... but the BM shows me a est. runtime for 1021 hours, aprox. 42.5 days, for this WU.

A up- or downgrade don't change the estimated time for your WUs. Therefor you have to check the duration_correction_factor in der client_state.xml. Close the BOINC manager and Client. Search in your BOINC data directory for the client_state.xml. Make a backup copy. Search for gpu in the file. Search for the next duration_correction_factor line in this file. There you find a entry in this form:

<duration_correction_factor>1.12345678</duration_correction_factor>

In dependence of your estimated time (eg. 100 hours instead of 10 hours) change the value to value/your correction (e.g. 1.12345678 / 10 = 0.01234568). Now save the file and restart BOINC. The estimated time should be 10% of the old value.

If you have more trouble as before, use the backup file.

---

Hi, you are not getting work because you have less than 256 of device memory.

2008-12-19 15:24:42.5519 [PID=21662] [debug ] CUDA mem < 256MB


gdf

Hi, you are not getting work because you have less than 256 of device memory.

2008-12-19 15:24:42.5519 [PID=21662] [debug ] CUDA mem < 256MB


gdf

Hi GDF, I don't know if you were referring to me in your post above, but my device has 512MB memory...

Hi, you are not getting work because you have less than 256 of device memory.

2008-12-19 15:24:42.5519 [PID=21662] [debug ] CUDA mem < 256MB


gdf

Hi GDF, I don't know if you were referring to me in your post above, but my device has 512MB memory...

Do you use XP 64bit? Then maybe your memory is used and you have to restart your PC. Otherwise forget this hint.

Are you sure, you have a 8400 GS with 512 MB? Normally this card has only 128 or 256 MB VidMem. Or would you say, you have 512 MB RAM in your PC? That's not the RAM of the graphic card.

---

It's an XP 32 bit... :)

I have 1GB of RAM in my pc and as I can see at the adapter information page, I have a GeForce 8400 GS with Memory size of: 512MB.

I have tried BOINC 6.4.3, 6.4.5, and 6.5.0 and I just tried resetting the npc error count to 0; none of it is working. My VRAM is 1 G, PC memory is 2.9 of installed 6 G (Tri-Channel)

WIndows XP Pro with SP 3

i7 processor.

And I did do at least 2 tasks successfully.

I have had a number of tasks failing with Computation error. I am not at all sure if the message is accurate or is simply a default server error message that is covering a multitude of sins.

What the heck, I just got a Cosmology download error (yes, there are still errors, though thankfully the INI file is now just missing instead of only D/L to 16%).

Anyway, I know my system was, and is, functional in that it has done work for the two projects that issue GPU work and that work has validated (an additional task at SaH has validated and one has a difference with the wingman where I found a pulse that they did not).

Anyway ... Guys, you need to look at the server code to see why the server is refusing to issue work to so many of us. It may be related to the issue policy. If you have a mix of tasks and they can only be issued to selected types of machines, you may be filling the queue with tasks of all one type and there is not enough room for the types of task that I and the others are asking for (or are capable of running.

Perhaps it is time that the feeder get a change so that instead of inserting tasks into one ring (or list) that the feeder should create a list type for each type of task that is to be issued. Then, when the request comes in the feeder can issue a task off of the correct list.

{edit}I just sent a post to the boinc dev mailing list{/edit}

Sometimes I had also the described error. But never for a long time. When I get the message with no work available, I make immediately a second call for work to get this situation:

Fresh provoked message:
19.12.2008 18:40:13|GPUGRID|Message from server: Not sending work - last request too recent: 27 sec

After this message the 24 hours delay is away and my box can make regular calls. One or two hours later I got in any case new work. But you can't force this things.

---

That wasn't working for me but if I suspended all other activity on my other projects I could then get gpugrid WU. Strange!!!

Pat

That wasn't working for me but if I suspended all other activity on my other projects I could then get gpugrid WU. Strange!!!

Pat

As I've written in another post ... how is the size of your work buffer? I have 2 days ... and I let my boxes call ever 0.02 days.

---

Mine is set to 4 days buffer and always connected. I think I might change that to every three hours (0.125)

Pat

Sometimes I can't let my fingers away from keyboard to test something ... ;)

I got this error (most of your know them ...)

20.12.2008 00:43:01|GPUGRID|Sending scheduler request: Requested by user. Requesting 48471 seconds of work, reporting 0 completed tasks
20.12.2008 00:43:06|GPUGRID|Scheduler request completed: got 0 new tasks
20.12.2008 00:43:06|GPUGRID|Message from server: No work sent
20.12.2008 00:43:06|GPUGRID|Message from server: Full-atom molecular dynamics on Cell processor is not available for your type of computer.
20.12.2008 00:43:06|GPUGRID|Message from server: Full-atom molecular dynamics for Cell processor is not available for your type of computer.

I set project to no new work and make a new call.

20.12.2008 00:44:52|GPUGRID|Sending scheduler request: Requested by user. Requesting 0 seconds of work, reporting 0 completed tasks
20.12.2008 00:44:58|GPUGRID|Scheduler request completed: got 0 new tasks

Now all error entrys in the client_state.xml are on 0. I change to new work, every other project on suspend and make a new call.

20.12.2008 00:46:33|GPUGRID|Sending scheduler request: Requested by user. Requesting 677441 seconds of work, reporting 0 completed tasks
20.12.2008 00:46:39|GPUGRID|Scheduler request completed: got 1 new tasks

Voilá! I get my fourth WU short time after I got this disliked failure. Maybe you can go the same way ...

---

Kokomiko,

Yeah, it now seems some massaging will get new WU's. I at least can get enough to sleep (away from my computer) an not have to hover over my rig during the day.

Still babysitting though!!

Pat

I have a GeForce 8400 GS with Memory size of: 512MB.

Idan,

your computer summary shows your card as 255 MB. The manufacturers have the bad habit of including the "turbo memory", i.e. system memory which can be accesses by the card, into the memory amount, because higher numbers are perceived as better. And generally it doesn't have any consequences, because these cards are too slow for games anyway..

MrS

---

Scanning for our furry friends since Jan 2002

I have a GeForce 8400 GS with Memory size of: 512MB.

Idan,

your computer summary shows your card as 255 MB. The manufacturers have the bad habit of including the "turbo memory", i.e. system memory which can be accesses by the card, into the memory amount, because higher numbers are perceived as better. And generally it doesn't have any consequences, because these cards are too slow for games anyway..

MrS

Thank you MrS, but, does it mean that I won't be able to crunch for GPUGrid using this CUDA capable card?

Unfortunately, yes! And it would also be very slow, possibly too slow to meet the 4 day deadlines of the current WUs. You may want to take a look at the seti@home @ GPU beta.

MrS

---

Scanning for our furry friends since Jan 2002