GIANNI ligand3-0-1 bad batch?

Message boards : Number crunching : GIANNI ligand3-0-1 bad batch?

Author	Message
Stoneageman Send message Joined: 25 May 09 Posts: 224 Credit: 34,057,374,498 RAC: 0 Level Scientific publications	Message 36376 - Posted: 18 Apr 2014 \| 7:14:49 UTC Last modified: 18 Apr 2014 \| 7:19:40 UTC
	All errorring out after 2 secs with ERROR: file pme.cpp line 85: PME NX too small
	ID: 36376 \| Rating: 0 \| rate: / Reply Quote

GDF Volunteer moderator Project administrator Project developer Project tester Volunteer developer Volunteer tester Project scientist Send message Joined: 14 Mar 07 Posts: 1957 Credit: 629,356 RAC: 0 Level Scientific publications	Message 36379 - Posted: 18 Apr 2014 \| 9:46:17 UTC - in response to Message 36376.
	sorry my fault testing a new submission system. It should be fixed now in lig3. g
	ID: 36379 \| Rating: 0 \| rate: / Reply Quote

Stoneageman Send message Joined: 25 May 09 Posts: 224 Credit: 34,057,374,498 RAC: 0 Level Scientific publications	Message 36395 - Posted: 18 Apr 2014 \| 16:33:29 UTC Last modified: 18 Apr 2014 \| 16:35:29 UTC
	They seem to be working OK now except this one I noticed was going at a snails pace, so I aborted it.
	ID: 36395 \| Rating: 0 \| rate: / Reply Quote

Richard Haselgrove Send message Joined: 11 Jul 09 Posts: 1629 Credit: 9,819,914,649 RAC: 8,558,238 Level Scientific publications	Message 36459 - Posted: 20 Apr 2014 \| 11:44:56 UTC - in response to Message 36395.
	They seem to be working OK now except this one I noticed was going at a snails pace, so I aborted it. I'd say 'long', rather than 'slow'. My GTX 670 machine got one of these too, and took over 14 hours to complete it. The credit awarded was proportional to the time spent, so I presume the team knew in advance how big these tasks are. By coincidence, the next two tasks allocated to the host were 'GIANNI_trypben', at the opposite end of the size range: well under 4 hours on this card. Gianni, if your tasks are going to have such a variable runtime, and if you're working on a new job submission system, could you extend that system to adjust <rsc_fpops_est> to match the job sizes, please? Our BOINC clients have a hard time scheduling everything properly when the estimates turn out not to match reality.
	ID: 36459 \| Rating: 0 \| rate: / Reply Quote

Stoneageman Send message Joined: 25 May 09 Posts: 224 Credit: 34,057,374,498 RAC: 0 Level Scientific publications	Message 36462 - Posted: 20 Apr 2014 \| 16:42:02 UTC
	When I said 'snails pace' I meant it had stopped making any meaningful progress rather than just slow. Since then I've had two error after 7hrs on different cards (780 & 570). Need to watch these like a hawk.
	ID: 36462 \| Rating: 0 \| rate: / Reply Quote

Matt Send message Joined: 11 Jan 13 Posts: 216 Credit: 846,538,252 RAC: 0 Level Scientific publications	Message 36463 - Posted: 20 Apr 2014 \| 16:50:38 UTC
	I've received and successfully processed one of these so far. They do run much longer than anything else I've seen.
	ID: 36463 \| Rating: 0 \| rate: / Reply Quote

skgiven Volunteer moderator Volunteer tester Send message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level Scientific publications	Message 36478 - Posted: 21 Apr 2014 \| 14:57:51 UTC - in response to Message 36463. Last modified: 21 Apr 2014 \| 15:03:07 UTC
	I will be keeping a lookout for tasks that don't progress; I've been caught out by those before! I've just finished a GIANNI_lig3 on a GTX770@1267MHz (W7). Uses 76% Power, 90% GPU utilization, ~750MB GDDR. It took 12h 39min. This of course means if I have a large cache I will miss the 24h bonus. So a nice low cache will be needed by many, and WRT the power usage, I suggest people check these don't push some cards too hard (check your temps and OC's). I would expect these to sneak in under 12h on a GTX770 on Linux and XP (with a good setup). The upload proportionally larger, 77MB. The time on a GTX670 should be about 13h10min (W7), possibly 15min under 12h on Linux but a bit too close to call. It's the first proper LONG run I've seen in a while, but does fit the description of 8h to 12h on the fastest card, the Titan Black, which is almost 50% faster than a GTX770. ____________ FAQ's HOW TO: - Opt out of Beta Tests - Ask for Help
	ID: 36478 \| Rating: 0 \| rate: / Reply Quote

Retvari Zoltan Send message Joined: 20 Jan 09 Posts: 2364 Credit: 16,537,396,044 RAC: 1,268,483 Level Scientific publications	Message 36487 - Posted: 21 Apr 2014 \| 16:35:17 UTC - in response to Message 36478.
	I would expect these to sneak in under 12h on a GTX770 on Linux and XP (with a good setup). You are right. This one took 12 hours 1 minute and 41 seconds on a factory clocked (1084MHz) GTX 680 under WinXPx64.
	ID: 36487 \| Rating: 0 \| rate: / Reply Quote

Mumak Send message Joined: 7 Dec 12 Posts: 92 Credit: 225,897,225 RAC: 0 Level Scientific publications	Message 36530 - Posted: 22 Apr 2014 \| 10:30:33 UTC
	Took 19h:43m on a 750 Ti. GPU Temp ~48 C.
	ID: 36530 \| Rating: 0 \| rate: / Reply Quote

biodoc Send message Joined: 26 Aug 08 Posts: 183 Credit: 10,085,929,375 RAC: 645 Level Scientific publications	Message 36533 - Posted: 22 Apr 2014 \| 11:03:07 UTC
	9 hours, 38 minutes on my GTX780Ti (underclocked; 906 MHz, linux). 2 have finished successfully.
	ID: 36533 \| Rating: 0 \| rate: / Reply Quote

Carlesa25 Send message Joined: 13 Nov 10 Posts: 328 Credit: 72,619,453 RAC: 0 Level Scientific publications	Message 36592 - Posted: 23 Apr 2014 \| 18:02:01 UTC - in response to Message 36533.
	Hello: Finished a task correctly on my GTX 770 Linux 11H 20min. The CPU load is too low does not pass of 9% over the GPU 98%.
	ID: 36592 \| Rating: 0 \| rate: / Reply Quote

dskagcommunity Send message Joined: 28 Apr 11 Posts: 462 Credit: 892,415,458 RAC: 451,598 Level Scientific publications	Message 36632 - Posted: 25 Apr 2014 \| 9:55:07 UTC Last modified: 25 Apr 2014 \| 9:55:58 UTC
	~18:40 on 570gtx with XP 32 ____________ DSKAG Austria Research Team: http://www.research.dskag.at
	ID: 36632 \| Rating: 0 \| rate: / Reply Quote

Post to thread

Message boards : Number crunching : GIANNI ligand3-0-1 bad batch?

	About	Science	Volunteers	Performance	Forum	Join us	Donate