I was wondering whether anyone can tell me why my machine has only been getting CUDA work for over a week. I have settings to accept both CPU and GPU work but for some reason I have only been getting 99.994% CUDA tasks. I see that I did get one CPU task today but I used to get fairly balanced work between CPU and GPU. Is there something about the project that decides the type of work a certain machine should get? Or is it simply the case where the project has only been making GPU work?

Keith

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I was wondering whether anyone can tell me why my machine has only been getting CUDA work for over a week. I have settings to accept both CPU and GPU work but for some reason I have only been getting 99.994% CUDA tasks. I see that I did get one CPU task today but I used to get fairly balanced work between CPU and GPU. Is there something about the project that decides the type of work a certain machine should get? Or is it simply the case where the project has only been making GPU work?

Keith

I am getting (and returning) mostly CUDA tasks, but did get one Astropulse also. I had also joined Climateprediction over the weekend since the machine idle. So between the Astro and the Climate the main CPU has plenty of work, and the GPU is getting work, so I am happy.

I guess it is just sort of random which machines get tasks in this situation, coming out of the big outage. But Keith, does your machine accept Astro WU? If you catch one that will keep one CPU busy for several days!

Also, the Climate WUs give partial credit as you go along, so no need to wait for several days to get credit.

---

are you receiving Einstein@home work? or are you set to only receive Seti?

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

Hi Hobo, yes I also do AP work too but that crunches through in only about half a day. Have a couple still pending and I validated one about a week ago. Not worried about credit. In fact in a short two months I've already surpassed all my 10 years of accumulative SETI work by the work credit I've received from MilkyWay.

Skildude - I am getting both CPU and GPU from Einstein project.

By the by .....is anyone else doing MilkyWay@Home work? If so, can you take a look over in the Number Crunching Forum at my thread topic:

"How does a MW 0,50 CPU task affect your computer...?"

I haven't received any response from anyone in the forum about my particular question.

Cheers,

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Well I've been getting plenty of MB CPU work here all morning on my 3 PC's and as far as I can see they've all been VLAR's so far.

Cheers.

---

I'd bet that your BOINC is running into some sort of debt with Einstein which is probably why you aren't getting the seti work

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I was wondering whether anyone can tell me why my machine has only been getting CUDA work for over a week. I have settings to accept both CPU and GPU work but for some reason I have only been getting 99.994% CUDA tasks. I see that I did get one CPU task today but I used to get fairly balanced work between CPU and GPU. Is there something about the project that decides the type of work a certain machine should get? Or is it simply the case where the project has only been making GPU work?

Keith

The main determinant is what it asks for. Which you can see in the message/event log on your own computer, and nowhere else. You see it even more clearly if you set the <work_fetch_debug> logging flag.

So tell us what you're asking for, and then we can try to work out why you are, or aren't, getting it.

I keep reading through all my message log and I never see a request for CPU work. I'll add the <work_fetch_debug> flag to my client file and get back to you.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I see I have picked up a CPU VLAR task since I posted. This is my debug output:

3/23/2011 3:56:53 PM | | [work_fetch]: work fetch start
3/23/2011 3:56:53 PM | | [work_fetch] ------- start work fetch state -------
3/23/2011 3:56:53 PM | | [work_fetch] target work buffer: 8640.00 + 345600.00 sec
3/23/2011 3:56:53 PM | | [work_fetch] CPU: shortfall 0.00 nidle 0.00 saturated 850238.15 busy 0.00 RS fetchable 100.00 runnable 500.00
3/23/2011 3:56:53 PM | Einstein@Home | [work_fetch] CPU: fetch share 0.00 LTD -43347.57 backoff dt 0.00 int 0.00 (comm deferred)
3/23/2011 3:56:53 PM | Milkyway@home | [work_fetch] CPU: fetch share 1.00 LTD -560.96 backoff dt 0.00 int 4800.00
3/23/2011 3:56:53 PM | SETI@home | [work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred)
3/23/2011 3:56:53 PM | | [work_fetch] NVIDIA GPU: shortfall 137833.85 nidle 0.00 saturated 256522.09 busy 0.00 RS fetchable 0.00 runnable 500.00
3/23/2011 3:56:53 PM | Einstein@Home | [work_fetch] NVIDIA GPU: fetch share 0.00 LTD -112617.56 backoff dt 0.00 int 0.00 (comm deferred)
3/23/2011 3:56:53 PM | Milkyway@home | [work_fetch] NVIDIA GPU: fetch share 0.00 LTD -13748.98 backoff dt 322.12 int 600.00
3/23/2011 3:56:53 PM | SETI@home | [work_fetch] NVIDIA GPU: fetch share 0.00 LTD 0.00 backoff dt 563.27 int 600.00 (comm deferred)
3/23/2011 3:56:53 PM | Einstein@Home | [work_fetch] overall LTD -3875772.32
3/23/2011 3:56:53 PM | Milkyway@home | [work_fetch] overall LTD -495808.97
3/23/2011 3:56:53 PM | SETI@home | [work_fetch] overall LTD -4586385.83

Any insights? Are my debts too great to other projects now?

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I see I have picked up a CPU VLAR task since I posted. This is my debug output:

3/23/2011 3:56:53 PM | | [work_fetch]: work fetch start
3/23/2011 3:56:53 PM | | [work_fetch] ------- start work fetch state -------
3/23/2011 3:56:53 PM | | [work_fetch] target work buffer: 8640.00 + 345600.00 sec
3/23/2011 3:56:53 PM | | [work_fetch] CPU: shortfall 0.00 nidle 0.00 saturated 850238.15 busy 0.00 RS fetchable 100.00 runnable 500.00
3/23/2011 3:56:53 PM | Einstein@Home | [work_fetch] CPU: fetch share 0.00 LTD -43347.57 backoff dt 0.00 int 0.00 (comm deferred)
3/23/2011 3:56:53 PM | Milkyway@home | [work_fetch] CPU: fetch share 1.00 LTD -560.96 backoff dt 0.00 int 4800.00
3/23/2011 3:56:53 PM | SETI@home | [work_fetch] CPU: fetch share 0.00 LTD 0.00 backoff dt 0.00 int 0.00 (comm deferred)
3/23/2011 3:56:53 PM | | [work_fetch] NVIDIA GPU: shortfall 137833.85 nidle 0.00 saturated 256522.09 busy 0.00 RS fetchable 0.00 runnable 500.00
3/23/2011 3:56:53 PM | Einstein@Home | [work_fetch] NVIDIA GPU: fetch share 0.00 LTD -112617.56 backoff dt 0.00 int 0.00 (comm deferred)
3/23/2011 3:56:53 PM | Milkyway@home | [work_fetch] NVIDIA GPU: fetch share 0.00 LTD -13748.98 backoff dt 322.12 int 600.00
3/23/2011 3:56:53 PM | SETI@home | [work_fetch] NVIDIA GPU: fetch share 0.00 LTD 0.00 backoff dt 563.27 int 600.00 (comm deferred)
3/23/2011 3:56:53 PM | Einstein@Home | [work_fetch] overall LTD -3875772.32
3/23/2011 3:56:53 PM | Milkyway@home | [work_fetch] overall LTD -495808.97
3/23/2011 3:56:53 PM | SETI@home | [work_fetch] overall LTD -4586385.83

Any insights? Are my debts too great to other projects now?

Red: you're full already. Your CPU doesn't need any more work.
Orange: When you need new CPU work, it'll show up here. Like it does for GPU now.
Green: You've contacted SETI within the last five minutes, probably requesting CUDA work. SETI will throw away your request if you pester it too soon.

So, because the Seti project ran out of work for too long, I ended up with too much CPU work from MW and Einstein, correct? Refresh my knowledge of how the Manager handles project work .... FIFO algorithm is no longer used for clients > 6.10.58, correct?

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

So, because the Seti project ran out of work for too long, I ended up with too much CPU work from MW and Einstein, correct? Refresh my knowledge of how the Manager handles project work .... FIFO algorithm is no longer used for clients > 6.10.58, correct?

FIFO only ever was a (recent) problem for GPU work. Yes, for GPUs (only), v6.10.58 operates on strict FIFO, and won't preempt a task once started. We gradually got those restrictions whittled down during testing of the v6.12 client, and that will now pause GPU work, switch projects, respect Task Switch Interval and so on - just as v6.10.58 and its predecessors do for CPU work, and have done for years.

Thanks for the confirmation of my understanding. Running the 6.12.18 client for now. I don't have to worry about pestering Einstein much because it defaults to a 4 hour backoff anytime you run a app_info file. Wish they would change that behavior. The project that seems to monopolize my computer is the MilkyWay project. For some reason they set really short report times. I'm always getting the other projects preempted for High Priority tasks from MW. Right now I have a 100/100/300 resource share allocation for Einstein/MilkyWay/Seti and I'm thinking of biasing it even more heavily toward Seti. Not too happy that Seti seems to be getting the short end of the stick with regards to credit lately. My primary allegiance is still to Seti as I've been with the project now for almost 10 years. I just added the other projects when I built this new computer and updated operating systems so I can for the first time run other projects for backup when Seti goes offline for long periods of time. Never had that luxury before this new computer.

Cheers, Keith

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I was wondering whether anyone can tell me why my machine has only been getting CUDA work for over a week.

In my case it was just since this last outage. I ran of of CPU work, but still had plenty of CUDA work.
It would ask for both CPU & GPU work a couple of times, then it'd be GPU only.
Looking at my project properties in the Manager, it would show "CPU work fetch deferred for" & "CPU work fetch deferral interval" as anything from 1 min/ 5min to 2 hours/ several hours (and combinations there of).

It appears that it got work several times when asking for GPU work, but not CPU work- that's when the CPU work fetch deferred came into play.
I reset the values several times, but after a couple of failed attempts to get CPU work allocated they'd soon be back.

While i was at work, the system finally snagged some CPU work & now those values are null (---) & it's asking for both CPU & GPU work.

---

Grant
Darwin NT

While I have gotten a very few CPU tasks in the last day or so, it is nothing like I used to get. I have to play games with MilkyWay@Home, if I don't set it to NNT when I do an Update, then MW always puts itself at the front of the queue in demanding CPU work. If I let it get work ahead of Seti, then Seti immediately runs into the CPU dept inflicted my MW. I'd say besides the work dept that builds up on other projects and puts Seti at the bottom, the primary reason I haven't gotten any CPU work is that there hasn't been any to send. I still get tons of GPU work.

Keith

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

While I have gotten a very few CPU tasks in the last day or so, it is nothing like I used to get. I have to play games with MilkyWay@Home, if I don't set it to NNT when I do an Update, then MW always puts itself at the front of the queue in demanding CPU work. If I let it get work ahead of Seti, then Seti immediately runs into the CPU dept inflicted my MW. I'd say besides the work dept that builds up on other projects and puts Seti at the bottom, the primary reason I haven't gotten any CPU work is that there hasn't been any to send. I still get tons of GPU work.

Keith

You could always reschedule GPU work to the CPU.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

So how would I do that? I haven't found any docs or features in the tabs in the Manager that says reschedule. Where is that feature hiding?

Keith

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

...the primary reason I haven't gotten any CPU work is that there hasn't been any to send. I still get tons of GPU work.

On the server side, there's no difference between GPU and CPU tasks - it only depends on what is asked for and they get marked as such.

Gruß,
Gundolf
[edit]The rescheduler is a separate (external) tool. There are several threads about it. (Here's one of them:-)[/edit]

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours

So how would I do that? I haven't found any docs or features in the tabs in the Manager that says reschedule. Where is that feature hiding?

Keith

Like the Opt Apps it is 3rd party. The thread about Freds rescheduler is here.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

yeah, i get mostly cuda now too, that and seti_enhanced, which i assume to be the cpu portion. i stopped working on other projects to support seti more fully. i use my computer alot, but i give seti typically at LEAST 8 hours of full run time, usually overnight, or during the day, when i'm working on new computers.

---