...so I started to experiment with different settings of <count> in the app_info.

atm I run that with a settting of 0.3 Seems to work, but display of the progress-bar acts strange. Sometimes two wu's show progress, sometimes none...

Did you set the -instances value according to your <count> variable? (Not sure if that applies to your application(s))

Gruß,
Gundolf

Did you set the -instances value according to your <count> variable? (Not sure if that applies to your application(s))

Gruß,
Gundolf

??? Ahmm .. No.

<cmdline>-period_iterations_num 5 -instances_per_device 1</cmdline>

Looks like there is more to do.
instances per device is selfexplaining (I hope so), but what is period iterations num?

I found this:

I'm not so sure what you can do with the ATI MB app, i'd leave it as it is until you're lowered the APR enough that you're no longer at risk of -177's, then try and run two instances again, (but with an increased -period_iterations_num number too)

With the ATI AP app, try increasing the unroll value and the -ffa_block value up to 6144, again probably best if you did these next 9 Wu's as slow as possible to get the APR down first,

Claggy

Sorry, this raises more questions than answers were given.

Is there more info available?

Alexander

---

First of all install other drivers.
As Raistmer said try 11.2.
With 11.6 you won´t get it work.

instances 1 needs count 1
instances 2 count 0.5
instances 3 count 0.33 and so on.

But you should get it work with instances 1 first since Raistmers card seems not to work with 2 instances.

---

With each crime and every kindness we birth our future.

It's starting and stopping Boinc in the background that affects the running processes.

And it's the background boinc.exe process which has to be stopped and restarted for an application change to take effect.
...

Richard, many Windows updates require restarting Windows. When done as a manual install the final step of the process may have two buttons, one to immediately Restart and the other allowing the user to do the restart later. Is something similar possible here? The actual install is done, it's just a matter of stopping and restarting BOINC so it will read the app_info.xml and thereby activate the changes.

                                                                Joe

It's starting and stopping Boinc in the background that affects the running processes.

And it's the background boinc.exe process which has to be stopped and restarted for an application change to take effect.
...

Richard, many Windows updates require restarting Windows. When done as a manual install the final step of the process may have two buttons, one to immediately Restart and the other allowing the user to do the restart later. Is something similar possible here? The actual install is done, it's just a matter of stopping and restarting BOINC so it will read the app_info.xml and thereby activate the changes.

                                                                Joe

Hmmm. There would be no problem with generating a new app_info.xml, or even with putting new applications into the project directory ready for use. But the current installer moves your old applications into a backup directory, and cleans up the clutter in the working directory - I don't think that housekeeping can be done while SETI is active. And although we can suspend the project (as opposed to closing down BOINC), I suspect that CPU executables would remain locked if the user had 'keep applications in memory' selected - I don't think we can over-ride that and force a dismount without exiting BOINC.

Windows, and application installers, can pull off the trick because it has hooks to run any unfished file operations at an early enough stage in the next reboot cycle for the files not to be in use. I don't think BOINC has any similar facility during the BOINC startup phase, and I suspect users here would be even less happy about being asked for a full Windows restart after a Lunatics installation. Perhaps we should take this over to the development discussion area.

First of all install other drivers.
As Raistmer said try 11.2.
With 11.6 you won´t get it work.

instances 1 needs count 1
instances 2 count 0.5
instances 3 count 0.33 and so on.

But you should get it work with instances 1 first since Raistmers card seems not to work with 2 instances.

THX to the help of all of you it works now with a count of 0.5 and GPU-usage not less than 80% (and a max of 98%) --> goal reached!

@Mike: ols check your PM.

Cheers to all,

Alexander

---

First of all install other drivers.
As Raistmer said try 11.2.
With 11.6 you won´t get it work.

instances 1 needs count 1
instances 2 count 0.5
instances 3 count 0.33 and so on.

But you should get it work with instances 1 first since Raistmers card seems not to work with 2 instances.

THX to the help of all of you it works now with a count of 0.5 and GPU-usage not less than 80% (and a max of 98%) --> goal reached!

@Mike: ols check your PM.

Cheers to all,

Alexander

Replied already.

---

With each crime and every kindness we birth our future.

I have a Core i7 Extreme 980X CPU should i select the SSE4.1 Option or should i stay with the SSE3 (x64, generic)?

And whats with the MultiBeam (Cuda - WARNING requieres 236MiB AVAILABLE Video RAM) option, should i enable processing MultiBeam tasks on NBidia Cuda GPU or is this for special CUDA cards? I have and GTX480

---

* Hi! I'm a signature virus! Copy me into your ~/.signature to help me spread! *

I have a Core i7 Extreme 980X CPU should i select the SSE4.1 Option or should i stay with the SSE3 (x64, generic)?

And whats with the MultiBeam (Cuda - WARNING requieres 236MiB AVAILABLE Video RAM) option, should i enable processing MultiBeam tasks on NBidia Cuda GPU or is this for special CUDA cards? I have and GTX480

For most Intel Core-series processors, it's a toss-up between SSE4.1 and SSSE3x - note that triple 'S'. The balance usually falls in the direction of SSS3x, unless you have a super-fast memory subsystem - but there won't be much in it. Either will be significantly faster than the generic SSE3.

The memory warning on the CUDA app is for users of first-generation CUDA cards with only 256 MB video RAM on-card - your GTX480 will be absolutely fine. Just watch out for temperatures - the CUDA app pushes the card much harder and more continuously than gaming does.

[Post deleted by user]

[Post deleted by user]

Why did you delete the question?

I just PM'd Jason asking him to look at it.

---

"Time is simply the mechanism that keeps everything from happening all at once."

[Post deleted by user]

Why did you delete the question?

I just PM'd Jason asking him to look at it.

If it was an installer question, Carola or I might be able to deal with it - but unfortunately I wasn't watching at the right moment.

Hi Guys, It was a highly technical optimisation related question that Peter elected to ask me via PM. Sorting shortly.

Jason.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Hi Guys, It was a highly technical optimisation related question that Peter elected to ask me via PM. Sorting shortly.

Jason.

That good then....I thought you would be interested in the question.
Glad he went directly to you so you were aware of it.

---

"Time is simply the mechanism that keeps everything from happening all at once."

That good then....I thought you would be interested in the question.
Glad he went directly to you so you were aware of it.

Well hopefully my answer helped him figure what was going on (i.e. lower contention induced by x38g over previous builds). Yes my mods appear to favour high performance machines (such as your own) at this time, and there are a bunch of relevant optimisations I can point to that contribute to that each in a small way. Apart from a few more small hurdles to jump, it should be smooth sailing into the next stages.

Jason

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

it should be smooth sailing into the next stages.

Not if I keep running tests for you!! :-)

---

PROUD MEMBER OF Team Starfire World BOINC

Not if I keep running tests for you!! :-)

You're doing just fine :)

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

I am experiencing 30 secs of system-pausing when the GPU is reloaded since installing v0.38. Where can I backtrack to the v0.37 or is there anything else I can do?

Thanks.

I am experiencing 30 secs of system-pausing when the GPU is reloaded since installing v0.38. Where can I backtrack to the v0.37 or is there anything else I can do?

Thanks.

Hi there. Your original files from a previous installation should have been backed up & kept in a backup folder within your project folder. You could either look for those and replace them in the project folder (manually), or uninstall the 0.38 installation via control panel. That *should* put everything back.

I've seen no reports yet of 30 second freezes, but I'll certainly be on the lookout for them now.

Jason

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

I am experiencing 30 secs of system-pausing when the GPU is reloaded since installing v0.38. Where can I backtrack to the v0.37 or is there anything else I can do?

Thanks.

Hi there. Your original files from a previous installation should have been backed up & kept in a backup folder within your project folder. You could either look for those and replace them in the project folder (manually), or uninstall the 0.38 installation via control panel. That *should* put everything back.

I've seen no reports yet of 30 second freezes, but I'll certainly be on the lookout for them now.

Jason

I have noticed that intermittently on 5530010. I figured it had to do with the higher task priority in that version. Watching task manager the GPU app maxes out a core for the first few seconds it starts & then everything goes back to normal. Since tasks normally take a few hours on that GPU it doesn't really effect me enough to worry about it.
I also I I remember you mentioned else ware the thread priority was or will be lowered later.

---

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