I know there are problems at present with getting work and eventually this will pass. However, my question is about the number of units I get for CPU processing as opposed to units for CPU processing.

I installed my new GPU only recently and it appears to have been runnnig fine. I get dozens and dozens of units for the GPU to process and have never run out. I currently operate a 10 day cache in the hope of increasing the number of CPU tasks received. I now have GPU task overload!

The CPU always gets far fewer units to process. I can process two tasks at a time and it seems to get through a great deal of work. However, I never get a full supply of CPU tasks.

Is my set-up wrong somewhere?
Should I be getting through vastly more GPU units than I am?
Is there any way to increase the number of CPU tasks but not GPU tasks?
Any help or suggestions will be most gratefully received.

(Tried to search for answers for this but the threads are now so voluminous that I could find nothing of help elsewhere. Sorry)

---

Hi, I see, you're running the SETI stock app. I wonder if you can use the REScheduler, to divide task's between CPU and GPU.

It's always a surprise, what kind of work you'll get, not very much you can do about that.
Sometimes I've got too much CPU work, so the GPU's go idle, or crunch for another project, Einstein f.i.
Or it's the other way around, only (CUDA)GPU and no CPU.
I run all my rigs, with an optimized app. and use RESChedule (1.9).

---

Many thanks Fred.
I may be a very long-term Seti user but I remain a computer illiterate and ignoramus!
My next naive question is: where do I find and how di I use the REScheduler?

---

Hi, I've looked around and there appeared to be a few forum-threads, about the REScheduler and it's use.
Reschedule 1
And also,
this thread.
It looks like, you can only use RESchedule with the optimized app.'s , f.i. use the LUNATIC's INSTALLER.

After some searching, found a link to Unified Installer

Well, i can't play the piano, either ;-) .

---

That is perfect but I shall have to wait until I have a tame "advanced user" to help me install and run. Looks as if it is not a job for one as faint-hearted as me!

Part of the trouble seems to be that BOINC Manager seems to expect me to take 40 or more hours to complete a CPU task and only up to about 2 hours to complete a GPU task. Therefore I am inundated with GPU work and very quickly run short of CPU work. The latter time is about correct but the former is so exaggeratedly over the top as to be ridiculous.

Is this a problem that needs to be sorted out in a future BOINC Manager release? Must be possible to work out these times better and make appropriate allowances.

If all else fails, I shall leave my cache at 10 days and hope that I can collect enough CPU work for it not to run out at crisis times.

Many thanks again. Anyone else any ideas that an old simpleton can manage?

---

That is perfect but I shall have to wait until I have a tame "advanced user" to help me install and run. Looks as if it is not a job for one as faint-hearted as me!

Part of the trouble seems to be that BOINC Manager seems to expect me to take 40 or more hours to complete a CPU task and only up to about 2 hours to complete a GPU task. Therefore I am inundated with GPU work and very quickly run short of CPU work. The latter time is about correct but the former is so exaggeratedly over the top as to be ridiculous.

I can be quite tame ;-) .
That is another good reason to use optimized app's, if experiencing rather long run-times on Multi Beam WU's .

I've to go out, but this evening I can 'walk you through the Install of the LUNATIC's Installer' .
Important is which SSE version is supported by your E8500, (it is?)
(Probably SSSE3x, for MB and SSE3 for AP WU's)
You can use CPUz for this, to find out.
And DECompress the *.tar/rar RESchedule 1.9 and install it.

---

You are dealing with an idiot!
The processors are E8400. I imagine this will make a difference. I have no idea about SSE3 or SSE3x (doh!)

---

If you don't feel comfortable "tweaking" your computer just leave it as it is.

I see your GPU tasks (SETI@home Enhanced v6.09 (cuda23)) take about 0.5 hour
and your CPU tasks (SETI@home Enhanced v6.03) take about 6-7 hours

You have 2 CPU cores and 1 GPU so you have to see about
6 times more CUDA tasks than CPU tasks.

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

It's a pitty BOINC X.xx.xx doesn't have a build-in devider/rescheduler, so you can 'split/devide' your WU's for CPU{cores}('s) & GPU('s).

Easy said, not quite easily done.

And you can forget about, set & forget ;-)
And the forum will be too small, for all the questions about how to manage it. . . .

---

OK, I am more than grateful for all suggestions. At this stage I think it best to let sleeping processors lie, as it were, and wait until July when my son will arrive with his experise and we'll have a fresh look at it then.

I'll leave this thread open for a bit longer because others may contribute or be interested in the same problem. I'll lock it up some time in the next day or so.

Carry on shrubbing!

---