A very strange thing is happening on my host.

I´m using BOINC 6.3.10 with default settings (no messing around with ncpu+1 or something like that)

After I uploaded a new WU for GPU Grid last night I found my PC crunching happily two CPU Seti units alongside one GPUgrid WU.... (Core2 Duo)

The WU will be finished in about 1h so I can give you the time per step / estimated completion time.



Cu KyleFL

Hi

Same thing here, I am not using ncpu+1

Boinc 6.3.10 - Vista64

Same here but only on 2 from 4 computers. Is there any reason for that ?

I am using Boinc 6.3.10

Hello,

I had the same behavior last evening. Suddenly boinc started an extra GPUGrid-WU additional to the running CPDN and EAH. It is an old WU which is crunched before as an extra task on my Dualcore. For information: before that a new WU was downloaded. Yesterday evening I stopped Boinc overnight and in the morning I started BOINC again.
Now this behavior I see. Boinc starts GPU-Grid as a third tasks and after a short while (I think on the first checkpoint) the GPU-Grid-WU is stopped. Of course the GPU-Grid has the highest STD of all projects and on "right" scheduling this WU has to run with first priority. After on hour BOINC started GPU-Grid again and stops immediately.

I show you the stdoudae:

Same here, had running 6 tasks on both Quad running PS3Grid. Had reboot both machines, same effect. Have killed the ncpus-entry in the cc_config.xml and now running 5 tasks on both machines.

Something changed in the code? Looks like PS3Grid is now running like DepSpid in the past (http://depspid.net/).

Edit: First only my two 64 bit Vista PCs affected, after upload the calculaed WU on the PC of Cebion now also our 32 bit XP Quad affected.
____________

Same behaviour on my computer.
And i saw that the CPU time on different computer have decrease.

Before this update, on my Server 2003-x64, the CPU time was on average 2400 seconds, and for one task this morning, i can see only 800 seconds! But an increase of the total running time from 49000 seconds to 53000 seconds.

I don't understand what happened, because there has been a server upgrade last night, as GDF said.

Jim PROFIT

Ok -- my GPU WU is now finished:
12h instead of 9h.

After it finished the GPU-WU, it startet the next one -- running it alongside 2CPU WUs.
But it seems, that the GPU-WU is preempted for ~1minute after calculating one step. It trepeats infinity. Calculating -- preempted -- calculating -- and so on...

I´ve stopped the CPU-WU (Seti) tasks for now until we have sorted out the problem. After that the GPU-WU seems to be back on full crunching-speed.





Cu KyleFL

In That case just set <ncpus> to cores-1 (so f.i. 3 for a quad core) that ensures that one core always is unused for the gpu tasks.

In That case just set <ncpus> to cores-1 (so f.i. 3 for a quad core) that ensures that one core always is unused for the gpu tasks.

Surely this will help, but the balancing with the resourceshare is blown away by this method.

I think there must be a solution from the "chefs". ;-)

Setting Boinc to use only "50% of Prozessors" (DualCore) seems to work.

Boinc still seems to be using both cores (one for GPU - one for CPU) and not putting both tasks on one core.

That should do the trick for now.
Update: I roughly calculated the crunching time and the GPU-WU seems to be on full force like it was before.


But the question is: What had happend to change the behavior of Boinc?


Cu KyleFL

Could you repeat what I did (previous post) to make my GPU run again? Maybe it helps.

MrS

just encountered it myself! No changes suddenly 4 wus + grid running.

Changed to ncpus=3 worked

I had too this morning 2 cpu + 1 GPU WUs running in Core 2 Duo and 4 cpu + 1 GPU in Quad core. But I have experienced a 20% download in my points per day, so I have put 50% and 75 % CPU use restriction respectively and now, as before, all is OK.

Back from work now, and after my early morning post in the other thread, I checked my boxes again.

They were running the same as this morning. So, I've just updated them all with a cc_config.xml file, setting ncpus-1 (ncpus=1 for dual core and ncpus=3 for quad) and have the same experience as other users: everything returns to normal.

Strange behaviour indeed??!
____________

Back from work now, and after my early morning post in the other thread, I checked my boxes again.

They were running the same as this morning. So, I've just updated them all with a cc_config.xml file, setting ncpus-1 (ncpus=1 for dual core and ncpus=3 for quad) and have the same experience as other users: everything returns to normal.

Strange behaviour indeed??!

BOINC seems to be completely out of shape. After finish the last WU I had again running 5 tasks on my Quad, but all are non-GPU-tasks. I had to stop BOINC and restart it, then are 4 CPU-tasks and one GPU-task are running. Another team-mate has on XP 32 bit the problem, that without a cc_config.xml 6 tasks are running. Nothing helps.
____________

You don´t need to make this change through the cc_config.xml file. Just go into the settings and on the CPU-Tab:

Thank you GDF

Problem found and fixed. Now the behavior is as before: 1GPU+1CPU

When there will be no performance drop in using the 0 CPU, we will do the change in the policy.

gdf

This should be configurable by the user. Someone like to use all cores for other projects and someone not.

A GTX280 needs for one WU

with ncpu-entry 8h 24m, thats 384,77 Credits/h.
without ncpu-entry 10h 28m, thats 308,99 Credits/h.

The difference is here 75,78 cr/h. The additional usable core with ncpus=5 can't reach such a amount of credits in one hour, also not on PrimeGrid. There I get 53,23 cr/h while crunching PSP Sieve.

So for a GTX280 a ncpu-entry cores+1 is not advisable.

That takes not the same effect on a 8800GT or smaller graphic card.

You have to calculate for your own.
____________

....... Still have the issue - Quad still wants to run 5 processes, and the PS3Grid process runs veeeerrrrryyyyyy slowly - how can this be fixed?

....... Still have the issue - Quad still wants to run 5 processes, and the PS3Grid process runs veeeerrrrryyyyyy slowly - how can this be fixed?

It may be look very slowly, but it's only showing the CPU using time and not the real time. The system should be run in normal state after the next connect to the server (if you don't have created a cc_config.xml with a ncpus-entry in the BOINC-data directory).

The new server software has a flag for GPU using software to add a non existent core for crunching with all CPU-cores and a additional virtual core for the GPU. This flag is switsched by GDF, so your client should work as normal after the next connect.
[/quote]


Thanks Kokomiko - I can see from today's WU that the step time has moved from 42ms to 70ms - so something has slowed my GTX260. Unfortunately, the project is mostly offline at the moment... so I'll have to wait for the cure!

Burdett:

You can do a workaround in your BOINC settings:

Just tell your BOINC, that (on Multiprozessor Systems use only xx Cores)

On a Quad: 75%
On a Duo: 50%

That should get you back to normal as long as the "wrong" server flag is active.
After getting the next new work, stop the cpu taks, take the setting back to 100% and restart the CPU Task.

That did the trick for me.


Cu KyleFL

Burdett:

You can do a workaround in your BOINC settings:

Just tell your BOINC, that (on Multiprozessor Systems use only xx Cores)

On a Quad: 75%
On a Duo: 50%

That should get you back to normal as long as the "wrong" server flag is active.
After getting the next new work, stop the cpu taks, take the setting back to 100% and restart the CPU Task.

That did the trick for me.


Cu KyleFL

I can't manage to make the cc_config.xml trick work anymore... any other with the same issue?

I can't manage to make the cc_config.xml trick work anymore... any other with the same issue?

I have a dual core and I tried Ncpus 1, 2 and 3 (-1, 0, +1) but none of the options worked.

I want to run two "normal" tasks + the GPUGrid one.

The fact is that, whatever I wrote in the cc_config file, it just ran normal tasks, with the GPUGrid one "waiting for elaboration".

Without the cc_config file, GPUGrid works but it reserves a core for itself (1 normal task + the GPUGrid one).

I have a dual core and I tried Ncpus 1, 2 and 3 (-1, 0, +1) but none of the options worked.

I want to run two "normal" tasks + the GPUGrid one.

The fact is that, whatever I wrote in the cc_config file, it just ran normal tasks, with the GPUGrid one "waiting for elaboration".

Without the cc_config file, GPUGrid works but it reserves a core for itself (1 normal task + the GPUGrid one).

He is running Linux, there the slowdown is almost not measurable, at least on my machines there seems to be no impact to the "time per step" value when I set ncpus+1

Plus I added the following entry to roots crontab to increase the priority of the acemd process automatically:

* * * * * renice 5 `ps -ef | grep acemd | grep -v grep | awk '{ print $2 }'`

So for a GTX280 a ncpu-entry cores+1 is not advisable.

That takes not the same effect on a 8800GT or smaller graphic card.

You have to calculate for your own.

So for a GTX280 a ncpu-entry cores+1 is not advisable.

That takes not the same effect on a 8800GT or smaller graphic card.

You have to calculate for your own.

I take it from reading this I can run the ncpu-entry cores+1 if I have a 8800GT because it doesn't need a CPU Core reserved for it ... ???

He is running Linux, there the slowdown is almost not measurable, at least on my machines there seems to be no impact to the "time per step" value when I set ncpus+1

Plus I added the following entry to roots crontab to increase the priority of the acemd process automatically:

* * * * * renice 5 `ps -ef | grep acemd | grep -v grep | awk '{ print $2 }'`

I have a dual core and I tried Ncpus 1, 2 and 3 (-1, 0, +1) but none of the options worked.

I want to run two "normal" tasks + the GPUGrid one.

The fact is that, whatever I wrote in the cc_config file, it just ran normal tasks, with the GPUGrid one "waiting for elaboration".

Without the cc_config file, GPUGrid works but it reserves a core for itself (1 normal task + the GPUGrid one).

You´ll get ~30% longer computing time on the GPU-WU if you run a CPU-WU on the Core that the GPU-WU is using.
Is it possible that you did put the cc_config file in the wrong directory?
Or that you didn´t restart your Boinc after making your changes?


Cu KyleFL

[quote
On Windows I would wait a few additional weeks, until the problem with the scheduler is sorted out and the NCPU+1 can be run without a penatly.
Maybe it´s possible to let Boinc run the GPU-WU with a higher prozess priority than the CPU WUs. That way the GPU-WU always would get CPU time and the CPU-WU would only take the idle-cycles left...


Cu KyleFL

I've suddenly got work hanging. Since PS3GRID switched to allowing me to run 2 CPU + 1CPU/GPU on 6.3.10 it has not been running all that well. not using ncpus+1, just running normal. I did upgrade to 178.13 about same time.

Yesterday I found the GPU task had been hung, the GPU was cold and had been for 15 hours. I think I ended up resetting computer. It ran from about 14% to 50% (in high priority). Now it is stuck again, this morning. was running when I went to sleep last night.

From what I can see in the log, the 2 CPUs are still running tasks for another project, have finished some (uploaded) and running different tasks.

If I suspend the hung task, another will not start.

I suspend all work. It looked like it resumed (logged), the % went from 52.388 to 53.400 but the GPU temp is still cold.

I suspended, quit boinc and restarted. This started another GPU task which after a few seconds had a computational error. That too was now going to be late and I was going to abort anyway.

I'm just going to abort this hung one because it is hanging, due in 4 hours with 8 hours runtime left and has wasted over 29 hours of time.

I hope I get some new tasks and this does not repeat, I will check back when I get home from work.

Except now I only have 1 CPU + 1 CPU/GPU running, 2 total, not 3.

Addition to last post.

The work unit is http://www.ps3grid.net/workunit.php?wuid=51416

I see someone else already had a go at it and it failed for them too.