Links available from
gpugrid.net/join.php

Please also do a project reset. The client moved from cpu time to elapsed time estimates based on flops, so we had to correct it on the server too. This create confusion between old and new clints.

BOINC will only try to debug problems on 6.4.x. So, after the update, if you still don't get WUs, write here your host id, so that I can scan it in the log.

many thanks,
gdf

OK ..... took host 17415 40 6.4.2. Reset project after version change - checked DCF manually - it was at 1.000000.
Suspended all other project work - manually updated GPUGrid - got 2 x 1 WU estimate to completion 238 hours.
Third attempt to get more WUs (this host has 2 x 260s) received the response below.

18/12/2008 13:47:13|GPUGRID|Sending scheduler request: To fetch work. Requesting 4497293 seconds of work, reporting 0 completed tasks
18/12/2008 13:47:18|GPUGRID|Scheduler request completed: got 0 new tasks
18/12/2008 13:47:18|GPUGRID|Message from server: No work sent
18/12/2008 13:47:18|GPUGRID|Message from server: Full-atom molecular dynamics for Cell processor is not available for your type of computer.
18/12/2008 13:47:18|GPUGRID|Message from server: Full-atom molecular dynamics on Cell processor is not available for your type of computer.

Not sure where to go from here.

P.

That method doesn't work for me, I've already tried to downgrade to 6.4.2 and also reseting the project. My host ID is 18432.

Reverted to 6.4.2 and still getting no work available messages for host ID 8882

Messages:

Thu 18 Dec 2008 15:28:20 GMT||Starting BOINC client version 6.4.2 for x86_64-pc-linux-gnu
Thu 18 Dec 2008 15:28:20 GMT||log flags: task, file_xfer, sched_ops
Thu 18 Dec 2008 15:28:20 GMT||Libraries: libcurl/7.18.0 OpenSSL/0.9.8g zlib/1.2.3.3 c-ares/1.5.1
Thu 18 Dec 2008 15:28:20 GMT||Data directory: /home/richard/Desktop/Boinc/BOINC
Thu 18 Dec 2008 15:28:20 GMT||Processor: 2 GenuineIntel Intel(R) Core(TM)2 CPU          6700  @ 2.66GHz [Family 6 Model 15 Stepping 6]
Thu 18 Dec 2008 15:28:20 GMT||Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx lm constant_tsc arch_perfmon pebs bts rep_good pni monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr lahf_lm
Thu 18 Dec 2008 15:28:20 GMT||OS: Linux: 2.6.24-19-generic
Thu 18 Dec 2008 15:28:20 GMT||Memory: 1004.55 MB physical, 2.44 GB virtual
Thu 18 Dec 2008 15:28:20 GMT||Disk: 55.84 GB total, 49.99 GB free
Thu 18 Dec 2008 15:28:20 GMT||Local time is UTC +0 hours
Thu 18 Dec 2008 15:28:20 GMT||Not using a proxy
Thu 18 Dec 2008 15:28:20 GMT||CUDA devices found
Thu 18 Dec 2008 15:28:20 GMT||Coprocessor: GeForce 8800 GT (1)
Thu 18 Dec 2008 15:28:21 GMT||Version change (6.4.5 -> 6.4.2)
Thu 18 Dec 2008 15:28:21 GMT|GPUGRID|URL: http://www.gpugrid.net/; Computer ID: 8882; location: (none); project prefs: default
Thu 18 Dec 2008 15:28:21 GMT||General prefs: from GPUGRID (last modified 20-Aug-2008 22:12:44)
Thu 18 Dec 2008 15:28:21 GMT||Host location: none
Thu 18 Dec 2008 15:28:21 GMT||General prefs: using your defaults
Thu 18 Dec 2008 15:28:21 GMT||Preferences limit memory usage when active to 753.42MB
Thu 18 Dec 2008 15:28:21 GMT||Preferences limit memory usage when idle to 904.10MB
Thu 18 Dec 2008 15:28:21 GMT||Preferences limit disk usage to 27.92GB
Thu 18 Dec 2008 15:28:21 GMT||Running CPU benchmarks
Thu 18 Dec 2008 15:28:21 GMT||Suspending computation - running CPU benchmarks
Thu 18 Dec 2008 15:28:52 GMT||Benchmark results:
Thu 18 Dec 2008 15:28:52 GMT||   Number of CPUs: 2
Thu 18 Dec 2008 15:28:52 GMT||   2784 floating point MIPS (Whetstone) per CPU
Thu 18 Dec 2008 15:28:52 GMT||   7879 integer MIPS (Dhrystone) per CPU
Thu 18 Dec 2008 15:29:00 GMT|GPUGRID|Resetting project
Thu 18 Dec 2008 15:29:06 GMT|GPUGRID|Sending scheduler request: Requested by user.  Requesting 172802 seconds of work, reporting 0 completed tasks
Thu 18 Dec 2008 15:29:11 GMT|GPUGRID|Scheduler request completed: got 0 new tasks
Thu 18 Dec 2008 15:29:11 GMT|GPUGRID|Message from server: No work sent
Thu 18 Dec 2008 15:29:11 GMT|GPUGRID|Message from server: Full-atom molecular dynamics for Cell processor is not available for your type of computer.
Thu 18 Dec 2008 15:29:11 GMT|GPUGRID|Message from server: Full-atom molecular dynamics on Cell processor is not available for your type of computer.
Thu 18 Dec 2008 15:29:13 GMT|GPUGRID|Started download of logops3grid.png
Thu 18 Dec 2008 15:29:13 GMT|GPUGRID|Started download of project_1.png
Thu 18 Dec 2008 15:29:15 GMT|GPUGRID|Finished download of logops3grid.png
Thu 18 Dec 2008 15:29:15 GMT|GPUGRID|Finished download of project_1.png
Thu 18 Dec 2008 15:29:15 GMT|GPUGRID|Started download of project_2.png
Thu 18 Dec 2008 15:29:15 GMT|GPUGRID|Started download of project_3.png
Thu 18 Dec 2008 15:29:17 GMT|GPUGRID|Finished download of project_2.png
Thu 18 Dec 2008 15:29:17 GMT|GPUGRID|Finished download of project_3.png

edit #1 ....and the same for host ID 5889 :-(

/edit #2 ....detach and then re-attach host ID 5889 still receives the same no work messages

---

Idem for hostID 16551

Kind regards

Alain

50% succes rate here!

The success:
Host 17875 got a WU at the first attempt and is running with 4 other project's tasks. All 5 tasks are running in high priority mode although nothing is remotely in deadline trouble.

The failure:
Host 117880 CANNOT get a wu now, although with 6.3.21 it never had any trouble.

Both updates were done by suspending all task activity; running the boinc installer; waiting for benchmarks; resetetting GPUgrid; checking that dcf was reset to 1.000000 (it was) before restarting task activity. I've done several manual updates I can't get anything for 17880. It doesn't matter if I suspend the other project either.

I've also tried 6.4.5 on this machine again this afternoon with the same result. First attempt below with other project running; the second with it suspended.

18/12/2008 15:49:14|GPUGRID|Sending scheduler request: Requested by user. Requesting 175 seconds of work, reporting 0 completed tasks
18/12/2008 15:49:19|GPUGRID|Scheduler request completed: got 0 new tasks
18/12/2008 15:49:19|GPUGRID|Message from server: No work sent
18/12/2008 15:49:19|GPUGRID|Message from server: Full-atom molecular dynamics on Cell processor is not available for your type of computer.
18/12/2008 15:49:19|GPUGRID|Message from server: Full-atom molecular dynamics for Cell processor is not available for your type of computer.
18/12/2008 15:56:25|GPUGRID|Sending scheduler request: Requested by user. Requesting 349 seconds of work, reporting 0 completed tasks
18/12/2008 15:56:30|GPUGRID|Scheduler request completed: got 0 new tasks
18/12/2008 15:56:30|GPUGRID|Message from server: No work sent
18/12/2008 15:56:30|GPUGRID|Message from server: Full-atom molecular dynamics on Cell processor is not available for your type of computer.
18/12/2008 15:56:30|GPUGRID|Message from server: Full-atom molecular dynamics for Cell processor is not available for your type of computer.

To keep this machine in work I'm having to go back to 6.3.21. It is looking as if the effect of the latest changes is to make high-end boards unable to get work.

By the way I note the recommended Nvidia drivers have gone back a version or two, is that a mistake or must we also revert those to make any progress with 6.4.2??


Phoneman1

Phoneman1 - is 6.3.21 actually working for you?

Burdett - it was. It is not now!

I can't get anything for the GTX260 machine this afternoon on 6.4.2; 6.4.5; or 6.3.21. It doesn't matter if I suspend the other project or not. Always get the messages I posted earlier in this thread.

8600GTS machine is chugging away in 1 plus 4 mode on 6.4.2 with no real problem at all save for all 5 tasks in high priority for no reason I can see. There is no GPU tasks queued up there just the one that is running.

Phoneman1

..... frustrating isn't it.
It has got to be server side - you can see the 'results in progress' total dropping as machines start to run dry/low cache.

My Host-Id is 7521.
I made a reset of the projekt an have now DCF 1.00000
I have BOINC 6.4.5
Now I get the messages:

18.12.2008 17:47:39|GPUGRID|Sending scheduler request: Requested by user. Requesting 168179 seconds of work, reporting 0 completed tasks
18.12.2008 17:47:44|GPUGRID|Scheduler request completed: got 0 new tasks
18.12.2008 17:47:44|GPUGRID|Message from server: No work sent
18.12.2008 17:47:44|GPUGRID|Message from server: Full-atom molecular dynamics on Cell processor is not available for your type of computer.
18.12.2008 17:47:44|GPUGRID|Message from server: Full-atom molecular dynamics for Cell processor is not available for your type of computer.

I have reduced the number of flops requested by a factor 10 on type of workunits.
To see if things improve.

gdf

GDF - can you explain why the amount of work requested &/or a DCF issue could trigger a 'not available for your type of computer' message?

Thanks,
P.

I think that is because if the estimate is longer than the deadline, the WU is not sent.

gdf

I'm now back on 6.4.2. after seeing GDF's post at 16:54.At first no luck again. Same sort of messages. Then I remembered a trick from my days cruncing cpdn...

I suspended all processing. Used notepad to edit client_state.xml and found the dcf tag in the block for GPU. Between the GPU URL and the DCF there is nrpc_failures. That was 6, changed that to 0; saved the file and exited notepad. Resumed boinc processing and updated GPU manually. Bingo! I now have GPU plus 4 other tasks running in high priority (for no good reason) on the GTX260 machine. I also have 2 GPU tasks ready to go when this finishes in 11 hours.

Note. Usual provisio with editing client_state.xml: make sure BOINC can't update it and preserve the < > </> signs or it will take some time on your part to correct it! You shouldn't need to do this edit unless you can't get work and were getting work previously.

Phoneman1

........ great tip. Got 4 WUs - but coming down at 240+hours, even though my DCF is 1.000000.

Got 4 WUs - but coming down at 240+hours, even though my DCF is 1.000000.

Phew! Yes just gone onto BOINCView for the first time in several hours and seen similar on my machines. I guess that is why the GPU tasks are running high priority. I am still puzzled why the other project's tasks are running in high priority. There is at most two days of work there for a 10 day deadline project. When I couldn't get a GPU task to run they were running in normal mode - very odd.

Phoneman1

Yippee! For me, that tip works host ID 5889

Boooo! It does not work on host ID 8882

---

Just got this from Beta mail list.

Hi Rom and David,

I just downloaded 6.4.5 to my home XP64 machine. Because I use
firefox it took me to the wrong download page and gave me the Win32
version.

The Win32 version has been fixed to report Win64 when run on a Win64
box. Which is fine for projects that have Win64 applications, but
apparently it doesn't report Win32 as an alternate platform, so I
don't get any work on projects that only support Win32. That's a
substantial problem. Haven't been monitoring the lists to see if that
problem has been reported before.

Eric

Could this be the issue here on GPUGrid..?

It does not work on host ID 8882

Nightlord if it is not the wrong version of Boinc as just suggested, I'd suspend Boinc processing take another look at the client_state.xml file on that machine.

Between the:
<project>
<master_url>http://www.gpugrid.net/</master_url>
<project_name>GPUGRID</project_name>

and the

<duration_correction_factor>1.000000</duration_correction_factor>

we have been looking at you should find these two lines amongst the other stuff:

<nrpc_failures>0</nrpc_failures>
<master_fetch_failures>0</master_fetch_failures>

both need to be 0 - Burdett and I only needed to change the top one. It is just possible your second machine has a number greater than 0 for master fetch failures.

Phoneman1

Using version 6.4.2, did project reset and still getting no work units. Host ID is 19685. Also I was not getting the "Full-atom molecular dynamics is not available for your type of computer." message until I did a couple of version changes based on recommendations from this forum. Everything else was pretty much the same though. Btw I am a new user, so this may not be related to the current series of troubles.

18/12/2008 21:22:48||Libraries: libcurl/7.19.0 OpenSSL/0.9.8i zlib/1.2.3
18/12/2008 21:22:48||Data directory: D:\aplw\boinc-data
18/12/2008 21:22:48||Running under account Administrator
18/12/2008 21:22:48||Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU T7500 @ 2.20GHz [x86 Family 6 Model 15 Stepping 10]
18/12/2008 21:22:48||Processor features: fpu tsc sse sse2 mmx
18/12/2008 21:22:48||OS: Microsoft Windows XP: Professional x86 Editon, Service Pack 2, (05.01.2600.00)
18/12/2008 21:22:48||Memory: 2.00 GB physical, 3.85 GB virtual
18/12/2008 21:22:48||Disk: 141.04 GB total, 16.96 GB free
18/12/2008 21:22:48||Local time is UTC +1 hours
18/12/2008 21:22:48||Not using a proxy
18/12/2008 21:22:48||CUDA devices found
18/12/2008 21:22:48||Coprocessor: GeForce 8600M GT (1)
18/12/2008 21:22:48||Version change (6.3.21 -> 6.4.2)
18/12/2008 21:22:48|GPUGRID|URL: http://www.gpugrid.net/; Computer ID: 19685; location: (none); project prefs: default
18/12/2008 21:22:48||No general preferences found - using BOINC defaults
18/12/2008 21:22:48||Reading preferences override file
18/12/2008 21:22:48||Preferences limit memory usage when active to 1023.18MB
18/12/2008 21:22:48||Preferences limit memory usage when idle to 1841.72MB
18/12/2008 21:22:48||Preferences limit disk usage to 5.00GB
18/12/2008 21:22:48||Running CPU benchmarks
18/12/2008 21:22:48||Suspending computation - running CPU benchmarks
18/12/2008 21:23:19||Benchmark results:
18/12/2008 21:23:19|| Number of CPUs: 2
18/12/2008 21:23:19|| 2152 floating point MIPS (Whetstone) per CPU
18/12/2008 21:23:19|| 5050 integer MIPS (Dhrystone) per CPU
18/12/2008 21:23:53|GPUGRID|Resetting project
18/12/2008 21:23:58|GPUGRID|Sending scheduler request: Requested by user. Requesting 60480 seconds of work, reporting 0 completed tasks
18/12/2008 21:24:03|GPUGRID|Scheduler request completed: got 0 new tasks
18/12/2008 21:24:03|GPUGRID|Message from server: No work sent
18/12/2008 21:24:03|GPUGRID|Message from server: Full-atom molecular dynamics for Cell processor is not available for your type of computer.
18/12/2008 21:24:03|GPUGRID|Message from server: Full-atom molecular dynamics on Cell processor is not available for your type of computer.
18/12/2008 21:24:03|GPUGRID|Message from server: Full-atom molecular dynamics is not available for your type of computer.