Pendings are very high

Message boards : Number crunching : Pendings are very high
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Profile Chris S Crowdfunding Project Donor*Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 19 Nov 00
Posts: 40477
Credit: 41,361,524
RAC: 402
United Kingdom
Message 1901499 - Posted: 17 Nov 2017, 13:30:17 UTC

I don't know if anyone else has noticed this, but over the last couple of months my pendings are getting ever bigger in numbers. About 40% of everything I crunch is going in the pendings list. Yet some of my wingmen are running i5's and 970 cards and are regularly contacting Boinc. So why aren't they reporting work?

No wonder my RAC won't increase ......
ID: 1901499 · Report as offensive
kittyman Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 49848
Credit: 914,567,858
RAC: 161,225
United States
Message 1901500 - Posted: 17 Nov 2017, 13:31:57 UTC - in response to Message 1901499.  

I don't know if anyone else has noticed this, but over the last couple of months my pendings are getting ever bigger in numbers. About 40% of everything I crunch is going in the pendings list. Yet some of my wingmen are running i5's and 970 cards and are regularly contacting Boinc. So why aren't they reporting work?

No wonder my RAC won't increase ......

Like wayward kitties, pendings always come home eventually.......
Meow.
What meowing lurks in the hearts of man? The kittyman knows....MEOWhahahahahahha!

Have made friends here.
Most were cats.
ID: 1901500 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 31104
Credit: 63,312,511
RAC: 25,687
Germany
Message 1901518 - Posted: 17 Nov 2017, 15:41:38 UTC

Your pendings aren`t higher than mine.
So nothing suspicious.
With each crime and every kindness we birth our future.
ID: 1901518 · Report as offensive
Tutankhamon
Volunteer tester
Avatar

Send message
Joined: 1 Nov 08
Posts: 7113
Credit: 44,224,049
RAC: 4,082
Sweden
Message 1901523 - Posted: 17 Nov 2017, 15:47:25 UTC

Pendings=Money in the bank :-)
ID: 1901523 · Report as offensive
Profile Chris S Crowdfunding Project Donor*Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 19 Nov 00
Posts: 40477
Credit: 41,361,524
RAC: 402
United Kingdom
Message 1901534 - Posted: 17 Nov 2017, 16:05:18 UTC

Ok, thanks for the replies. They just seemed a lot higher that what I remember themto be. First time I've ever run two cards in one rig, was rather hoping to see the RAC go up a bit quicker. No matter.
ID: 1901534 · Report as offensive
kittyman Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 49848
Credit: 914,567,858
RAC: 161,225
United States
Message 1901535 - Posted: 17 Nov 2017, 16:08:01 UTC - in response to Message 1901534.  

Ok, thanks for the replies. They just seemed a lot higher that what I remember themto be. First time I've ever run two cards in one rig, was rather hoping to see the RAC go up a bit quicker. No matter.

The faster you turn the work around, the higher your pendings will be.
They will stabilize at some point and hover around that value.
What meowing lurks in the hearts of man? The kittyman knows....MEOWhahahahahahha!

Have made friends here.
Most were cats.
ID: 1901535 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6530
Credit: 185,628,208
RAC: 44,523
United States
Message 1901564 - Posted: 17 Nov 2017, 18:36:04 UTC

I don't really pay attention to tasks in pending or inconclusive .
I really only find In Progress, Valid, Invalid, and Error useful
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the BP6/VP6 User Group today!
ID: 1901564 · Report as offensive
Cruncher-American Special Project $75 donor

Send message
Joined: 25 Mar 02
Posts: 1448
Credit: 265,554,225
RAC: 150,692
United States
Message 1901584 - Posted: 17 Nov 2017, 21:25:00 UTC - in response to Message 1901564.  

High Pendings tell me that I have gotten ahead of my wingmen (on average), so I know that they consist of more older, slower machines than the usual mix. That's all...but I then know I WILL get a bump in credit when they show up, to offset the current lower credit as the current Ps accumulate.
ID: 1901584 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 9899
Credit: 128,434,126
RAC: 80,904
Australia
Message 1901591 - Posted: 17 Nov 2017, 23:36:32 UTC - in response to Message 1901534.  

Ok, thanks for the replies. They just seemed a lot higher that what I remember themto be. First time I've ever run two cards in one rig, was rather hoping to see the RAC go up a bit quicker. No matter.

As kittyman pointed out, the more work, and the faster you do it, the higher the number of pendings will be.
Generally Pendings are up to 1/3 of the All number.
For Petri
All 17,237 Pending 5628
For my i7
All 2,194 Pending 817
My C2D
All 679 Pending 242

The more work you do, the higher the All number, and the higher the Pending number.

RAC is a lagging indicator- putting a new system online it takes 6-8 weeks for RAC to settle. Adding another card, or just upgrading a card, it still can take 6-8 weeks for the RAC to settle around it's new value. And that's without changes in the work mix (GBT v Arecibo, amount of AP work if you do that), let alone Server downtime above & beyond the weekly outage.
Grant
Darwin NT
ID: 1901591 · Report as offensive
Profile betreger Project Donor
Avatar

Send message
Joined: 29 Jun 99
Posts: 7664
Credit: 18,417,196
RAC: 10,010
United States
Message 1901628 - Posted: 18 Nov 2017, 1:29:25 UTC - in response to Message 1901591.  

Generally Pendings are up to 1/3 of the All number.

State: All (440) · In progress (85) · Validation pending (161)
Not on the box I run Seti on
ID: 1901628 · Report as offensive
Profile Brent Norman Special Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2129
Credit: 203,897,642
RAC: 508,018
Canada
Message 1901637 - Posted: 18 Nov 2017, 1:39:10 UTC - in response to Message 1901628.  
Last modified: 18 Nov 2017, 1:40:37 UTC

I generally find myself running higher pendings than valids, but not right now it seems.

3.5 hour turnaround
All (7945) · In progress (717) · Validation pending (3439) · Validation inconclusive (175) · Valid (3614)

12 hour turnaround
All (2170) · In progress (300) · Validation pending (906) · Validation inconclusive (35) · Valid (939)
... typically this one has lower percentage of pendings.
ID: 1901637 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 9899
Credit: 128,434,126
RAC: 80,904
Australia
Message 1901638 - Posted: 18 Nov 2017, 1:40:39 UTC - in response to Message 1901628.  
Last modified: 18 Nov 2017, 1:42:04 UTC

Generally Pendings are up to 1/3 of the All number.

State: All (440) · In progress (85) · Validation pending (161)
Not on the box I run Seti on

?
2.7 is more than a half, but less than 1/3. And it's closer to 1/3 than it is to 1/2.

The faster you return work, the higher pendings tend to be, the slower you return work, the lower pendings tend to be.
Hence my use of "Generally" and "up to" and "tend to".
Grant
Darwin NT
ID: 1901638 · Report as offensive
Profile Chris S Crowdfunding Project Donor*Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 19 Nov 00
Posts: 40477
Credit: 41,361,524
RAC: 402
United Kingdom
Message 1901662 - Posted: 18 Nov 2017, 6:51:39 UTC

You've all made sensible observations, and yes py pendgs are around 1/3 of the total which is normal. Still don't quite remember it taking 6-8 weeks to settle down after an upgrade or new rig, but it's just me being impatient a bit I guess. A bit like going on holiday in the car with the parents "are we there yet, are we there yet!!!".
ID: 1901662 · Report as offensive
Ghia
Avatar

Send message
Joined: 7 Feb 17
Posts: 164
Credit: 13,805,687
RAC: 23,686
Norway
Message 1901682 - Posted: 18 Nov 2017, 9:39:28 UTC

My pendings are also high atm. Luck of the draw with wingmen, I guess...sometimes I have to wait for time-outs, and then the WU gets sent to a very slow computer. Oh well...all WUs will clear out sooner or later ;-)

...Ghia...
Humans may rule the world...but bacteria run it...
ID: 1901682 · Report as offensive
Cap'n Bludd

Send message
Joined: 8 Sep 13
Posts: 2
Credit: 19,440,804
RAC: 13,798
United States
Message 1902213 - Posted: 21 Nov 2017, 10:09:59 UTC
Last modified: 21 Nov 2017, 10:13:41 UTC

I've noticed a big drop in RAC, not just for myself but for my group. I'd been in the high 20k range and this week I'm dropping more than 10k points and so has the guy that usually has the same numbers as I do, the rest the group is dropping by the same proportion. I also have the 1/3 pendings folks say is normal.

I usually only look at my RAC and our small group's active list now and then but I can't recall something like this happening before. I was into the gold badge range last week or the week before.
ID: 1902213 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 9899
Credit: 128,434,126
RAC: 80,904
Australia
Message 1902215 - Posted: 21 Nov 2017, 10:18:04 UTC - in response to Message 1902213.  

I can't recall something like this happening before.

It's happened before, it'll happen again.

For quite a while, we had mostly Arecibo WUs, which take less time to process and give more Credit than GBT WUs. Over the last few weeks the percentage of WUs have moved over to more GBT than Arecibo, so longer to process and less Credit. Add to that the current batch of GBT WUs are the longest running yet- so even less Credit than in the past with a similar mix of work.
At some stage in the future there will probably be a drop in GBT WUs, and a corresponding increase in Arecibo WUs and RAC will rise again; till there's more GBT WUs once again & RACs will start to fall, once again.
Rinse and repeat.
Grant
Darwin NT
ID: 1902215 · Report as offensive
Profile Chris S Crowdfunding Project Donor*Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 19 Nov 00
Posts: 40477
Credit: 41,361,524
RAC: 402
United Kingdom
Message 1902296 - Posted: 22 Nov 2017, 7:02:55 UTC

It could be said could it not, that all work supplied from Arecibo or GBT is still Seti work at the end of the day. So why is there different credit awarded? We know and accept that Astropulse work is rarer than hens teeth and is always welcome when we get it, but the algorithms used there I believe are quite different. At present I'm getting about 60% of Guppi work in the mix,

I've been around long enough to remember the original intention and premise behind Boinc when it first came out to replace Classic. Which was that to avoid favouritism to a project for paying more credit than the others, all projects would pay the same. People would then choose which project to support depending upon the perceived scientific merit.

Well that didn't last 5 minutes did it, each project admin stuck two fingers up and paid whatever they wanted to draw the punters in. Now we can't even have parity within the same project!! I am more than happy to lend our Richard Haselgrove to Boinc to help sort out the mess they have got themselves in, but we want him back here at some point OK??
ID: 1902296 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 9899
Credit: 128,434,126
RAC: 80,904
Australia
Message 1902301 - Posted: 22 Nov 2017, 7:22:23 UTC - in response to Message 1902296.  

It could be said could it not, that all work supplied from Arecibo or GBT is still Seti work at the end of the day. So why is there different credit awarded?

Credit New.
Grant
Darwin NT
ID: 1902301 · Report as offensive
rob smith Special Project $250 donor
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 16154
Credit: 312,517,053
RAC: 257,431
United Kingdom
Message 1902304 - Posted: 22 Nov 2017, 7:52:25 UTC

Grant got it in one
The reason being that CreditNew is not properly calibrated to cope with the different sources of data. It also has flaws that make it difficult to maintain such calibration. A "proper" credit granting system would grant the same amount of credit on a given processor on a particular computer for a benchmark task last year, this week and next year. Because CreditNew tries to continuously re-calibrate itself it will not do that. Add to that the differences in the actual data the amount of calculation required for the two major sources (and effectively six task types) it is hardly surprising that the amount of credit per hour on a given processor varies so much.

It is not beyond the wit of man to come up with a credit system that would cope with the six task types and award credit consistently, based on the actual task (server side action), not on the run-time of the computer (client side action). CreditNew is predominantly a client side process, with a bit of "interfering" from the server.....
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1902304 · Report as offensive
Profile Chris S Crowdfunding Project Donor*Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 19 Nov 00
Posts: 40477
Credit: 41,361,524
RAC: 402
United Kingdom
Message 1902321 - Posted: 22 Nov 2017, 11:00:07 UTC
Last modified: 22 Nov 2017, 11:00:59 UTC

But no funding any more to put it right Rob. And if the truth be known little interest either. This is already a 6 day a week project, it only appeals to the true supporters these days.
ID: 1902321 · Report as offensive
1 · 2 · Next

Message boards : Number crunching : Pendings are very high


 
©2018 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.