Hi. Something a bit odd happened last night. I think the scheduler has gone nuts or something. Usually when i request more work units, at most I get 10 at a time.
Last Night I got 50+ work units. a mix of seti ver 7.00 & cuda ver 3.20. I think it migth have sent even more, had I not had a 30 meg per 7 days bandwith limt in place. So I hope to process them all before their deadlines. Come to that, I wonder how avg will cope with cuda 3.20. The last lot of cuda 3.20's that were sent to me all ended in error after avg's Identy protection module decided cuda 3.20 were a virus. I have added the cuda 3.20 executiuble to the exceptions list & now pray that this time that avg will let cuda crunch all day & night long without cooking any part of my pc or ending the tasks in error. Until I clear this backlog, I will turn on no new tasks in bonic. After listing the tasks by report deadline, I decided to override Bonics choices on what units to process first & suspended all tasks except those that are due
on 01/08/2013 (18 tasks are due on that date>. The next deadline date after that is 02/09/13 & more tasks due on 03/09/13
Any thoughts welcome.
Rae

As you appear to have handled the AVG False Positive situation http://setiathome.berkeley.edu/forum_thread.php?id=71784&postid=1373641

The download of 50 or so workunits is normal for SETI V7.
I've got 55 SETI V7 workunits myself

SETI V7 It's a hybrid application essentially using the CPU for the floating point operations and passing the straightforward integer operations to the faster GPU.

at least that's my experience with my AMD/ATI setup.

The Improved throughput will be such that the computer should comfortably through the workunits required for your first deadline.

Hi. Something a bit odd happened last night. I think the scheduler has gone nuts or something. Usually when i request more work units, at most I get 10 at a time.
Last Night I got 50+ work units. a mix of seti ver 7.00 & cuda ver 3.20. I think it migth have sent even more, had I not had a 30 meg per 7 days bandwith limt in place. So I hope to process them all before their deadlines. Come to that, I wonder how avg will cope with cuda 3.20. The last lot of cuda 3.20's that were sent to me all ended in error after avg's Identy protection module decided cuda 3.20 were a virus. I have added the cuda 3.20 executiuble to the exceptions list & now pray that this time that avg will let cuda crunch all day & night long without cooking any part of my pc or ending the tasks in error. Until I clear this backlog, I will turn on no new tasks in bonic. After listing the tasks by report deadline, I decided to override Bonics choices on what units to process first & suspended all tasks except those that are due
on 01/08/2013 (18 tasks are due on that date>. The next deadline date after that is 02/09/13 & more tasks due on 03/09/13
Any thoughts welcome.
Rae

Since you have your computer(s) hidden we can't see what tasks you are attempting to work on. Are you sure that you have the dates right? It would seem to me that if they are due in 2014 instead of 2013 as stated then you should have more then enough time to process them. If the dates are correct, then they are already late and you should abort them as you will never get credit for them.

Another question that I have is why just exclude the Cuda 3.20 .exes instead of the entire data directory? If for some reason AVG decides that another type of .exe and/or wu is a virus in the future then you have to go back and exclude that type also, while if you exclude the entire data directory it's once and done.

---

I don't buy computers, I build them!!

Thanks for your reply. my desktop is a museum piece compared to todays windows 8 machines, but it has run win 8. Its an old Hp, with 2 cores & a nvida 9400GT with 512K on the graphics card. the main board only supports a max of 2 gig of ram & short of replacing it, I am stuck with that hardware restriction. Thanks to vista home basic & later versions of windows, I can use an sd card as extra ram using windows ready boost to improve the performance a bit. 18 tasks in 18 days is the easy bit, completing the other 40ish tasks in 30 days might be pushing it, but I can but try & see how close i get. who knows, it might even surprise me if it completes it early. 2 ordinary tasks at a time plus the gpu & 0.08 of 1 cpu for the cuda tasks. As for the wing-people for the abandoned tasks, I am unsure why Bonic did that, I didn't tell it to give up on those tasks, so sorry if you are stuck waiting for vslidation. 1 minute it was crunching away nicly, the next minute Bonic abandoned those tasks & Grabbed the 50+ new 1's.

---

A city destroyed by an earthquake is an opportunity to Rebuild, redeign & make it a better place to be. Better, stronger, faster like the 6 Million Dollar Man

The workunits are due in 2013 not 14. I Have just ticked the box so my desktop / tasks should now be viewible. As for why not add the entire directory to the exceptions, I have tryed but it won't for what ever reason. For now all the required executibles have exceptoions.

{edit} The dates posted earlier in this thread are in New Zealand date format dd/mm/yy not mm/dd/yy.

I have already got execeptions in place for cuda 4.2 & 5.0 as well as setiathome v7.00

---

A city destroyed by an earthquake is an opportunity to Rebuild, redeign & make it a better place to be. Better, stronger, faster like the 6 Million Dollar Man

SETI V7 It's a hybrid application essentially using the CPU for the floating point operations and passing the straightforward integer operations to the faster GPU.

Where did you get such wrong impression?

All apps for SETI V7 are CPU only or GPU only, there is no 'hybrid application'

---

 


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

... for the abandoned tasks, I am unsure why BOINC did that, I didn't tell it to give up on those tasks ...

This happens (for still unknown reason) if you have very bad Internet connection.

(your Computers are still shown 'hidden', it takes some time (1 hour?) for the setting ('Should SETI@home show your computers on its web site? yes') to come into effect)

How much of those ~50 SETI V7 tasks are for CUDA?
Your GPU probably can do 20-30 tasks per day (?)

---

 


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

How much of those ~50 SETI V7 tasks are for CUDA?
Your GPU probably can do 20-30 tasks per day (?)

25 of the 50+ tasks are cuda 3.20. They Vary in size in estimated times to complete. On an average day, the gpu might complete 1 or 2 tasks & the cpus might complete up to 4 tasks, unless its an astroplulse unit, which can take up to a week, depending on how long the computer is turned on & would otherwise be "twiddling its thumbs" before I turn it off for the night.

As for my internet connection, most of the time its quite good, but the city I live in is rebuilding its infrastructure from the sewer pipes up in some places, not to mentsion houses/ businesses that were destroyed in earthquakes that started 2 3/4 years ago & still happen on a regular Basis http://www.christchurchquakemap.co.nz/
As such, sometimes when they are working on the infrastructure, , a lot of peoples internet etc goes off line for short or sometimes hours at a time, while magor repairs are in progress.
The Old joke of "Did the Earth Move for you?" has had a new meaning here since sept 4 2010, when the first quake struck my city.

(edit) Why would my internet connection cause Bonic to abandon already downloaded tasks? Its not like they had timed out or anything like that or had their download interupted.

---

A city destroyed by an earthquake is an opportunity to Rebuild, redeign & make it a better place to be. Better, stronger, faster like the 6 Million Dollar Man

The Old joke of "Did the Earth Move for you?" ...

Joking... That´s happens with me every day after few shoots... and here we don´t have earthquakes.

Back to topic. Check your cache settings, you could have a large number of days on the cache, try a lower number. In most of my lowest hosts i use a 3+0,1 day cache. After the change of the server to Colo we realy don´t need a large cache on SETI anymore.

---

Cache setting:
Maintain enough tasks to keep busy for at least
(max 10 days). 2 days
... and up to an additional 1 days

---

A city destroyed by an earthquake is an opportunity to Rebuild, redeign & make it a better place to be. Better, stronger, faster like the 6 Million Dollar Man

Cache setting:
Maintain enough tasks to keep busy for at least
(max 10 days). 2 days
... and up to an additional 1 days

I don´t realy understand why, but try a small amount (0.1 for example) in the adittional cache... not the main cache, 2 days is a good number i use 1 on my fastest hosts... that could help, at least works here.

---

On an average day, the gpu might complete 1 or 2 tasks & the cpus might complete up to 4 tasks (SETI V7)

I now can see your results (there is only one for CUDA and one for CPU)

The CUDA task finished in 16,430 s (4.5 h) so your GPU can do ~5 tasks/day (by 'day' I mean 24 h)
The CPU task finished in 39,176 s (11 h) so your (2 core) CPU can do ~4 tasks/day

This is if BOINC run tasks all the time (24/7) and computing on CPU & GPU happens while you are using the computer.
(yes, not all tasks complete in the same time)

---

 


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

I guess I will try reducing the cache settings a bit more. As for running Bonic 24/7, not really practical since my Desktop is in my bedroom & the fan would keep me awake. Depending on what I am doing on the computer, most of the time Bonic is allowed to work while I am, as long as non Bonic is Less than 60% of cpu time. Bonic is also set to to suspend work durring defrags of the hard drive. Speaking of sleep, the local time is close to 2:20am & I guess I should call it a nite. Thanks all for your kind & helpful thoughts.

---

A city destroyed by an earthquake is an opportunity to Rebuild, redeign & make it a better place to be. Better, stronger, faster like the 6 Million Dollar Man

(edit) Why would my Internet connection cause BOINC to abandon already downloaded tasks? Its not like they had timed out or anything like that or had their download interrupted.

I already said: for reason unknown exactly

Server have anti-cheating check (to not allow several computers to pretend be only one (by using the same HostID))

Every sched_request have <rpc_seqno>
So the theory is - your second request somehow arrives at the server before your first request.
Server see <rpc_seqno> go backwards and punish the computer (for supposed cheating) by abandoning tasks.

---

 


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

SETI V7 It's a hybrid application essentially using the CPU for the floating point operations and passing the straightforward integer operations to the faster GPU.

Where did you get such wrong impression?

All apps for SETI V7 are CPU only or GPU only, there is no 'hybrid application'

I made my assumption after observing the completion of 206 SETI 7 (7.03ati) on my AMD/ATI paired machine.

The GPU is a HD 6670 so it can't do floating point operations, so I assumed the floating point operations were being done by the CPU for the 0.03-0.05% of the CPU resource it was requesting.

I also assumed that Version 7 of SETI aside from the extra bit of science it was performing was to better utilise lower end GPUs (ones that do not do floating point operations themselves) by allocating instructions to the best resource to utilise them so simpler integer operations to the GPU which can perform those much faster than the CPU and then the floating point operations are done by the CPU arithmetic [processing] unit because those are the instructions that are left. So I see it as a hybrid application because it's not solely performing all the instructions on a single resource

For the GPUs that perform floating point operations also then I guess there is still the pure CUDA or ATI V7 equivenents to the Version 6 Applications.

But then I've only ever had Astropulse V6 (6.03) performed by low end ATI GPUS if it wasn't a CPU only as the only other hybrid application experience and SETI V7 behaves in a very similar fashion from my point of view.

All GPGPUs are nothing more than really high-speed floating point processors. Its simply a matter whether they support single precision or double precision math. SETI@home only requires single-precision.

[Edited to add] According to this chart, the Radeon HD 6670 supports single-precision floating point math.

---

Hi. Something a bit odd happened last night. I think the scheduler has gone nuts or something. Usually when i request more work units, at most I get 10 at a time.
Last Night I got 50+ work units. a mix of seti ver 7.00 & cuda ver 3.20. I think it migth have sent even more, had I not had a 30 meg per 7 days bandwith limt in place. So I hope to process them all before their deadlines. Come to that, I wonder how avg will cope with cuda 3.20. The last lot of cuda 3.20's that were sent to me all ended in error after avg's Identy protection module decided cuda 3.20 were a virus. I have added the cuda 3.20 executiuble to the exceptions list & now pray that this time that avg will let cuda crunch all day & night long without cooking any part of my pc or ending the tasks in error. Until I clear this backlog, I will turn on no new tasks in bonic. After listing the tasks by report deadline, I decided to override Bonics choices on what units to process first & suspended all tasks except those that are due
on 01/08/2013 (18 tasks are due on that date>. The next deadline date after that is 02/09/13 & more tasks due on 03/09/13
Any thoughts welcome.
Rae

Since you have your computer(s) hidden we can't see what tasks you are attempting to work on. Are you sure that you have the dates right? It would seem to me that if they are due in 2014 instead of 2013 as stated then you should have more then enough time to process them. If the dates are correct, then they are already late and you should abort them as you will never get credit for them.

Another question that I have is why just exclude the Cuda 3.20 .exes instead of the entire data directory? If for some reason AVG decides that another type of .exe and/or wu is a virus in the future then you have to go back and exclude that type also, while if you exclude the entire data directory it's once and done.

The tasks are due in early September from what I see.

---

Server have anti-cheating check (to not allow several computers to pretend be only one (by using the same HostID))

Every sched_request have <rpc_seqno>
So the theory is - your second request somehow arrives at the server before your first request.
Server see <rpc_seqno> go backwards and punish the computer (for supposed cheating) by abandoning tasks.

[/quote]

ok, but since seti is only running on my desktop, who & how has someone gottem hold of its host id?

---

A city destroyed by an earthquake is an opportunity to Rebuild, redeign & make it a better place to be. Better, stronger, faster like the 6 Million Dollar Man

The tasks are due in early September from what I see.

scroll down a bit further <next page and the 1 after> and you will see 18 tasks due on 31 july 20:30 UTC <N.Z. time 1 aug, since I live at GMT +12>

---

A city destroyed by an earthquake is an opportunity to Rebuild, redeign & make it a better place to be. Better, stronger, faster like the 6 Million Dollar Man

I think you have just experienced a glitch in the scheduler, and been given extra work. Seems to happen. Don't worry about the 2 week deadline jobs. If the system thinks it's not going to get to them in time it will switch those to high priority and take care of them first, then come back to the longer deadlines. I would expect your machine to be able to get through the 50 units, even if it only does a couple per day.

I also see the temperature in Christchurch is expected to go down to 0 C tonight, so feel free to leave the PC on a bit longer and huddle around it for warmth :D

Ian (From an equally cold Stratford)