I am using all my computers 24/7. Especially 5867601. But my rac is dropping dramatically. I have over 3k pendings. Anyone know how this is going to be fixed?Anything to do with recent server outage?

State: All (9786) | In progress (5475) | Pending (3123) | Valid (1162) | Application: All (9786) | Astropulse v505 (7) | SETI@home Enhanced (9779)

---

As far as things go here, everything is working fine at the S@H end. There are various reasons why your RAC may be dropping, one being that your RAC depends on how quickly your wing-man gets through the WU that you've done or are doing. Errors and invalids also add to the mix. Looking at the Application Details, the number of consecutive valid tasks for the nVidia GPU (optimised app?) looks a little low, so you might have a bit of a problem there. Inconclusives will get sent out to another machine, so even on a 'shortie' you could be waiting for a month or so, depending on the other machine(s) involved, for any credit or otherwise.

---

Don't take life too seriously, as you'll never come out of it alive!

And with the server woes of recent days there have been a number of wingmen dumping their loads and returning home. Which only adds to the delays for those of who are prepared to fly on to the great target...

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

And with the server woes of recent days there have been a number of wingmen dumping their loads and returning home. Which only adds to the delays for those of who are prepared to fly on to the great target...

Also, on the flip side some are increasing their cache which means they have more work waiting on their rigs to get them through these outages. I had mine set to 6 days but kept running out before they could get the servers back online so I went up to 10 days. I've been making changes to my rig and BOINC was/is having trouble figuring out how much work I actually need. :-)

---

PROUD MEMBER OF Team Starfire World BOINC

And with the server woes of recent days there have been a number of wingmen dumping their loads and returning home. Which only adds to the delays for those of who are prepared to fly on to the great target...

Also, on the flip side some are increasing their cache which means they have more work waiting on their rigs to get them through these outages. I had mine set to 6 days but kept running out before they could get the servers back online so I went up to 10 days. I've been making changes to my rig and BOINC was/is having trouble figuring out how much work I actually need. :-)

What you need is one WU and a spare (for each CPU/GPU), plus a fast, responsive server :-) What you would like is maybe different.

Perryjay, have you ever tried turning that telescope round, and - instead of thinking about your response to the servers' "problems" - thought about the servers' response to your "problems"?

We had a look at that recently, and worked out that the servers had to hold Two thousand five hundred gigabytes to service people's extended caches. Since then, the number of MB tasks "out in the field" has increased by 11%, in just two weeks. More storage, on more elderly disks prone to failure. Longer recovery time when they do fail. More bandwidth wasted (have you looked at the size of your sched_request_setiathome.berkeley.edu.xml file recently?). Bigger database tables, longer to backup/compact, slower to query. Need I go on?

All true Richard but it is really annoying when I had a six day cache set and would run out after a day and a half. I have run it up to ten until BOINC figures out exactly, or at least closely, how long it's gonna take me to do the work. Once it gets somewhere close I will go back to a lower cache. And before you say it, I do have E@H as a backup project so I can stay busy but I would rather run SETI.

---

PROUD MEMBER OF Team Starfire World BOINC

As far as things go here, everything is working fine at the S@H end. There are various reasons why your RAC may be dropping, one being that your RAC depends on how quickly your wing-man gets through the WU that you've done or are doing. Errors and invalids also add to the mix. Looking at the Application Details, the number of consecutive valid tasks for the nVidia GPU (optimised app?) looks a little low, so you might have a bit of a problem there. Inconclusives will get sent out to another machine, so even on a 'shortie' you could be waiting for a month or so, depending on the other machine(s) involved, for any credit or otherwise.

And what about all those pendings. Will they go to zero at some point?

---

In the style of Mr Spock Logic dictates that unless the output from the servers ceases completely and until all work in progress is validated under those conditions, then there will always be pendings.

---

Don't take life too seriously, as you'll never come out of it alive!

And what about all those pendings. Will they go to zero at some point?

With a RAC like yours you'll never run out of pendings! :-)

If you mean the pending credit list on your account page then yes, none are being added to it because with the new credit scheme they don't figure credit until both work units are completed. If you mean pendings in general then no, they will never go away as long as people make mistakes, join and then quit without clearing their cache, or machines fail. Very rarely you will see one that doesn't give credit because it had too many errors returned but this is something that doesn't happen often. Most all your pending will eventually finish and you will get credit for them unless something was wrong with your return.

---

PROUD MEMBER OF Team Starfire World BOINC

And what about all those pendings. Will they go to zero at some point?

With a RAC like yours you'll never run out of pendings! :-)

If you mean the pending credit list on your account page then yes, none are being added to it because with the new credit scheme they don't figure credit until both work units are completed. If you mean pendings in general then no, they will never go away as long as people make mistakes, join and then quit without clearing their cache, or machines fail. Very rarely you will see one that doesn't give credit because it had too many errors returned but this is something that doesn't happen often. Most all your pending will eventually finish and you will get credit for them unless something was wrong with your return.

I do not have problem with the new system. But I drop my rac from 88000 last week ( number 6 at top computer list) and now i am straggling to keep at the top 20 with 50000.
The only change i did was to go from 6.12.26 to 6.10.60 ( far better version ).

---

Love it! ;)

I do not have problem with the new system. But I drop my rac from 88000 last week ( number 6 at top computer list) and now i am straggling to keep at the top 20 with 50000.
The only change i did was to go from 6.12.26 to 6.10.60 ( far better version ).

Tim:
What you appear not to understand is that your RAC is a moving average, and is dependent not just on your production, but also on those machines you are paired with. No matter how fast you crunch and return a Task, nobody gets credit until BOTH crunchers report and the results are validated. If you are paired with slower machines (like mine), or ones who have large caches and report results in big bunches, it may take days or weeks to validate your tasks and get credit. And if one of your wingmen doesn't return his Result, or it doesn't validate and requires a third opinion....

And server problems can also delay reporting and validation. We ALL see our RACs drop during and for a short time after an outage. (Some more than others, but still...)

So when you have a lot of "Pending" tasks, your RAC drops. But when those tasks finally validate, your RAC goes back up. Your RAC will eventually stabilize, and then fluctuate around that stable value.

But unless your machines are throwing off errors or invalid Results, don't lose sleep over it.

---

Donald
Infernal Optimist / Submariner, retired

''What you appear not to understand is that your RAC is a moving average, and is dependent not just on your production, but also on those machines you are paired with. No matter how fast you crunch and return a Task, nobody gets credit until BOTH crunchers report and the results are validated. If you are paired with slower machines (like mine), or ones who have large caches and report results in big bunches, it may take days or weeks to validate your tasks and get credit. And if one of your wingmen doesn't return his Result, or it doesn't validate and requires a third opinion....''

So if this occurs, it doesn't matter if i work 4 gtx 590, or 1 gt 9500.
The faster machine will wait for the slower.
Thank you

---

''What you appear not to understand is that your RAC is a moving average, and is dependent not just on your production, but also on those machines you are paired with. No matter how fast you crunch and return a Task, nobody gets credit until BOTH crunchers report and the results are validated. If you are paired with slower machines (like mine), or ones who have large caches and report results in big bunches, it may take days or weeks to validate your tasks and get credit. And if one of your wingmen doesn't return his Result, or it doesn't validate and requires a third opinion....''

So if this occurs, it doesn't matter if i work 4 gtx 590, or 1 gt 9500.
The faster machine will wait for the slower.
Thank you

Not necessarily. Many slower machines are left on BOINC's default cache setting of 0.5 days. Even if it takes them a day to do the computation, their turnround time should be no more than two days.

On the other hand, many 'fast' machines take ten days to return a result. In the case, the 'slow' machine has to wait for the 'fast' machine to get around to doing the work.

Continuing the RAC analysis, an immed 'average' or idea of production can be got from the statistics graphcs in the BOINC client.

Also of course you can go by your lifetime average i.e since your start-up date.

As a proud owner of a P4, I agree with Richard. I run with a 3 day cache and I have 13 pendings as of a few moments ago. I only checked four, One im waiting on a quad 4, another is a P4 with dual core. one a fermi card ( to be fair that one is marked inconclusive, ) and another old P4 onelunger.

All my machines are down in rac. Id say due to tons of shortys.

---

[/quote]

Old James

Hey guys!

I'm only using one laptop for s@h and have had it running 24/7, so it's an easy thing to monitor. For the best part of 6 months it had a fairly constant RAC of (around) 1500. During the outages it would dip down, of course, but eventually climb back up to a steady 1500. High Score: 1550

But towards the end of April (and for 6-7 weeks) my RAC "decided" it wanted to steadily rise, and eventually reach 1800! That's 20%! All things being equal on my side (yes pendings too) I can only consider this behaviour "artificial". For example: maybe it's got something to do with the shorty-storm.

Long story short, my RAC is declining too and tbh I expect it to normalize back at 1500 eventually.

Edit: Looks like everyone is blaming the shorties:D Maybe tiny cuda cards behave differently than the Fermi monsters... who knows? Also I'd like to add:
No OC, No Lunatics and proc@80%