I got bit by this bug also and wondered why the installation deleted the 521 AP NV app. Luckily I had the original unpacked directory in my backup and since there hasn't been any work available for a couple of days now.... no harm, no foul. The installer should be fixed to not delete this app even though you still have to edit your app_info to get all your mods back in place.

Cheers, Keith

---

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

A proud member of the OFA (Old Farts Association)

Skildude, Now that I've sent in a bunch more I still don't have that many -9s but I do have a lot of completed that are only giving five or six credits. Very strange. Just an example http://setiathome.berkeley.edu/workunit.php?wuid=869407967

it looks like 11.9 is stable enough for now. I'm not getting any errors.

Your results look interesting. I've not seen that sort of shorty before

---

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

Regarding -9 overflow results:

We very rarely see extended runs of -9 data on the tapes. You do see short clusters of -9 sometimes, but usually only on tasks which are closely grouped together, according to the other numbers in the task names. Check the numbers too: if a run of -9 continues into a different group of tasks, it's probably a problem - suspend processing and investigate.

First, check and see if your wingmates have reported the same tasks. If both computers report -9, and they validate together, things are probably OK.

Otherwise, some common -9 problems are:
on CPUs - overclocking or overheating. If you haven't changed anything recently, look for dust bunnies.
on NVIDIA GPUs - a computer reboot may cure it. If overclocked, back down a bit: GPUs can pass standard OC tests, but still fail to produce valid SETI results, especially under the added stress of the optimised CUDA apps.
on AMD/ATI GPUs - possible driver incompatibilities. If you downgrade from a newer to an older driver, run a driver cleaner to ensure all the newer components have been fully removed - installation programs, even ours, might be programmed not to overwrite newer files with older files. And the comments on rebooting and overclocking can apply to ATI cards too.

If the problems continue after you've checked all that, post here.

Nvidia does that on more recent drivers, I think It started around 275, just click on Custom, then check off clean install, then proceed to install what You want or don't want.

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

I got bit by this bug also and wondered why the installation deleted the 521 AP NV app. Luckily I had the original unpacked directory in my backup and since there hasn't been any work available for a couple of days now.... no harm, no foul. The installer should be fixed to not delete this app even though you still have to edit your app_info to get all your mods back in place.

Cheers, Keith

ATM the installer rewrites app_info.xml from scratch each time. Hence the warning to make a note of any manual edits.
Next time I'll make that more prominent and emphasise it concerns all app_info.xml edits.

The .exe being deleted from the project directory is the overenthusiasm displayed by some BOINC versions when it comes to cleaning up 'unused apps'

We are currently discussing at Lunatics what we want from a GUI app_info.xml editing tool (flops, counts, cmd line settings, ...). Considering the limited time of all parties involved (especially those who could actually code such stuff), it may or may not be ready by the time we need an installer for V7.

Nvidia does that on more recent drivers, I think It started around 275, just click on Custom, then check off clean install, then proceed to install what You want or don't want.

That's exactly the problem, ATI tends to leave some files locked, that need manual intervention, especially when downgrading.

Skildude, Now that I've sent in a bunch more I still don't have that many -9s but I do have a lot of completed that are only giving five or six credits. Very strange. Just an example http://setiathome.berkeley.edu/workunit.php?wuid=869407967

One word: CreditNew

DA is deaf with regard to complaints about his crappy statistical approach to credit.

Could people posting problems please be specific about which application processed the individual task or tasks they're commenting on?

Dave - we could look into that blank stderr_txt, if we knew. Or could you link the ResultID page? After all, you'll have had it open to read/copy that.

It just saves a lot of searching. Thanks.

Richard,
Sorry, with going to bed last night and the site being down today...
Anyway, here's one:
blank stderr

-Dave

[edit]
and one that has credit granted...

Cuda GPU gives also blank results:
blank stderr

That's a host running 6.08 stock - nothing to do with us.

IIRC empty stderrs are a result of problems in the old BOINC api. Shouldn't happen with x41g, but I can't remember if Raistmer switched to the fixed BOINC api for his current builds.

That's a host running 6.08 stock - nothing to do with us.

IIRC empty stderrs are a result of problems in the old BOINC api. Shouldn't happen with x41g, but I can't remember if Raistmer switched to the fixed BOINC api for his current builds.

I'll add that it can still happen with any app, for other reasons closer to (or within) the servers. It's thankful that results don't appear to suffer the same damage (at least so noticeably often), but still frustrating when trying to work out what app was run :)

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 got bit by this bug also and wondered why the installation deleted the 521 AP NV app. Luckily I had the original unpacked directory in my backup and since there hasn't been any work available for a couple of days now.... no harm, no foul. The installer should be fixed to not delete this app even though you still have to edit your app_info to get all your mods back in place.

Cheers, Keith

ATM the installer rewrites app_info.xml from scratch each time. Hence the warning to make a note of any manual edits.
Next time I'll make that more prominent and emphasise it concerns all app_info.xml edits.

The .exe being deleted from the project directory is the overenthusiasm displayed by some BOINC versions when it comes to cleaning up 'unused apps'

We are currently discussing at Lunatics what we want from a GUI app_info.xml editing tool (flops, counts, cmd line settings, ...). Considering the limited time of all parties involved (especially those who could actually code such stuff), it may or may not be ready by the time we need an installer for V7.

I'd also add that the 521 AP NV app is still classified as a beta app. It's a moot point - well worthy of discussion - whether the installer (which is designed as an ease-of-use tool for less technically-inclined users) should make special allowance for manually-installed test applications. There will never be - or at least, it's not even in long-term planning to provide - an equivalent of the installer for placing Beta apps in the SETI Beta project.

In short, I don't think it's right to classify the the removal and required re-installation of a Beta app as a 'bug': but we certainly could, should, and will in future give more prominence to the specials needs of Beta testers in future release announcements.

Richard, OK .. too strong a language to use "bug" for the installer. I do think you are correct in stating that the next installer have a note saying ..... " oh, by the way .... you will have to reinstall your beta apps because they will be deleted by the installer." Just to decrease the panic mode when you restart BOINC and see the red lines in the message log.

Cheers, Keith

---

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

A proud member of the OFA (Old Farts Association)

Had problems with 'normal' install with nVidia card (GT430 updated to latest driver 285.62) slowing/stopping/starting computations and after BOINC Manager restarting after install.

Solved problem by:

#1 - Stopping BOINC Manager.

#2 - Uninstalling Lunatics Application

#3 - Reinstalling Lunatics 0.39

#4 - Restarting computer

Perhaps uninstalling 0.38 before installing .39 is the safest way to go.

There really shouldn't be any need, but if it works on your system, no reason why not.

But if you do:

#1 - it isn't the Manager which you need to stop, but the underlying 'core client'. If you use the Manager to shut down, make sure that 'Stop running tasks when exiting the BOINC Manager' is checked in the Manager exit confirmation dialog.

#2, #3 - ensure that you do not allow BOINC (the core client, again) to restart between these two stages - in particular, be careful not to allow BOINC to start automatically if you reboot here, rather than at the end as Clyde suggests. You'll lose work if you restart BOINC here.

Running x39e with 280.26

Intend to upgrade over weekend, First bit of free time.

One thing that I have not seen mention so far is downclocking, how is this release with regards to downclocking with the last few releases of drivers?

---

Kevin

Running x39e with 280.26

Intend to upgrade over weekend, First bit of free time.

One thing that I have not seen mention so far is downclocking, how is this release with regards to downclocking with the last few releases of drivers?

This is fixed in the new release.

---

With each crime and every kindness we birth our future.

Running x39e with 280.26

Intend to upgrade over weekend, First bit of free time.

One thing that I have not seen mention so far is downclocking, how is this release with regards to downclocking with the last few releases of drivers?

This is fixed in the new release.

Good thank you.

Will upgrade soonest.

---

Kevin

What new funtions of the apps is it that would pursuade me to upgrade from the 0.38 release?

No errors with the apps I'm running now, neither with the ATI apps or the CPU apps.

This was the primary reason for the update.

This is to fix a bug in the triplet finding - in rare cases triplets were being missed. As it took over two years to notice this bug, we believe it only affects a very minor amount of tasks.

---

What new funtions of the apps is it that would pursuade me to upgrade from the 0.38 release?

No errors with the apps I'm running now, neither with the ATI apps or the CPU apps.

The CPU apps have had a bug fixed that resulted in possible missing of triplets. This may lead to only inconclusive with consecutive invalid, but may in cases (i.e. when two faulty apps are paired on the task in question) result in wrong data in the science database - like the desaster with the Fermis and V12/6.09.

Please upgrade.

What new funtions of the apps is it that would pursuade me to upgrade from the 0.38 release?

No errors with the apps I'm running now, neither with the ATI apps or the CPU apps.

This was the primary reason for the update.

This is to fix a bug in the triplet finding - in rare cases triplets were being missed. As it took over two years to notice this bug, we believe it only affects a very minor amount of tasks.

And that bug concerns which app? Meaning which app in the old releases is it I "need" to upgrade? Is it all, or only one, or what?

It affects MB applications - pretty much all of them, in particular CPU and ATI.

Your favourite AP is not affected, but if you ever use MB as winter warmers - you do need to upgrade.

What new funtions of the apps is it that would pursuade me to upgrade from the 0.38 release?

No errors with the apps I'm running now, neither with the ATI apps or the CPU apps.

This was the primary reason for the update.

This is to fix a bug in the triplet finding - in rare cases triplets were being missed. As it took over two years to notice this bug, we believe it only affects a very minor amount of tasks.

And that bug concerns which app? Meaning which app in the old releases is it I "need" to upgrade? Is it all, or only one, or what?

The CPU AKV8b2 MB apps have the Bug fix, as dooes the ATI OpenCL MB apps,

The x41f Cuda app also has a different Bug fix for Nvidia GPU's without Double Precision, (replaced nVidia's shortcut for the <cc1.3 chirp with full df64, to reduce inconclusives. NO FMAD Allowed!!!)

Claggy

Oh my, oh my. Oh well, I'll download and extract what's needed then. For my ATI machine I'll upgrade the AK_v8b_win_x64_SSSE3x.exe then, and the MB_6.10_win_SSE3_ATI_r177.exe apps. For my Nvidias it'll be the AK_v8b_win_x64_SSSE3x.exe and the x38g_cuda32 to upgrade.

Oh the pain, the pain....Here goes, time to upgrade and edit app_info. Pain pain pain :-)

By NVidias, you'll include the i3? When testing AK v8b2 before release, I was surprised to find that the good ol' SSE3 app was significantly faster that SSSE3x on my 64-bit i5 installation: not so much at VHAR, but definitely so for the VLARs that tend to get picked up by CPUs. We've flagged that in the installer, but if you're unpacking manually, you wouldn't see that....

We didn't have time to locate an i3 for specific speed testing, but it's worth checking.

OK, the i3 is now also upgraded. It will take some days though, before it has chewed up all the AP's it has in the cache, so we can test the speed of the SSE3 app for a while, and then shift to the SSSE3x app for comparison.

Thanks. Keep us posted.