New version 6.3.13 is released:

http://boinc.berkeley.edu/dl/boinc_6.3.13_windows_x86_64.exe
http://boinc.berkeley.edu/dl/boinc_6.3.13_windows_intelx86.exe

Anyone know what specific changes were made to this release?

Total crap like 6.3.12

Ps3grind runs one WU all other projects are stopped, NO wus are calculated. Even if i start other projects by hand and stop Grid - when other projects WUs are finished no new one is started!

so i recommend to everyone to stay with 6.3.10.

That's too bad. I guess there's no need for anyone else to try it?

Actually we could use an update of the scheduler to make sure the GPUs are always busy. 6.3.10 sometimes runs only CPU tasks.

MrS

---

Scanning for our furry friends since Jan 2002

Total crap like 6.3.12

Comments like that do not help.

I'm a little behind and just installed 6.3.12 this morning. I observed the behavior of it not starting CPU and GPU tasks, This is the same as it was back in 6.3.8 or so (not sure exact version).

It was easier to spot this time, as one host started two CPU tasks (since it was out of GPU work). The other host started only 1 GPU task. Having seen this before I knew exactly what the problem was.

I have reported this behavior to the developer.

I don't understand, did you find a problem in 6.3.13? What are the problems exactly?

gdf

Finally I've installed the 6.3.13 and found this:

After running benchmark I got the message:

03.10.2008 22:21:14|PS3GRID|Restarting task GHy8859-GPUTEST3-9-10-acemd_0 using acemd version 645

But my other 4 or 3 free cores does nothing. The running task for PS3Grid shows:

Aktiv (0.47 CPUs, 1 CUDA)

So I've made a downgrade to the 6.3.10, the last version running with no problems.

---

Exactly what the others are reporting.. so I'm not going to test it myself ;)

MrS

---

Scanning for our furry friends since Jan 2002

There is a problem in 6.3.12
It would only start 1 CUDA task and no CPU tasks. To get it to run 3, I had to suspend CUDA project, it would start 2 CPU tasks, then I could resume CUDA project and it would run that too, so 2 CPU + 1 CUDA. But when the CPU tasks finished, it would not start another. After a while I was left just running 1 CUDA task. This is similar to the startup problem that was back in 6.3.8 or so, but the problem after startup is different.

6.3.13 has a different problem.
Upon startup it will only run 1 CUDA task (using 0.73 CPU). It does not start using all CPU's.

If you suspend the CUDA project (PS3GRID) it will start 2 regular CPU tasks (I only have 2 CPU's), now using 100% CPU. Then upon resuming the CUDA project, it remains running the 2 CPU tasks. I let it run. After one of the CPU tasks finished in like 2 hours, it switched to running 1 CPU task and 1 CUDA task. I think it is trying to stay within resource shares but do not know for sure. It however will not run 2 CPU + 1 CUDA.

Host #2 left untouched is still only running 1 CUDA task after about 4 hours, it has 8% to go so I'm waiting to see what happens.

I only have about 4 hours of run time on this version as I just installed it today. I can try these verisons at home this week-end where I have more time.

As I posted the previous, the problem gets worse.

The other CPU task finished and now I'm left only running 1 CUDA task.

After testing the 6.3.13 and downgrading to the 6.3.10 I don't get the GPU task running for longer than some seconds. With a ncpu-entry 5 CPU-tasks are running, without 4 CPU-tasks. No task is running in high priority mode. The GPU-task starts for a short time, then he stops and a CPU task is starting. The first try was a complete deinstallation and a new fresh installation. The same effect. Now I've manually stopped all other tasks exclude 3 CPU-tasks.

What's wrong? When I set any WU to "ready to start", she is starting immediately and the GPU task shows "waiting to run".

The 3 active CPU projects are set in summary to 50% of the CPU-time, the GPU-task is also set to 50% to stop never.

While writing this text the GPU-task is starting as a fifth task, without a ncpus-entry. Could it be, that GDF is testing something at the server and the situation is changing from one minute to the other? Otherwise I don't understand the behavior of the client atm.

---

These are side effects of what was reported in " 6.3.10-runs 2CPU+1GPU on a CoreDuo since yesterday" and which should be fixed on the server side by now. I'm not sure though what's neccessary to make the client notice the change (my BOINC hasn't despite of it contacting the scheduler).

MrS

---

Scanning for our furry friends since Jan 2002

I've still 5 tasks running without a cc_config.xml, also after a update with the project. I'm now waiting for the next upload and hope, that this will change.

---

I've still 5 tasks running without a cc_config.xml, also after a update with the project. I'm now waiting for the next upload and hope, that this will change.

It is necessary to download a WU for the change to take effect. If you update without downloading a new WU, no change will happen.

At least this is what I have found.....

---

Not sure why I am special, but it started off running 3 tasks (1-GPU, 2-CPU) on my Core2 system and has pretty consistently kept three tasks running since updating.

At one point it was down to just the 2 CPU task, and I can't swear to it, but I think it was when was waiting for a WU for the GPU.

Then again, although it clearly had problems, I think I had more luck with 6.3.12 than what was reported by others in that it would intermittently keep 3 tasks running (WU pending or not).

For the record, I am running a 8800 GT w/178.13

Ahh (I should do more searching) ...

6.3.10-runs 2CPU+1GPU on a CoreDuo since yesterday:

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.

After I downloaded the new (fixed) GPU-WU, I stopped the CPU Tasks and then restarted it.
After that, everything went back to normal.


Cu KyleFL

After that, everything went back to normal.

Are you saying that 6.3.13 just works normally for you?

MrS

---

Scanning for our furry friends since Jan 2002

After that, everything went back to normal.

Are you saying that 6.3.13 just works normally for you?

MrS

Hello MrS.
Sorry, that answer was related to the question 2-3Posts above -- not to the 6.3.13 Topic...
I´m still on 6.3.10


Cu KyleFL

I know it's the wrong thread and all.

I don't check my Linux box very often, it runs great, But...

On Friday, my Linux Quad was running 4 cpu and 1 gpu tasks.

Today it's back to 3 cpu and 1 gpu...

on...Boinc 6.3.8
and Ubuntu 8.04

---

Consciousness: That annoying time between naps......

Experience is a wonderful thing: it enables you to recognize a mistake every time you repeat it.

These are caused by an unintentional change in the server config but it's fixed by now.

MrS

---

Scanning for our furry friends since Jan 2002