I am currently running these option for my Seti@home:

Run only the selected applications SETI@home Enhanced: yes
Astropulse: yes
Astropulse v5: yes
If no work for selected applications is available, accept work from other applications? yes
Use Graphics Processing Unit (GPU) if available yes

What I want to know is, if I just leave it alone, will it request its own CUDA work or do I have to manually edit these preferences each time I need it to get work? And if so, is there any way I can force it to download CUDA scans?

Thanks

---

I am currently running these option for my Seti@home:
Run only the selected applications SETI@home Enhanced: yes
Astropulse: yes
Astropulse v5: yes
If no work for selected applications is available, accept work from other applications? yes
Use Graphics Processing Unit (GPU) if available yes

What I want to know is, if I just leave it alone, will it request its own CUDA work or do I have to manually edit these preferences each time I need it to get work? And if so, is there any way I can force it to download CUDA scans?

In theory, the settings you have are correct for a mix of Astropulse 5.00, 5.03 and multibeam (CUDA, assuming you have a CUDA capable card and drivers). In practice, the scheduler that fetches work units does not discriminate between CPU and GPU work for clients running BOINC manager 6.4.x . The net result is usually nothing but Astropulse 5.03 downloads. To force CUDA multibeam downloads, Astropulse 5.00 and 5.03 must be deselected, and the option to fetch work for other applications must be disabled. This means nothing but multibeam work units will be downloaded - no CPU work if you are CUDA capable. Consequently, if you want a mix to keep your CPU and GPU busy, you have to alternate between everything selected and only SETI@Home enhanced (multibeam) selected every other day or so.

Alternatively, you could use the development version 6.6.15 of BOINC manager which has a scheduler that discriminates between CPU and GPU work. Bear in mind this is still officially a "test" version. In my hands, it works much better for CUDA capable clients.

This is a long explanation but I hope it's clear.

---

--
Classic 82353 WU / 400979 h

Thanks for your quick reply. I have upgraded to 6.6.15 upon your recommendation and it is successfully running.
I currently have a cc_config.xml file that contains:

<ncpus>5</ncpus>

Because I am running an AMD phenom x4 and my graphics card. Is this needed and are there any other things i need to add?

Thanks

---

No, you don't need that entry in cc_config.xml anymore when running BOINC 6.6, the client will do this all by itself.
Also, do know that there are still some bugs in the present 6.6 clients, so don't keep this one running forever. Preferably, when a new version of 6.6 is released and recommended, upgrade to that one.

---

Ok! Thanks for you help/advice!

---

I have removed the cc_config.xml file and restarted BOINC, however, it is not running CUDA. It is running 4 AP scans and the CUDA says "Waiting to run". Should I just leave it and see if it starts automatically?

---

Yes, it should just continue eventually on its own, once one of the CPUs is not completely working on an AP. This is one of the bugs that is still in 6.6.15

Comparing this BOINC with the one you were running before is a bit difficult, as the 6.6 range have a new way of calculating debt, a whole new (and separate) CPU and GPU scheduler and some other fine changes as well. So do not expect it to work as the old one was. As long as you can do that, it should work reasonably OK and that CUDA task will eventually continue. ;-)

---

Ok thanks for your help!!!
It all makes sense now

One final thing, when I am running CUDA will it suspend the GPU work whilst the computer is in use if i tick that box in the preferences tab?

---

One final thing, when I am running CUDA will it suspend the GPU work whilst the computer is in use if i tick that box in the preferences tab?

Yes, it will do that. That option has been supported since 6.6.7
It will even unload the task from the videocard's memory, as long as the task has checkpointed at least once. Something that 6.4 didn't do.

---

Ok thanks!

---

Could you post a link to the higher test versions of boinc so I can check them out please. This is exactly what I'm looking for. Something to keep every ounce of processors working. Thank you.

---

Could you post a link to the higher test versions of boinc so I can check them out please. This is exactly what I'm looking for. Something to keep every ounce of processors working. Thank you.

It's the All versions link on the BOINC download page.

Thank you, don't know how I missed that, lol. Thank you, again.

---