I was browsing my Validation Pending tasks of my SETI@home v7 tasks. I did notice that 10 of my oldest Pendings are from computers/users that did attach to S@H project and did abandon/uninstall it.

At that time when I did this list I did have 245 Pendings, these are ten oldest ones:

Unit		Sent		Deadline	Last contact	Tasks in progress
1566777435	14 Aug 2014	9 Oct 2014	14 Aug 2014	2			
1567917890	15 Aug 2014	8 Oct 2014	15 Aug 2014	19			
1569588988	17 Aug 2014	9 Oct 2014	17 Aug 2014	64			
1569588988	17 Aug 2014	9 Oct 2014	17 Aug 2014	64			
1571712823	20 Aug 2014	13 Oct 2014	20 Aug 2014	69			
1572032343	20 Aug 2014	12 Oct 2014	20 Aug 2014	85			
1572531021	21 Aug 2014	13 Oct 2014	21 Aug 2014	6			
1572758934	21 Aug 2014	5 Oct 2014	21 Aug 2014	75			
1576074890	25 Aug 2014	18 Oct 2014	26 Aug 2014	75			
1576511841	25 Aug 2014	17 Oct 2014	25 Aug 2014	81

As You can see, no of those computers has contacted S@H since first contact. And they do have 476 work units in progress.

So, in bigger scale, how many of these work units in progress are from computers that will never send them back? How many of those 3,258,255 Results out in the field are going to be Timed Out? Average Deadline for those 10 work units listed above is about 52 days, so multiply it in bigger scale.

How much of those "sleeper" units will stress S@H servers databases? How much smaller those databases will be if deadline is decreased to, well, like 20 days?

---

"Please keep Your signature under four lines so Internet traffic doesn't go up too much"

- In 1992 when I had my first e-mail address -

 
20 days may be too short for:
- old computers (like my 500 MHz K6-2+)
- slow Android devices
- when you go on 2-3 week vacation
- people with 'Suspend work while computer is in use'
 

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Yeah, but what percentage of WU's are actually uncompleted after 20 days? Maybe 5% of them? Some trickle in later, some time out and get reissued. '

But no matter what the exact number, I don't see it reducing the size of the database by more than a few percent?

The vast majority do get completed in the first week by fairly reliable machines, so I just don't see how there will be any worthwhile gain to be had.

Ian

I think having a server side feature for a user to mark their host as no longer in use. So that any In Progress tasks would get marked as abandoned, aborted, or timed out would be an ideal solution. That kind of feature probably should also not allow the host to download any new tasks while set. Which might be able to be done by setting the daily limit to 0. Sort of like when the daily limit is reduced when several errors occur.

Really since they implemented the task limits I have seen the number of really old pending tasks drop quite a lot.
Out of all my ~400 pending tasks I have 31 that are older than 30 days & most of those are ones stuck in limbo. So cleaning those up on a regular basis would probably have a greater effect than reducing the deadlines and forcing slower machines out.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Out of all my ~400 pending tasks I have 31 that are older than 30 days

That's about 8% of the total.

So assuming you return work on the average time, then only 1/2 your tasks go into pending at all. So maybe 4% of the total are actually lost?

Out of all my ~400 pending tasks I have 31 that are older than 30 days

That's about 8% of the total.

So assuming you return work on the average time, then only 1/2 your tasks go into pending at all. So maybe 4% of the total are actually lost?

My Pending to Valid numbers are generally pretty close to 50/50. It is actually 54.5/45.5 right now, but it sometimes it goes the other way too.
With my current numbers, & only looking at my pending tasks, 2.5% are stuck in limbo. That's about half of my tasks older than 30 days. So just using 5% as an average across the project. That would give us about 158941 MB & 11223 AP that are in limbo or ones that may time out. With about half, 79471 MB & 5612 AP, of them going in the "may time out" column. I imagine not all of the tasks over 30 days old would time out & get sent to another host, but even if it is over 80% of the 79471 MB & 5612 AP. That is a pretty small number for the whole project.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

A Simple Question:

Why Seti@Home v7 tasks have longer deadlines than Astropulse v7?

Runtimes in S@H v7 are VERY MUCH SHORTER than AP v7.

---

"Please keep Your signature under four lines so Internet traffic doesn't go up too much"

- In 1992 when I had my first e-mail address -

A Simple Question:

Why Seti@Home v7 tasks have longer deadlines than Astropulse v7?

Runtimes in S@H v7 are VERY MUCH SHORTER than AP v7.

The main reason is to allow users with very old (read slow) computers running part time to still be able to contribute to SETI.

---

Boinc V7.2.42
Win7 i5 3.33G 4GB, GTX470

I look at this problem the other way - many other projects are in just such a rush to get work through they do not allow for going on holiday, having a slow computer or suing the computer, or shutting it down for more than a few micro seconds so actually suffer from a higher proportion of re-sends than S@H does.

From another thread:

My Atom N450 has finished it first two AP v7 7.00 Wu's, only 1.3 million seconds each (around 375 hours), that'll give NewCredit something to think about.

Which would have failed on a time out on just about every other project....



(If you want a really long turn around take a look at the CPDN which regularly sends out work with three or four moths deadlines for the first return)

---

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

(If you want a really long turn around take a look at the CPDN which regularly sends out work with three or four moths deadlines for the first return)

Actually, some CPDN WU have a 9 month deadline and they take less time to process then the 3 to 4 month one.

---

Boinc V7.2.42
Win7 i5 3.33G 4GB, GTX470

The main reason is to allow users with very old (read slow) computers running part time to still be able to contribute to SETI.

Yes, I know that explanation.

My Atom N450 has finished it first two AP v7 7.00 Wu's, only 1.3 million seconds each (around 375 hours), that'll give NewCredit something to think about.

AP v7 units took about 15 days running 24/7 (I assume).

Let's say that MB v7 runtimes are 1/3 of AP v7. So that Atom would take about 5 days to complete MB v7 unit. So when deadline for MB v7 is over 50 days, this Atom has to do work about 2,5h/day.

How many computers like that Atom are in S@H? And how much they contribute to S@H? Electricity is surely wasted...

It is just frustrating to see that some of MB v7 units has to wait over 50 days to be validated because someone did attach to S@H and then never again contacted servers.

Anyway, I let this case rest.

---

"Please keep Your signature under four lines so Internet traffic doesn't go up too much"

- In 1992 when I had my first e-mail address -

...Thus sayeth the man who has aborted a fair number of AP7 tasks....

---

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

...Thus sayeth the man who has aborted a fair number of AP7 tasks....

How Your post relates that this thread is about talking long deadlines in S@H v7 units?

Yes, I did abort several (about 50) number of AP7 tasks. To free them to another S@H user to crunch them. Not to keep them knowing I can't finish them before deadline.

All of them have been crunched and validated.

When you abandoned those tasks you had no idea when the next, or the next, or the next user would process them. You INCREASED the risk that there would be an extended validation period for those tasks.
If you are so concerned about the turn-around time for AP tasks then why not set the AP option for all your crunchers to reject them. That way you wouldn't induce any delay in their processing.

---

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

It is just frustrating to see that some of MB v7 units has to wait over 50 days to be validated because someone did attach to S@H and then never again contacted servers.

Why would that matter, unless you happen to be an Atom cruncher yourself :-)
The score pending from a few MB tasks isn't going to make much difference for most.

...Thus sayeth the man who has aborted a fair number of AP7 tasks....

What's wrong with aborting? I often do it for tasks I haven't started, in case I need to do some maintenance (and there's a good chance those task will be completed quicker, as they can be given to some faster crunchers).

When you abandoned those tasks you had no idea when the next, or the next, or the next user would process them. You INCREASED the risk that there would be an extended validation period for those tasks.
If you are so concerned about the turn-around time for AP tasks then why not set the AP option for all your crunchers to reject them. That way you wouldn't induce any delay in their processing.

Why You still are talking about AstroPulse version 7 tasks?

This thread is about Seti@Home MultiBeam version 7 tasks. And it's long deadlines.

---

"Please keep Your signature under four lines so Internet traffic doesn't go up too much"

- In 1992 when I had my first e-mail address -

Because YOU brought them up!

---

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

Because YOU brought them up!

And You can stop sending messages which are not inside topic.

---

"Please keep Your signature under four lines so Internet traffic doesn't go up too much"

- In 1992 when I had my first e-mail address -

The main reason is to allow users with very old (read slow) computers running part time to still be able to contribute to SETI.

Yes, I know that explanation.

My Atom N450 has finished it first two AP v7 7.00 Wu's, only 1.3 million seconds each (around 375 hours), that'll give NewCredit something to think about.

AP v7 units took about 15 days running 24/7 (I assume).

Let's say that MB v7 runtimes are 1/3 of AP v7. So that Atom would take about 5 days to complete MB v7 unit. So when deadline for MB v7 is over 50 days, this Atom has to do work about 2,5h/day.

Only takes around a day to do MBv7 Wu with the Stock 7.00 app, a bit faster than my old PIII 800E,
If i crunch Optimised AP on the Atom N450, Wu's will take around 70 to 80 hours, against the heading up to 500 hours the PIII will take.

How many computers like that Atom are in S@H? And how much they contribute to S@H? Electricity is surely wasted...

While it might be slow at doing Crunching, it's also uses a lot less power, around 35 Watts against the 100 Watts the PIII uses,
There will be thousands of slow computers like this, including all the android devices, they are certainly no faster at crunching, But only use a couple of Watts.

Claggy

And not forgetting all those that only crunch for a short time a day, or use the screen saver because of its interesting animations of the search - from memory the screen saver is ten times or more slower than the same PC running in "compute only" mode.

---

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