wow, I am impressed with all the super fast attempts to help me get my gpu started. Thanks.

I'm virtually certain that I do not have the gpu on my 8400gs rig disabled or deselected, and AFAIK, I did the Lunatics install properly. In fact Boinc reports GPU is enabled.

I have tried to reduce my video ram being used, but to no avail. I tried reducing screen resolution...but I don't know what else to do. Googleing is no help. If you have any suggestions in this regard...???

I am not able to check my GT 210 right now, but I will tomorrow.

I don't think screen resolution will really make any significant difference, the numbers are too small. But with Windows 7, you'll need to disable Aero window effects - and maybe let LadyL help you pick out a nice NVidia driver version.

wow, I am impressed with all the super fast attempts to help me get my gpu started. Thanks.

I'm virtually certain that I do not have the gpu on my 8400gs rig disabled or deselected, and AFAIK, I did the Lunatics install properly. In fact Boinc reports GPU is enabled.

I have tried to reduce my video ram being used, but to no avail. I tried reducing screen resolution...but I don't know what else to do. Googleing is no help. If you have any suggestions in this regard...???

I am not able to check my GT 210 right now, but I will tomorrow.

I'm not familiar with Win7 but on a Vista system, video RAM usage is greatly reduced if you use a smaller colour scheme (e.g. 16 bit instead of 32 bit - on Vista that disables aero as well iirc)

I tried all that...reducing resolution, colors, and I turned off Aero...little to no help.

You'll have to repeat that event log experiment when the preject servers are fully back up, and there's work available.

For the time being,'Project has no tasks available' means exactly what it says. But if you ask at a time when work is available, you may get a message back with a reason for the rejection, like for example the lack of free memory - that would show up.

I agree...seems like I'll just have to wait...cricket, cricket.

Thanks everyone. I check back in a few days.

You'll have to repeat that event log experiment when the preject servers are fully back up, and there's work available.

For the time being,'Project has no tasks available' means exactly what it says. But if you ask at a time when work is available, you may get a message back with a reason for the rejection, like for example the lack of free memory - that would show up.

It won't show lack of free memory! With stock he'd get the 'you need at least X MB' message. With x38g it would do CPU fallback on any GPU tasks assigned.

As before - if you can't get at least 230 MiB free VRAM on the 8400 you can't use it to crunch SETI. The x38g application will detect lack of sufficient resources and do CPU fallback - crunching veeeeeeery slowly on the CPU instead. This leaves the CPUs overcommited.

Better disable GPU work on the 8400!

I am using an 8500 GT with only 256mb of ram as well. In Windows 7 with all the whiz-bang graphics enabled BOINC is reporting:
NVIDIA GPU 0: GeForce 8500 GT (driver version 28026, CUDA version 4000, compute capability 1.1, 240MB, 45 GFLOPS peak)
I've not had any issue with it working other then not being able to connect due to the current router issue.

---

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

Maybe an older driver and 16BIT color depth, but have doubts if it'll work and be
worth it?!
(I don't even bother to try and use an 8500GT, too slow, but thats also personal
and still better then nothing).

And almost forgot the availabillity of work...............

---

Well, both gpus started crunching today, although I'm not so sure either one is worth it. Seti seems to hog up one cpu core for what seems like minimal output from the gpu...especially the 8400gs.

Appears the 8400 host's 6.10 task may have done CPU fallback as LadyL mentioned earlier in this thread. That task hasn't been reported by the host yet, probably due to current upload / report issues at the project. It doesn't have any other 6.10 tasks, so I'd advise disabling it pending advice from others t;hat know more than I about that card. (Activity menu).

The host with the 210 completed and reported two 6.10 tasks and they are pending validation, but no errors. Not sure on the other 3, should have crunched by now but may be hung up in the upload/report process.

---

Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.

Appears the 8400 host's 6.10 task may have done CPU fallback as LadyL mentioned earlier in this thread. That task hasn't been reported by the host yet, probably due to current upload / report issues at the project. It doesn't have any other 6.10 tasks, so I'd advise disabling it pending advice from others t;hat know more than I about that card. (Activity menu).

The host with the 210 completed and reported two 6.10 tasks and they are pending validation, but no errors. Not sure on the other 3, should have crunched by now but may be hung up in the upload/report process.

yeah...I think I am learning about cpu fallback. Seems like it occurs if you switch between two gpu projects...

Appears the 8400 host's 6.10 task may have done CPU fallback as LadyL mentioned earlier in this thread. That task hasn't been reported by the host yet, probably due to current upload / report issues at the project. It doesn't have any other 6.10 tasks, so I'd advise disabling it pending advice from others t;hat know more than I about that card. (Activity menu).

The host with the 210 completed and reported two 6.10 tasks and they are pending validation, but no errors. Not sure on the other 3, should have crunched by now but may be hung up in the upload/report process.

yeah...I think I am learning about cpu fallback. Seems like it occurs if you switch between two gpu projects...

I think it is more likely that you are having an issue with the x38g app raising it's priority to normal. Since going to the newest version I often see the GPU maxing out a CPU core instead of only using 2-5%. If I change the task to low then everything works fine for a while. It is problematic enough for me I keep thinking about going back to the previous version. However, With the connection problems and other issues the machine doesn't really get enough work for me to send any to the GPU.

---

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

I think it is more likely that you are having an issue with the x38g app raising it's priority to normal. Since going to the newest version I often see the GPU maxing out a CPU core instead of only using 2-5%. If I change the task to low then everything works fine for a while. It is problematic enough for me I keep thinking about going back to the previous version. However, With the connection problems and other issues the machine doesn't really get enough work for me to send any to the GPU.

The only time I see the x38g app using that kind of CPU time is when it is first starting up and "priming the pump", when it is filling the GPU with initial data. Once it actually starts processing the work unit it drops down to 5% or less. This is normal, and happens with the unoptimized app as well.

---

I think it is more likely that you are having an issue with the x38g app raising it's priority to normal. Since going to the newest version I often see the GPU maxing out a CPU core instead of only using 2-5%. If I change the task to low then everything works fine for a while. It is problematic enough for me I keep thinking about going back to the previous version. However, With the connection problems and other issues the machine doesn't really get enough work for me to send any to the GPU.

The only time I see the x38g app using that kind of CPU time is when it is first starting up and "priming the pump", when it is filling the GPU with initial data. Once it actually starts processing the work unit it drops down to 5% or less. This is normal, and happens with the unoptimized app as well.

Yes the app does use CPU time while loading the work up. Which is perfectly normal. It is just when you look at the task manager and see the GPU app has been using the CPU for the past minutes that it is an issue. When I see that I just lower the app priority and then everything is fine.
There was some discussion about the app priority causing issues back in June http://setiathome.berkeley.edu/forum_thread.php?id=64699

---

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

Appears the 8400 host's 6.10 task may have done CPU fallback as LadyL mentioned earlier in this thread. That task hasn't been reported by the host yet, probably due to current upload / report issues at the project. It doesn't have any other 6.10 tasks, so I'd advise disabling it pending advice from others t;hat know more than I about that card. (Activity menu).

The host with the 210 completed and reported two 6.10 tasks and they are pending validation, but no errors. Not sure on the other 3, should have crunched by now but may be hung up in the upload/report process.

yeah...I think I am learning about cpu fallback. Seems like it occurs if you switch between two gpu projects...

If you think you're running out of GPU memory (or you don't like the higher priority) with the x38g app, try the x39e diagnostic build,
or you can always try running the Stock 6.08 app, that should run with ~220Mb free, i did an app_info for Ricky Graham's 256Mb 8400 GS recently:

http://setiathome.berkeley.edu/forum_thread.php?id=65518

Claggy

Been looking into a new vid card. Trying to stay under $100. How about ATI cards?

There you go
or these slightly older models
Don't expect world class speed for under $100

---

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

Been looking into a new vid card. Trying to stay under $100. How about ATI cards?

I wanted to replace my 8500 with a GT 430, but the PO for it at work was never approved.

---

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

As in everything, I guess you get what you pay for.

I guess gddr5 is better than ddr3?

How about when they say 96 or 192 process cores...how many WUs get crunched at one time?

I do have an sli motherboard...680iLT

...
How about when they say 96 or 192 process cores...how many WUs get crunched at one time?
...

For parts of the processing which can be done using parallel processing of 192 or more nearly identical sequences, 192 cores will be about twice as fast as 96 given the same clock rate. But the overall time spent doing GPU tasks is heavily affected by the parts which cannot be easily broken down that far, parts which need significant amount of data transfer between host memory and the GPU card, and delays until the CPU can tell the GPU what to do next. IOW, the more cores the better in general, but it's not a simple ratio.

                                                                   Joe