First of all: a Happy New Year to everybody :-)

here my problem:
earlier today, on two of my PCs, in the Boinc Manager I have added other projects, in addition to WorldCommunityGrid (WCG) which has run on the machines solely so far.
However, only tasks from WCG continue to be chosen and worked on, no tasks from the other projects.
What is wrong? How can I get the BoincManager to take tasks from the newly added projects?

Without knowing what the other projects are it's a bit difficult to guess. Do those projects have work? Also for your operating system and hardware?

---

Patience... It will take time for the new projects to build debt so that they get their turn.

edit: If you don't have any work from the new projects yet, try suspending WCG for a few minutes and see if work downloads from the new projects.

Without knowing what the other projects are it's a bit difficult to guess. Do those projects have work? Also for your operating system and hardware?

sorry for the lack of information:
On my Vista 32-bit, it's DistrRTGen
On my Win7-64bit, it's SETI@home

I choose these to projects explicitly in order to find out whether computation works with my graphic cards (NVIDIA GeForce 8600GTS - 256MB; ATI Radeon 5400 - 512MB).

edit: If you don't have any work from the new projects yet, try suspending WCG for a few minutes and see if work downloads from the new projects.

this worked :-)

edit: If you don't have any work from the new projects yet, try suspending WCG for a few minutes and see if work downloads from the new projects.

this worked :-)

From Seti@Home project I got the task AstroPulse v7, but no way for me to tell whether this uses GPU for computing or not

Hello Eric,

I run Astropulse 7 on seti.

When I get a WU that can run on GPU, the BOINC (advanced view) task page
shows, on the status column, "Running(1 CPUs + 1 AMD/ATI GPU)

You should get something similar.

If not, check the event log at start-up time and verify that BOINC detects your GPU card.

I have an AMD/ATI card and wrote up an app_config.xml file that
limits me to only 1 task running in the GPU at a time,
and having a WU running on the CPU to assist the GPU.

There are a lot of forum events under SETI that give examples of app_config.xml

Hope this helps.

Have fun,
Jay

-- edit --

SETI does have WU for GPU all of the time.
They may be temporarily out. Today, Sunday 18 Jan, there are some - at least for AMD/ATI.

EGAD!
I left out the word "not"

Seti does not have WU for all types of GPU all of the time.

(Threw in an edit too.)

They also don't give out WU during their maintenace hours -
Usually starting at noon Pacific Time on Tuesdays.

Have fun,
Jay

I know I saw this somewhere...

I have an AMD/ATI card and wrote up an app_config.xml file that limits me to only 1 task running in the GPU at a time, and having a WU running on the CPU to assist the GPU.

I have to ask, could you clarify why you do this? It's the default way that BOINC runs things, one task on any GPU and one task per CPU core. No need to write an app_info.xml file for that.

---

I know I saw this somewhere...

I have an AMD/ATI card and wrote up an app_config.xml file that limits me to only 1 task running in the GPU at a time, and having a WU running on the CPU to assist the GPU.

I have to ask, could you clarify why you do this? It's the default way that BOINC runs things, one task on any GPU and one task per CPU core. No need to write an app_info.xml file for that.

Because we Lunatics recommend that users free a CPU core when running OpenCL GPU apps, other wise there is an app slowdown:

Best usage tips:

For best performance it is important to free 2 CPU cores running multiple instances.
Freeing at least 1 CPU core is necessity to get enough GPU usage.*

* As alternate solution try to use -cpu_lock / -cpu_lock_fixed_cpu N options.
This might only work on fast multicore CPU`s.

There are various ways of doing this, setting 'On multiprocessors, use at most' low enough so one less CPU task is run,
to changing to the following in the app_info.xml to achieve the same thing:

<avg_ncpus>1.0</avg_ncpus>
<max_ncpus>1.0</max_ncpus>

to changing to the following in an app_config.xml to:

<cpu_usage>1.0</cpu_usage>

They'll all achieve the same thing, althrough only the second two will show the desired:

Running(1 CPUs + 1 AMD/ATI GPU)

Claggy

Because we Lunatics recommend that users free a CPU core when running OpenCL GPU apps

That's fine, but as I read jay_e's post he's not saying that. As I read it, he's running a WU on the CPU that assists the GPU: and having a WU running on the CPU to assist the GPU

Which is why I am asking for clarification. Because if he's running one task on the GPU and one task on the CPU that assists the GPU, then you don't need an application configuration file for that. That's default behaviour. (He doesn't say how many CPU cores he has)

---

Because we Lunatics recommend that users free a CPU core when running OpenCL GPU apps

That's fine, but as I read jay_e's post he's not saying that. As I read it, he's running a WU on the CPU that assists the GPU: and having a WU running on the CPU to assist the GPU

Which is why I am asking for clarification. Because if he's running one task on the GPU and one task on the CPU that assists the GPU, then you don't need an application configuration file for that. That's default behaviour. (He doesn't say how many CPU cores he has)

He's talking about SETI/Astropulse.

There's no way he can be "having a WU running on the CPU to assist the GPU", to quote him exactly. Even if he had an application running on the CPU, it couldn't "help" the GPU - I don't know of anyone who has (yet) written that sort of bi-modal application to make separate use of CPU and GPU so that one resource 'helps' the other resource.

The nearest we know of is the common advice for OpenCL apps (like SETI/AP) to limit the number of apps running on the CPU by one, or even to one on a dual-core CPU, to leave some resources spare to support the GPU. He might be meaning that.