Panic Mode On (113) Server Problems?

Message boards : Number crunching : Panic Mode On (113) Server Problems?
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 30 · 31 · 32 · 33 · 34 · 35 · 36 . . . 37 · Next

AuthorMessage
Speedy
Volunteer tester
Avatar

Send message
Joined: 26 Jun 04
Posts: 1643
Credit: 12,921,799
RAC: 89
New Zealand
Message 1963595 - Posted: 6 Nov 2018, 7:53:55 UTC - in response to Message 1963582.  

Since all data split for more than a year has been the "new" format cancel your thought about anything we are seeing today being a "new format" because it isn't.
It's the data contained within the tape that is having an impact on the run-time -sometimes we get data that is much faster to process than others.

Getting data which is much faster to process helps us make a little bit of headway on all of data that needs processing
ID: 1963595 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1963653 - Posted: 6 Nov 2018, 16:56:18 UTC

. . Ruh Roh!

. . Seems to be a rather late start to the outrage ...

. . I wonder what is going on.

Stephen

:(
ID: 1963653 · Report as offensive
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1963654 - Posted: 6 Nov 2018, 16:57:34 UTC - in response to Message 1963653.  

Umm, voting ...
ID: 1963654 · Report as offensive
Sirius B Project Donor
Volunteer tester
Avatar

Send message
Joined: 26 Dec 00
Posts: 24879
Credit: 3,081,182
RAC: 7
Ireland
Message 1963656 - Posted: 6 Nov 2018, 21:01:07 UTC

Hmm, only 4 hours?
ID: 1963656 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1963690 - Posted: 7 Nov 2018, 2:13:59 UTC - in response to Message 1963582.  

Since all data split for more than a year has been the "new" format cancel your thought about anything we are seeing today being a "new format" because it isn't.
It's the data contained within the tape that is having an impact on the run-time -sometimes we get data that is much faster to process than others.


. . Actually there was no thought of another "new" format, quite the opposite, the recent tapes were old format. The current Blc01 series and the latest Blc04 series are the 'new' format from the past year. The run times of these and the previous series were too consistent and typical of one type or the other to make me believe the variance was simply the content of the recorded data.

. . But maybe you know something that I don't ...

Stephen

<shrug>
ID: 1963690 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1963691 - Posted: 7 Nov 2018, 2:38:41 UTC - in response to Message 1963690.  

What "recent" tapes are you talking about. The oldest in my list are the BLC22 and 23 in October. They were the "new" 2-bit format. We've been doing 2-bit for over a year now as has been already mentioned.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1963691 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13733
Credit: 208,696,464
RAC: 304
Australia
Message 1963707 - Posted: 7 Nov 2018, 5:59:57 UTC
Last modified: 7 Nov 2018, 6:05:05 UTC

Regardless of what the WU file name might say, we've been doing 4bit WUs for a bit over 2 years now, not too long after we first started processing GBT data.
Grant
Darwin NT
ID: 1963707 · Report as offensive
ShadowBot 3.75
Volunteer tester

Send message
Joined: 28 Mar 00
Posts: 9
Credit: 23,708,744
RAC: 176
United States
Message 1963708 - Posted: 7 Nov 2018, 6:02:12 UTC

It is interesting when you look at the results, the flopcount was significantly higher for the blc22/23 than the blc01/04

blc23 - Flopcounter: 37269949406435.922000
blc04 - Flopcounter: 23945220084970.641000

I am not sure what conclusions you can draw from this data, but it is pretty consistent for the different WU batches.
ID: 1963708 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1963713 - Posted: 7 Nov 2018, 7:11:06 UTC
Last modified: 7 Nov 2018, 7:13:17 UTC

It was mentioned already that certain tasks are just easier to process. Whether because of the internal structure or possibly that the amount of signal pulses seem to be reduced in the BLC01/04 tasks. Based on the number of spikes, pulses, triplets in the validated tasks, they seem to be 'quieter' with not as much 'noise' to have to filter through looking for signals. Thus the reduced flop counter totals.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1963713 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1963715 - Posted: 7 Nov 2018, 7:45:34 UTC - in response to Message 1963691.  
Last modified: 7 Nov 2018, 7:50:08 UTC

What "recent" tapes are you talking about. The oldest in my list are the BLC22 and 23 in October. They were the "new" 2-bit format. We've been doing 2-bit for over a year now as has been already mentioned.


. . We upgraded to 4 bit tapes way over a year ago (hence the 700Kb file sizes) but that is not what I was talking about. In October last year there was some change made with a series of Blc04 tapes that allowed them to run in the same time as Arecibo normal WUs. No overflows or "shorties" but full run files. That was followed by another couple of series that showed the original GBT run times of about 70% greater than norms. Then all the GBT tapes became this nice new type up until a month or 2 back when they reverted to the slow long run times. There was a Blc04, Blc06, Blc22 and the Blc23 series. Now the Blc01 and Blc04 series we are seeing are the nice "new' quick run times.

Stephen

. .
ID: 1963715 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13733
Credit: 208,696,464
RAC: 304
Australia
Message 1963717 - Posted: 7 Nov 2018, 8:00:03 UTC - in response to Message 1963715.  

In October last year there was some change made with a series of Blc04 tapes that allowed them to run in the same time as Arecibo normal WUs. No overflows or "shorties" but full run files. That was followed by another couple of series that showed the original GBT run times of about 70% greater than norms. Then all the GBT tapes became this nice new type up until a month or 2 back when they reverted to the slow long run times. There was a Blc04, Blc06, Blc22 and the Blc23 series. Now the Blc01 and Blc04 series we are seeing are the nice "new' quick run times.

There have been no changes made- it's all just data, and like Arecibo WUs some have larger & some have smaller Angle Ranges, hence the different run times.
Grant
Darwin NT
ID: 1963717 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22191
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1963719 - Posted: 7 Nov 2018, 8:13:49 UTC

For a given timeslot all collected data will have the same AR, but the amount of noise on the each receiver channel could well be different, and some channels due to their design are more "noise prone" than others.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1963719 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1963721 - Posted: 7 Nov 2018, 9:08:50 UTC

You have to pick apart the various different influences on runtime.

Noise - usually results in a task overflowing (30+ signals) and exiting early.
AR - almost all BLC tapes are vlar, which process slowly on, especially, NVidia GPUs.
Processing configuration - affects the number of flops needed, and can be seen on both GPU and GPU runs.

There was a significant change between the BLC22/23 tapes and the BLC01/04 tapes. That change affected the runtimes on both CPUs and GPUs. So I'd put it in the third group.
ID: 1963721 · Report as offensive
Profile Unixchick Project Donor
Avatar

Send message
Joined: 5 Mar 12
Posts: 815
Credit: 2,361,516
RAC: 22
United States
Message 1964135 - Posted: 9 Nov 2018, 14:27:27 UTC

Hoping the Seti team gives us more datafiles before going home for the weekend. No need for an early Monday morning panic.
ID: 1964135 · Report as offensive
Profile Jimbocous Project Donor
Volunteer tester
Avatar

Send message
Joined: 1 Apr 13
Posts: 1853
Credit: 268,616,081
RAC: 1,349
United States
Message 1964307 - Posted: 10 Nov 2018, 5:44:40 UTC - in response to Message 1964135.  

Hoping the Seti team gives us more datafiles before going home for the weekend. No need for an early Monday morning panic.

Or Tuesday, as the case may be. I think Monday is observed as a holiday on the left coast :)
ID: 1964307 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1964319 - Posted: 10 Nov 2018, 7:01:20 UTC - in response to Message 1964307.  

Yes the Berkeley campus will observe Veterans Day on Monday.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1964319 · Report as offensive
Profile Unixchick Project Donor
Avatar

Send message
Joined: 5 Mar 12
Posts: 815
Credit: 2,361,516
RAC: 22
United States
Message 1964377 - Posted: 10 Nov 2018, 15:38:21 UTC

It looks like someone is posting some Aricebo files for us so that helps, but not nearly enough to get us through to Tuesday. At least there is hope that someone is paying attention and will load more files when we get low.
ID: 1964377 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 5124
Credit: 276,046,078
RAC: 462
Message 1964400 - Posted: 10 Nov 2018, 16:23:28 UTC - in response to Message 1964377.  
Last modified: 10 Nov 2018, 16:32:53 UTC

It looks like someone is posting some Aricebo files for us so that helps, but not nearly enough to get us through to Tuesday. At least there is hope that someone is paying attention and will load more files when we get low.


Are those files something like "01no18aa........"? I am crunching them on my CPU at something like 1.5 to 2.5 hours. The blc04_.... were crunching in 1 hour 6-8 minutes.

My gpu is crunching those in about 4 minutes but were doing the blc's in 2 minutes 10 to about 3 minutes.

Tom
A proud member of the OFA (Old Farts Association).
ID: 1964400 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1964409 - Posted: 10 Nov 2018, 16:45:28 UTC - in response to Message 1964400.  

Arecibo files always have been the hardest to crunch. Especially now that Arecibo VLAR's are sent to Nvidia cards.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1964409 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1964412 - Posted: 10 Nov 2018, 16:49:12 UTC - in response to Message 1964409.  

Arecibo files always have been the hardest to crunch. Especially now that Arecibo VLAR's are sent to Nvidia cards.
And the Arecibo file naming convention - day, month, year, sequence letters - has been unchanged for 20 years.
ID: 1964412 · Report as offensive
Previous · 1 . . . 30 · 31 · 32 · 33 · 34 · 35 · 36 . . . 37 · Next

Message boards : Number crunching : Panic Mode On (113) Server Problems?


 
©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.