Astro Pulse 6 has a problem due to users taking the work but just not processing it. See this one user that has hung me out to dry "Rduke".

http://setiathome.berkeley.edu/show_host_detail.php?hostid=6566176

In this listing you will see my completed WU, plus I am almost finished with two other WU's with users waiting on me to finish & one more WU I have not started yet but neither has the other user.

He "RDUKE" won't abort the WU and keeps accepting new WU's and has not finished a one of them.

Now I have to wait until it times out & then the server finally re-sends it out again, and hope that someone will complete it. which will be a couple of months before I will get credit for a WU that took me 45 hours to crunch.

Now I know why nobody wants to crunch Astro pulse 6 because the WU's are too big and you will wait forever to finally get credit for the work being processed.

After I finish my four WU's I will not be accepting anymore unless something is fixed.

not everyone processes work at the same speed. If we all had the same processors and GPU's and everyone ran the project 24/7 then yes if there were delays you could call someone out. However, this is voluntary work. If that person chose to leave and not complete the work so be it. The work just gets reassigned to another person.

I looked at my pending tasks the other day. I have over 1000 pending. I saw that the oldest WU waiting on a wingman is from the middle of December.

I'm sorry if this bursts your bubble since the v6 AP's have only been out a couple weeks. Consider this a lesson in patience. You'll get your credit you'll just have to wait

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I know what you are trying to say, but you didn't read what I was saying. This person has a faster CPU plus a GPU that I don't have & has chosen to accept the work but has not completed a single one. Only finishes the WU's that run real fast on his GPU.

Yeah, so I wait till April 10th when he deadlines out & then when it gets reassigned I have to hope some one will finish it a month or so down the road.

If it was a hour or two of crunching I wouldn't mind but when you run the CPU for 55 hours & get a gigantic 609 points for credit, is it really worth the time?

So of the four WU's I got, I finished two (one pending) & I have two more to finish with a total 102 hours left to crunch.

1000 pending for you, yup! That has got to get you wondering where is everyone that is processing the WU's

I see that you have been doing this since 2000 & I started in 2003 the WU's were a lot smaller back then, I used a Pentium 3 @600 Mhz. back then & it would be a 20 minute crunch per unit of work. Now I'm running a I7 8 processor @2.2 Ghz with a 3.1 turbo mode on a laptop & these WU's have taken the fun out of it all.

Thanks for your reply.

---

Is there any evidence that this person has actually received the workunit in question? Just because you see that it was assigned to him, doesn't mean that he actually received it. We often call these "ghost" workunits - workunits the server thinks it sent to a client but the client did not actually receive.

The solution to ghost workunits is to have the client and server compare their lists of workunits they think they should have and send any missing ones. This functionality is actually built into BOINC, but has been turned off by SETI because it is too resource intensive for this project - they simply don't have powerful enough database servers to support the number of hosts connected and constantly performing this routine check.

Well, when I click on the work unit & then the person's computer id & then show his tasks all work units show up with status. What else can I say?

How would a ghost unit not show up on the users work task info?

How about I do a "print screen" & paste it up here to show everyone what I am trying to get across?


http://setiathome.berkeley.edu/results.php?hostid=6566176&offset=0&show_names=0&state=0&appid=12

Couldn't do the print screen for some reason, look at the above link.

Is there any evidence that this person has actually received the work unit in question? Just because you see that it was assigned to him, doesn't mean that he actually received it. We often call these "ghost" workunits - workunits the server thinks it sent to a client but the client did not actually receive.

The solution to ghost workunits is to have the client and server compare their lists of workunits they think they should have and send any missing ones. This functionality is actually built into BOINC, but has been turned off by SETI because it is too resource intensive for this project - they simply don't have powerful enough database servers to support the number of hosts connected and constantly performing this routine check.

Ghost workunits happen more frequently when SETI's outgoing traffic is maxed out for long periods of time.


The list you're looking at is what the server thinks it sent to the user - it is not reflective of what the user actually received. The only way to find that out would be for the user to check their local BOINC Manager and check to see if they have it.


This is one of the reasons why investigating what other users have in their task list is not a good way to determine what they actually have on hand and/or whether they're able to handle such a workload.

How would a ghost unit not show up on the users work task info?

Simply because although the server did appoint that work to him, his BOINC didn't download it. It could well still be trying to download, but since the Seti's data connection has been maxed out pretty much ever since last Tuesday, this could be problematic for him. We don't know what kind of internet he has, how fast it is.

But even then, Seti's upload (download to us) is max 100Mbit, which is 12.5MB that it can upload at the same time. Astropulse tasks are 8MB, so you just need 2 of them to completely saturate the upload connection, for as long as any one person is downloading that work. However, it's not just you, us and him doing this, but a lot of other people, which is why the downloads are just coming in very slowly. I've seen Multibeams come in at 1.1KB/sec. And that on my 120Mbit connection.

I've sat here nursing the work in, hitting Retry Now on the transfers whenever possible. Also on Astropulse tasks. And they also came in at between 1.1KB/sec and 23KB/sec. Only for the connection to then taper out and BOINC going back to a long deferral before going for a retry. So if that's happening to him....

If you really want to know all of that, or if he's ignoring the work, or that perhaps his BOINC is overworked and doing other work first, why not send hima private message? If you ask nicely, he may answer you.

---

I know what you are trying to say, but you didn't read what I was saying. This person has a faster CPU plus a GPU that I don't have & has chosen to accept the work but has not completed a single one. Only finishes the WU's that run real fast on his GPU.

His CPU is faster than yours? He has a laptop with 8 HT cores at 1.6 ghz. What do you have? He has completed a couple CPU WU's with really long completion times. Nearly 2 hours for a VHAR, for instance. My guess is that 7 of his 8 cores are doing the AP work, and his laptop gets shut off alot because it's overheating. As a relative new user (this month) he surely lacks the experience you have in getting things set up correctly.

---

I know what you are trying to say, but you didn't read what I was saying. This person has a faster CPU plus a GPU that I don't have & has chosen to accept the work but has not completed a single one. Only finishes the WU's that run real fast on his GPU.

CPU type GenuineIntel
Intel(R) Core(TM) i7 CPU Q 720 @ 1.60GHz [Family 6 Model 30 Stepping 5]
Number of processors 8
Coprocessors NVIDIA GeForce 310M (512MB) driver: 187.83


Yes faster than mine with the GPU sure the CPU is a bit slower, but the GPU is way faster than my "Sandy Bridge GPU" which is not supported.

Yup, I have noticed that running full out isn't the best way to crunch on any "CPU Grid computing" & I have used every precaution to keep my CPU running without over heating it. The max I will allow is 165 degrees F although supposedly I can go to 212 degrees, but why would I do that?

I didn't notice that this other person has only been on for a month, I may contact him "nicely" & suggest cooling problems & other good things to help him out. These laptops do get HOT!

Thanks for your informing reply to me, I do appreciate it.


His CPU is faster than yours? He has a laptop with 8 HT cores at 1.6 ghz. What do you have? He has completed a couple CPU WU's with really long completion times. Nearly 2 hours for a VHAR, for instance. My guess is that 7 of his 8 cores are doing the AP work, and his laptop gets shut off alot because it's overheating. As a relative new user (this month) he surely lacks the experience you have in getting things set up correctly.

Gatekeeper,

I sent a nice message to the user and gave him some "cooling" issues like I use for a laptop. I hope he responds to me, it really is a bear with a laptop & no real "GPU" the heat factor is horrible!

Thanks Man!

Kurt

I know what you are trying to say, but you didn't read what I was saying. This person has a faster CPU plus a GPU that I don't have & has chosen to accept the work but has not completed a single one. Only finishes the WU's that run real fast on his GPU.

CPU type GenuineIntel
Intel(R) Core(TM) i7 CPU Q 720 @ 1.60GHz [Family 6 Model 30 Stepping 5]
Number of processors 8
Coprocessors NVIDIA GeForce 310M (512MB) driver: 187.83


Yes faster than mine with the GPU sure the CPU is a bit slower, but the GPU is way faster than my "Sandy Bridge GPU" which is not supported.

Yup, I have noticed that running full out isn't the best way to crunch on any "CPU Grid computing" & I have used every precaution to keep my CPU running without over heating it. The max I will allow is 165 degrees F although supposedly I can go to 212 degrees, but why would I do that?

I didn't notice that this other person has only been on for a month, I may contact him "nicely" & suggest cooling problems & other good things to help him out. These laptops do get HOT!

Thanks for your informing reply to me, I do appreciate it.


His CPU is faster than yours? He has a laptop with 8 HT cores at 1.6 ghz. What do you have? He has completed a couple CPU WU's with really long completion times. Nearly 2 hours for a VHAR, for instance. My guess is that 7 of his 8 cores are doing the AP work, and his laptop gets shut off alot because it's overheating. As a relative new user (this month) he surely lacks the experience you have in getting things set up correctly.