX 780M Computer 3502292 NVIDIA GeForce GTX 680MX Computer 6902460 NVIDIA GeForce GTX 675MX Computer 7718750 NVIDIA GeForce GTX 680MX Computer 7494607 NVIDIA GeForce GTX 780M Computer 4726043 NVIDIA GeForce GTX 950 Computer 7183121 NVIDIA GeForce GTX 775M Computer 7752206 NVIDIA GeForce GTX 680MX Computer 7334464 NVIDIA GeForce GT 755M Computer 7592123 NVIDIA GeForce GTX 970 Computer 7453810 NVIDIA GeForce GTX 675MX Computer 7254833 NVIDIA GeForce GT 755M Claggy |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14654 Credit: 200,643,578 RAC: 874 |
I was beginning to get worried about the Oct 20, 2014, bugfix adding anonymous platform hosts to the lists (3136342575c42e7e3b2dc13fbb04feb54ae0c77d). Neither Milkyway not PrimeGrid seem to have that, judging by the missing ", measured by average elapsed time of tasks," in the rubric. Specifically, the first host we looked at - Jason's Xeon - is running anonymous platform: and the handler for anonymous platform ("if ($vendor == "cuda") {$av_ids .= "-3";}") uses "cuda" (only) as the vendor string. But that seems uniform throughout the code, and Claggy's first two examples are both running the stock app. Back to the drawing board... |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
I believe you will find most of the Mac nVidia results are coming from Laptops, running the Stock App, on the same Mobile GPUs. Which means it shouldn't be very difficult for the Server to see Hundreds of results from the exact same nVidia mobile GPU. I have seen Jason's 780 displayed a few times, and when I had My 750Ti in my Mac it did appear a couple times even though it was running 24/7 for weeks. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14654 Credit: 200,643,578 RAC: 874 |
David Anderson has found some Mac GPUs by increasing the result enumeration limit to 2000 per vendor. Evidently a low proportion of Macs are using their GPUs so far. |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
David Anderson has found some Mac GPUs by increasing the result enumeration limit to 2000 per vendor. Evidently a low proportion of Macs are using their GPUs so far. As well as some Windows and especially Linux GPUs. Claggy |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Yes, finally some Mac nVidia GPUs. Thank you very much. Now to see how long it lasts. ;-) |
TimeLord04 Send message Joined: 9 Mar 06 Posts: 21140 Credit: 33,933,039 RAC: 23 |
14 WUs, (processed between both my crunchers), have been marked Invalid. Most of them from Prometheus, my GTX-750 TI SC machine. Now up to 44 due to the Server issue... Plus, another 43 in Validation Inconclusive, ready to become more Invalids... :-( Now it's Exeter in the "lead" of most Invalids. (GTX-760.) Computer IDs: 7301566 - Exeter, GTX-760 7440554 - Prometheus, GTX-750 TI SC TL TimeLord04 Have TARDIS, will travel... Come along K-9! Join Calm Chaos |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13751 Credit: 208,696,464 RAC: 304 |
Was the issue of random abandonment of caches ever resolved? This host here appears to have suffered from the issue. They received a cache full of work on 15th Oct, every 5 minutes picking up 5-10 WUs. 7th Nov, 17:15:25 UTC they received another WU (doesn't appear to be any other communications between those dates), and at the same time all existing WUs were abandoned, 7th Nov, 17:15:25 UTC. On their machine, "Number of time client has contacted server" stands at 3, there were at least 6 contacts just to fill the cache on 15th Oct. EDIT- and I've noticed a lot of errored out WUs from ARM based systems, most common Stderr output, error: only position independent executables (PIE) are supported. this being on hardware with the application version SETI@home v7 v7.21 (armv7-vfpv3d16-nopie) eg ARM WU error Grant Darwin NT |
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
rob smith wrote: I'm not sure why you see "three results the same" - the stderr of the two results that validated look very similar, while the one that failed is very different. IIRC, two AP results were from (MAC) GPUs. My AP result was from the J1900 CPU. The result: single pulses: 0 repetitive pulses: 0 ...was at all 3 results the same. - - - - - - - - - - I myself wrote: ap_23jl11ac_B2_P1_00115_20150924_21168.wu TBar wrote: I'd say it's the same thing mentioned in this post, http://setiathome.berkeley.edu/forum_thread.php?id=78410&postid=1739228#1739228 It's the 2nd time 'I noticed' that my PC got an invalid for an AP result (because of the above mentioned). I don't look every day to the WUs/results overview... This 'resend thing which result in invalid result because a host which missed the deadline and send the result after deadline nevertheless' is already known by the admins? |
TimeLord04 Send message Joined: 9 Mar 06 Posts: 21140 Credit: 33,933,039 RAC: 23 |
[Rant On] Today, I'm up to 53 Invalid, 36 Inconclusive. This is getting really, REALLY stupid!!! These faulty units should just be purged from the system; especially if Eric and the rest know that these are faulty. Why are we being left to take the hit on these ludicrous units??? [Rant Off.] On a more positive note; both my computers put enough work in to raise my RAC back to a more reasonable level. I hope that trend holds. :-) TL TimeLord04 Have TARDIS, will travel... Come along K-9! Join Calm Chaos |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
[Rant On] They could be taking this opportunity to verify other systems in place are working correctly when there are known bad workunits in play. As I understand it. Trying to track down all of the bad workunits and flag them to be canceled by the server would be more work than just letting them run through the system. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Gary Charpentier Send message Joined: 25 Dec 00 Posts: 30692 Credit: 53,134,872 RAC: 32 |
[Rant On] He may have to do that as he really should give credit for work done on known bad work units. |
William Send message Joined: 14 Feb 13 Posts: 2037 Credit: 17,689,662 RAC: 0 |
Was the issue of random abandonment of caches ever resolved? Richard and I managed to find and eliminate one missing safety check, which could lead to tasks being abandoned though present on the host. Are you in touch with the user? If he did a reattach the tasks were rightly abandoned. Only if they are still crunching on the host but marked abandoned on the server, it is a problem. These days if you reattach you should get your old host id but times contacted server reset. [and any tasks still in progress from before marked as abandoned] EDIT- and I've noticed a lot of errored out WUs from ARM based systems, most common Stderr output, Wu is already purged. No idea if Eric has the time to look into it. A person who won't read has no advantage over one who can't read. (Mark Twain) |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13751 Credit: 208,696,464 RAC: 304 |
EDIT- and I've noticed a lot of errored out WUs from ARM based systems, most common Stderr output, I'll keep an eye out. Grant Darwin NT |
Louis Loria II Send message Joined: 20 Oct 03 Posts: 259 Credit: 9,208,040 RAC: 24 |
So, it wasn't just me. Good to know. 142 invalids, 154 errors. Some of those errors were mine, playing with drivers. Otherwise, I hope it gets straighten out. Good luck crunching folks! |
Swibby Bear Send message Joined: 1 Aug 01 Posts: 246 Credit: 7,945,093 RAC: 0 |
FYI - Matt posted a big update under Technical News. Thanks, Matt. Now we're eager for your next writeup! |
TimeLord04 Send message Joined: 9 Mar 06 Posts: 21140 Credit: 33,933,039 RAC: 23 |
Up to 58 Invalids; down to 23 Inconclusives... Yet, tomorrow is another Outrage day... What will that bring, next??? :-O TL P.S. [Rant On:] I'm with Gary; we should definitely be granted some credit for these faulty WUs... It's NOT the fault of our systems that they all error out! It's NOT our fault that time is wasted on them! We spend the time, we should get credit! [Rant Off.] TimeLord04 Have TARDIS, will travel... Come along K-9! Join Calm Chaos |
©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.