Welcome to the MacNN Forums.

Gecko_r7 · May 14, 2006, 01:51 AM

Originally Posted by lepetitmartien

Gecko, verify the app_info.xml reflex the correct enhanced client name, if it's still in 5.12 you're out of luck (it happened to me this saturday afternoon…), if the .xml is corrected already then the issue is elsewhere. I had to restart the computer to enable BOINC/a53/enhanced to process without errors. Note that if you have a WU on the go you may lost it but it'll be "normal".

I'd say there's something we don't have control in the equation that settles down nicely by itself…

Just in case, the content of the xml set in 513, even the <version_num>513</version_num>

uh… it doesn't show correctly, how can I make the xml tags show up?

Thanks. I spent all day trying to sort this out, but could not make the switching work.

It would process S@H Standard WUs just fine, but would trash all the Enhanced WUs and give the same error as shown in my post. I've re-installed Enhanced 5.13 Ap a couple of times. It just won't work in this switching mode. Are you using Boinc 5.4.9? It's very possible I may be doing something wrong, but I followed your previous posts that have worked for you and Jedimstr(?). I've got my Pent M-1.6 rig switching great w/ Crunch3rs Op aps but I'll admit that the migrations, updates, trouble-shooting/trial-error and forum monitoring between the Mac and PC setups has probably fried me by now.

I did a fresh install of 5.4.9, re-attached and got SE 5.13 and it's now working beautifully, albiet no switching, so no more Alpha 5.3

Unless someone can magically see from this limited info what I've somehow missed, guess I'll be riding the slow lane for a while until Alex and Rick can speed-up this Enhanced turtle

Suggestions are always welcomed from the gang. Best regards! Ian

jedimstr · May 14, 2006, 09:16 AM

Gecko, if you downloaded the new 5.13 client with 5.4.9 and want it to work with the A5.3, make sure you check the spelling of the 513 entry's filename in the app_info.xml. I had a problem recently similar to yours when I upgraded from 5.12 to 5.13. Apparently there was slight change to the filename other than just the version number.

setiathome-5.12-powerpc-apple-darwin became...
setiathome_5.13_powerpc-apple-darwin

Note the change from hyphens around the version number to underscores. Without changing that little typo in the app_info.xml, it caused a computation error everytime the boinc client tried to run a 5.13 WU.

lepetitmartien · May 14, 2006, 01:26 PM

I'm running BOINC 5.4.9, I have still not seen a switch between a53 and Enhanced as I've had only enhanced WU to crunch in 2 days…

Right now, the most important thing is we need an optimized enhanced

btw jedimstr retold in a more simple way what I wrote earlier, if you have the xml to match, it'll never work. Now welcome to the slow cruncher club again ! LOL

I'm seeing the Quads loosing their grip as the best crunchers for Seti it's depressing

sdubz · 05:43 PM

Hello all!

With all of the trouble and confusion going on with the new enhanced seti, I figured I could add a little to it.

I compiled seti enhanced 5.13 with the best optimizations I could to try to maybe tie myself over for a while until the guys here made something better. And it seems to be running faster on my g4 mini.

You may have to edit the app_info.xml to better suit your configuration, but here are the files:
EDIT: the new ones are over here: http://forums.macnn.com/showthread.php?t=295708

Gecko_r7 · May 14, 2006, 04:05 PM

Originally Posted by jedimstr

Gecko, if you downloaded the new 5.13 client with 5.4.9 and want it to work with the A5.3, make sure you check the spelling of the 513 entry's filename in the app_info.xml. I had a problem recently similar to yours when I upgraded from 5.12 to 5.13. Apparently there was slight change to the filename other than just the version number.

setiathome-5.12-powerpc-apple-darwin became...
setiathome_5.13_powerpc-apple-darwin

Note the change from hyphens around the version number to underscores. Without changing that little typo in the app_info.xml, it caused a computation error everytime the boinc client tried to run a 5.13 WU.

Thanks jedimstr and lepetitmartien! It works! Indeed, I missed the slight change in application description between 5.12 and 5.13 you pointed-out jedimstr. That's what was causing the problem. Ironic, I even checked this before, and didn't notice what was right in front of me. As I said, probably just X-eyed w/all the changes the past few days. Enhanced is taking a bloody eternity on my 1.33 G4...almost like going back to 99' when the original screen saver ap would take 120hrs on my PII 200mmx....well not quite that bad

arkayn · May 14, 2006, 04:57 PM

I think the new 5.4.9 manager is picky about names, it downloaded a new 4.1.8 processor even though I had everything pointing to the alpha 5.3.

I am going to see if changing the name of the alpha fixes the problem.

Gecko_r7 · May 14, 2006, 05:12 PM

Originally Posted by arkayn

I think the new 5.4.9 manager is picky about names, it downloaded a new 4.1.8 processor even though I had everything pointing to the alpha 5.3.

I am going to see if changing the name of the alpha fixes the problem.

I ran into the same thing, but at the time, I had only 5.13 loaded and I thought that since the downloader sent a couple of 4.18 WUs later on and didn't see an application in the folder, it sent a 4.18 Ap. Since then, I re-added Aplha 5.3 to the folder and the switching app_info.xml. So, far, so good. I'm waiting to see if it trys to send another 4.18 ap file with the next set of downloaded 4.18 WUs. If it, does, I'll try the rename as well. Cheers!

lepetitmartien · May 14, 2006, 07:04 PM

I've found another issue with the "mixed" mode, each time I start up BOINC I get this:

Dim 14 mai 23:41:59 2006|SETI@home|Scheduler request succeeded
Dim 14 mai 23:42:01 2006|SETI@home|Started download of file 28fe99aa.4634.2689.729828.1.71
Dim 14 mai 23:42:05 2006|SETI@home|Finished download of file 28fe99aa.4634.2689.729828.1.71
Dim 14 mai 23:42:05 2006|SETI@home|Throughput 97823 bytes/sec
Dim 14 mai 23:42:06 2006||Rescheduling CPU: files downloaded
Dim 14 mai 23:42:06 2006|SETI@home|Starting task 28fe99aa.4634.2689.729828.1.71_0 using setiathome version 418
Dim 14 mai 23:42:07 2006|SETI@home|Unrecoverable error for result 28fe99aa.4634.2689.729828.1.71_0 (process exited with code 2 (0x2))
Dim 14 mai 23:42:07 2006||Rescheduling CPU: application exited
Dim 14 mai 23:42:07 2006|SETI@home|Computation for task 28fe99aa.4634.2689.729828.1.71_0 finished

So the first WU is considered as a normal one and exit as an error. then the following is processed perfectly. (I get only enhanced)

I'm going to try your G5 client boog

I was too lazy to try a compile (to busy), and I'm not geek enough to dive into the code and to some real job there.

lepetitmartien · May 14, 2006, 07:11 PM

Installed boog's client, it continued processing the WU I had, no crash, I'll see if it ends up fine. I can't tell if it goes faster…

sdubz · May 14, 2006, 08:35 PM

Originally Posted by lepetitmartien

Installed boog's client, it continued processing the WU I had, no crash, I'll see if it ends up fine. I can't tell if it goes faster…

mine seemed to do ok, but now I'm starting to worry about it. Starting to get errors.

stderr out
<core_client_version>5.5.0</core_client_version>
<stderr_txt>
SETI@Home Informational message -9 result_overflow
NOTE: The number of results detected exceeds the storage space allocated.

</stderr_txt>

Gecko_r7 · May 14, 2006, 08:56 PM

Originally Posted by lepetitmartien

Installed boog's client, it continued processing the WU I had, no crash, I'll see if it ends up fine. I can't tell if it goes faster…

No problems w/ the switching on mine today. No errors. BTW, Boog, your application is working GREAT! on my G4 so far. Non-opt 5.13 was coming in @ 12 hours per WU, yours @ 5. Not too shabby

Thanks! Also curious if Boinc 5.4.9 is underclaiming compared to a couple of others, or if others are "over claiming" Check these out:

http://setiathome.berkeley.edu/worku...?wuid=78072116
http://setiathome.berkeley.edu/worku...?wuid=78070535

Funny, the OSX 5.2.13 & 5.4.9 are claiming the same on WUID 7870535, but the clients used by the AMD and Intel rigs in WUID 78072116 are claiming substantailly more and identical credit, even though WU completion times differ.

Interesting? Ideas?

arkayn · May 14, 2006, 09:25 PM

Renaming the optimized client resulted in download errors on my system, so I have went back to the actual names of the clients and installed boog's app_info and will see what that does.

lepetitmartien · May 14, 2006, 09:45 PM

I'm at about 37% of the WU, gone under Goos' from 10% on, still doing ok. Estimated time remaining about 4.5 hours, we'll see… It seems more altivec friendly like the old optimized (thanks to menumeter for guessing).

Just tried to process a classic WU, it makes the same error as the official

Lun 15 mai 03:46:28 2006|SETI@home|Unrecoverable error for result 28fe99aa.4634.3953.336096.1.190_2 (process exited with code 2 (0x2))

Is there only me having this error on the "classic" WU? :'( Don't let me out in the dark and the frize!

Gecko_r7 · May 14, 2006, 09:53 PM

Originally Posted by boog

mine seemed to do ok, but now I'm starting to worry about it. Starting to get errors.

stderr out
<core_client_version>5.5.0</core_client_version>
<stderr_txt>
SETI@Home Informational message -9 result_overflow
NOTE: The number of results detected exceeds the storage space allocated.

</stderr_txt>

Boog: Yikes! I'm getting the same here.

Each one has has crunched fine, but they are not validating on my G4.

http://setiathome.berkeley.edu/resul...ltid=325216316
http://setiathome.berkeley.edu/resul...ltid=325212451

I will go back to non-opt until the gremlin is found and squashed. This will be a good start on an Op Ap for us if you can get the problem sorted out. Good luck! Fingers X-ed!

Gecko_r7 · May 14, 2006, 09:55 PM

Originally Posted by lepetitmartien

I'm at about 37% of the WU, gone under Goos' from 10% on, still doing ok. Estimated time remaining about 4.5 hours, we'll see… It seems more altivec friendly like the old optimized (thanks to menumeter for guessing).

Just tried to process a classic WU, it makes the same error as the official

Is there only me having this error on the "classic" WU? :'( Don't let me out in the dark and the frize!

I'm not getting errors on S.Standard WUs...so far w/ the switching ap.

lepetitmartien · May 14, 2006, 10:33 PM

So, something is fishy here… doesn't seem to be the xml… could it be the wisdom file? let's go see about that…

sdubz · May 14, 2006, 10:36 PM

Originally Posted by Gecko_r7

Boog: Yikes! I'm getting the same here.

Each one has has crunched fine, but they are not validating on my G4.

http://setiathome.berkeley.edu/resul...ltid=325216316
http://setiathome.berkeley.edu/resul...ltid=325212451

I will go back to non-opt until the gremlin is found and squashed. This will be a good start on an Op Ap for us if you can get the problem sorted out. Good luck! Fingers X-ed!

Thanks! I'm working on trying to figure out what it may be. The waiting is the hard part, change something then crunch 1 wu, while suspending network activity and all but the 1 crunching wu.

I'll keep you guys posted on what I find. I'll start a new thread when I find something so that I will stop jacking this one.

now back to

sdubz · May 14, 2006, 10:43 PM

Originally Posted by lepetitmartien

So, something is fishy here… doesn't seem to be the xml… could it be the wisdom file? let's go see about that…

It could be because I am using the one from my files without mention of the wisdom an it works perfect at switching between apps. (can't figure out how to post the xml code here)

lepetitmartien · May 14, 2006, 10:50 PM

Well, I've tried both wisdom files available for my computer (there's 2 wisdom for G5 1.8…) I get the error. So it's not the wisdom itself. Could you send me your xml? Take my login here and add @macmusic.org it'll come along just fine

Thanks in advance.

On the time taken by boog's client, the WU was scheduled for about 5 hours and a half by the official client, it'll be done in about 3 hours it seems by boog's… if it finishes fine. I'm 89% thru and 22 minutes remaining.

sdubz · May 14, 2006, 11:05 PM

Originally Posted by lepetitmartien

Could you send me your xml? Take my login here and add @macmusic.org it'll come along just fine

Thanks in advance.

Sent!

Gecko_r7 · May 14, 2006, 11:15 PM

Originally Posted by boog

Thanks! I'm working on trying to figure out what it may be. The waiting is the hard part, change something then crunch 1 wu, while suspending network activity and all but the 1 crunching wu.

I'll keep you guys posted on what I find. I'll start a new thread when I find something so that I will stop jacking this one.

now back to

Thanks Boog. Keep the faith! If you are able to find and fix the problem, while preserving the speed, you've got the makes of a pretty speedy ap. I only had 2 WUs finish, so I'm not sure if others would have been longer or shorter, but these two finished VERY close to what I'm getting from a M-1.6 w/ Crunch3r's Opt 5.13 ap. Not bad for a G4 1.33 considering the M is quite efficient and so are Crunch3rs aps. In fact, w/ Crunch3rs Standard Ap for Pent M w/ SSE2, it would crunch in @ 3200secs avg vs. @ 5300 for A5.3 on my G4. Either Crunch3r's new SE Opt 5.13 still has a lot of headroom for improvement, or maybe Power PC users will comparatively benefit more from an Altivec Opt Ap w/ Seti E. ??? Maybe ALex or Rick can better weigh-in on this? 2 WUs certainly is not an edequate sample population, but if I understand correctly, one can compare True Angle Ranges to determine similar length units. If this is the case, my M 1.6 and G4 processed units w/ identical TARs, but the G4 w/ Boogs Ap took only 2000secs longer (18K for G4 vs. 16K for M-1.6 w/ Crunch3r's Opt 5.13) The closest gap ever between these two rigs IF this is consistant behavior, and the Ap can validate while preserving the speed. Exciting stuff!

Gecko_r7 · May 14, 2006, 11:22 PM

Originally Posted by lepetitmartien

Well, I've tried both wisdom files available for my computer (there's 2 wisdom for G5 1.8…) I get the error. So it's not the wisdom itself. Could you send me your xml? Take my login here and add @macmusic.org it'll come along just fine

Thanks in advance.

On the time taken by boog's client, the WU was scheduled for about 5 hours and a half by the official client, it'll be done in about 3 hours it seems by boog's… if it finishes fine. I'm 89% thru and 22 minutes remaining.

I sent you a copy of the one I'm using as well. No problems w/ the switching.

lepetitmartien · May 14, 2006, 11:24 PM

Well… seems to work… going to sleep now. Next episode, did it work till the end?

Thanks for the xml

(all the parts about the wisdom file with the relevant tags are removed)

lepetitmartien · May 15, 2006, 09:13 AM

The first classic WU finished in 5,965 seconds and has been validated, there's already been 2 Enhanced WU that went thru all right, both validated.

AND it's crunching faster

(about twice speed seemingly)

Boog, seems you're our saver

sdubz · May 15, 2006, 04:05 PM

Originally Posted by lepetitmartien

The first classic WU finished in 5,965 seconds and has been validated, there's already been 2 Enhanced WU that went thru all right, both validated.

AND it's crunching faster

(about twice speed seemingly)

Boog, seems you're our saver

you got my seti worker running and they are validating? cool, maybe I am banging my head for nothing?

Thanar · 04:44 PM

Checking boog's worker right now under BOINC 5.4.9 CLI... Working on an enhanced WU, guess it will take some time... Will wait for the premier result and then modify the app_info in order to include my bigfft_wisdom. Will let you posted!

btw... boog, I don't care what you say man about not being a developer, but if your worker really works, I say you should put aside your modesty and work together with Alex and Rick!

jedimstr · May 15, 2006, 05:18 PM

On my G5 Quad and Powerbook G4 1.5 I'm starting to see the same errors that Booq noted previously for his apps.

-9 result_overflow

In the interest of tracking down the issue and perhaps catching at least one (hopefully more than that) 5.13 WU that validates, I'll leave the Booq apps working overnight. I know this will mean tons of potentially lost credit, but I've been losing RAC for my biggest hitter (the Quad G5) during the switch to Seti Enhanced anyway. I figure the faster we help Booq (and maybe Alex and Rick) get the Altivec optimized Enhanced apps working correctly, the faster I can get something approaching my old RAC back.

For Booq and others if you want to check the results:

G4 PB - http://setiathome.berkeley.edu/show_...hostid=1621428
G5 Quad - http://setiathome.berkeley.edu/show_...hostid=2248487

Gecko_r7 · May 15, 2006, 05:47 PM

Can we confirm if anyone has actually had any validated WUs w/ 5.4.9 and Boog's speedy ap?
I inadvertantly trashed so many 5.13 units initially trying to get the ap switching to work (even w/ cache at 1/10 of a day), I felt guilty about trashing more after upgrading to the 5.13 opt ap and the first 2 WUs came back invalid w/ "-9 result_overflow" errors. If you need another G4 in the mix to test when you're ready, I'm glad to help.

sdubz · May 15, 2006, 05:50 PM

Originally Posted by jedimstr

On my G5 Quad and Powerbook G4 1.5 I'm starting to see the same errors that Booq noted previously for his apps.

-9 result_overflow

In the interest of tracking down the issue and perhaps catching at least one (hopefully more than that) 5.13 WU that validates, I'll leave the Booq apps working overnight. I know this will mean tons of potentially lost credit, but I've been losing RAC for my biggest hitter (the Quad G5) during the switch to Seti Enhanced anyway. I figure the faster we help Booq (and maybe Alex and Rick) get the Altivec optimized Enhanced apps working correctly, the faster I can get something approaching my old RAC back.

For Booq and others if you want to check the results:

G4 PB - http://setiathome.berkeley.edu/show_...hostid=1621428
G5 Quad - http://setiathome.berkeley.edu/show_...hostid=2248487

WOW! I love that you guys are willing to help! Thank you!

I have temp removed the files from my server, because it looks like they are erroring way bad! I have done tons of research today and may have another build to test soon.

I will keep you guys updated!

Thanar · May 15, 2006, 05:58 PM

Does the -9 error pop up after the WU computation is completed, or sometime in the middle? I am almost at 50% of an enchanced WU right now and the stderr file is still empty... Should I let it complete calculating overnight (it's 01:00 here)?

sdubz · May 15, 2006, 06:23 PM

Originally Posted by Thanar

Does the -9 error pop up after the WU computation is completed, or sometime in the middle? I am almost at 50% of an enchanced WU right now and the stderr file is still empty... Should I let it complete calculating overnight (it's 01:00 here)?

It seems to pop up at the end, I would say go back to the normal seti enhanced for now till I see if I can get at least 1 good wu out of my mini (rebuilding my compiling environment right now).

Gecko_r7 · May 15, 2006, 06:46 PM

Originally Posted by Thanar

Does the -9 error pop up after the WU computation is completed, or sometime in the middle? I am almost at 50% of an enchanced WU right now and the stderr file is still empty... Should I let it complete calculating overnight (it's 01:00 here)?

I didn't see the error until the end after the WU was uploaded. No problems showed up while crunching...looked to be going perfectly.

lepetitmartien · May 15, 2006, 09:22 PM

Well, after about 20 hours of crunching with boog's, mixed feeling…

http://setiathome.berkeley.edu/resul...?userid=957118
Boog's enhanced and the a53 are on from the WU reported at 15 May 2006 7:10:07 UTC on

you can see my WU, it's boog's on the enhanced and ye olde optimized a53 on the normal WU. Something is fishy.

It seemed on enhanced and one a53 validated.

(don't get lost in the "pink" lines, I had previously errors coming from the wisdom file named in the xml, so check the hours carefully to pick up the right ones)

Gecko_r7 · May 15, 2006, 10:01 PM

Lepetitmartien: Link doesn't work. Is your computer hidden?

lepetitmartien · May 15, 2006, 10:28 PM

Mhmm "Should SETI@home show your computers on its web site? yes" so I think it's ok…

my results
Just in case…

now I'm going

else I'll

May 15, 2006, 11:01 PM

I retired the 5.21.3 Menubar in favor of the 5.4.9 Manager, and removed the G4-a5 along with the xml file after setting SETI to no new work. All the 5.13 SETI files seem to have downloaded into the new project folder. And it worked fine for a day or two, but now SETI says it's trying to download a 4.18 WU, and the messages says it can't initialize file transfer of g4-a5... why would that be? What am I missing?

lepetitmartien · May 15, 2006, 11:06 PM

Have you still the app_info.xml in the project folder? It looks as if you have it still there and BOINC misses the g4-a5 to process a 4.18 WU. It tries to DL it but as it's not at SETI he can't retrieve it and you get an error…

May 15, 2006, 11:22 PM

No, that's why I'm so confused. I set to no new work, let the last 4.18 WU upload, then removed the xml and 4.18 app from the folder -- it was completely empty. When I restarted, it connected and only grabbed 5.13 app files. Would the .plist file still have some 4.18 preference info in it?

Gecko_r7 · May 16, 2006, 02:32 AM

Originally Posted by lepetitmartien

Mhmm "Should SETI@home show your computers on its web site? yes" so I think it's ok…

my results
Just in case…

now I'm going

else I'll

Sorry, I can't see them. When I click on the link, it says "unable to handle request" No Acccess.

lepetitmartien · May 16, 2006, 09:15 AM

And with this one?

The assimilators are down, so I can't tell how the WU I've done since this morning are doing. I'll move back to the official app at the end of the current WU.

andrew, I wonder if the stuff in the slots folder is not interfering in some way… (don't touch! it's atomic science). Have you tried to stop BOINC, do some basic maintenance (permissions, caches, system optimization) using something like OnyX or equivalent? Then restart the computer. :-/

Todd Madson · May 16, 2006, 10:54 AM

Geez. I think I'm just going to wait until this all gets sorted out before switching.

May 16, 2006, 02:31 PM

I restarted the computer, then reset the project. It re-downloaded all the 5.13 files into the proj folder and grabbed a new 5.13 enhanced WU. Both my machines say these will take 13 hrs to process. Yeeesh.

brysonda · May 16, 2006, 02:38 PM

Originally Posted by Todd Madson

Geez. I think I'm just going to wait until this all gets sorted out before switching.

I'm testing a hybrid a53, enhanced (boog compiled version) setup on one of my systems (1 GHz PB G4), but I definitely am going to wait it out before transitioning more. I of course want to support the science, but I'm going to keep my Quad at 5.2.13/a53 until I really have to switch.

lepetitmartien · 04:21 PM

well unable to upload the results, and now 2 4.18 WU went in error right on spot… going to restart all this soon…

[edit]My fault, a mix up in the xml files…
Else I had an error common with Enhanced, the unit that goes on crunching for ever and ever and ever…

It seems the unit unable to upload today are the enhanced only.

lepetitmartien · May 16, 2006, 06:58 PM

They uploaded at last, I've got another enhanced thru and validated, bu not the other. The bad WU worked too fast seemingly.

Gecko_r7 · May 16, 2006, 07:19 PM

Originally Posted by lepetitmartien

well unable to upload the results, and now 2 4.18 WU went in error right on spot… going to restart all this soon…

[edit]My fault, a mix up in the xml files…
Else I had an error common with Enhanced, the unit that goes on crunching for ever and ever and ever…

It seems the unit unable to upload today are the enhanced only.

This kind of error shows that boinc can't find the application, and will either be an xml issue as you mentioned, with a "mismatch" between the application and xml description, or that the application is not located where boinc is looking for it as a few others have run into, especially if the application is placed in the wrong folder.

BTW, your most recent link works and I can see you've experienced the same errors as Boog and I, with the exception of 2 WUs, which may be tied-up w/ the server outage since they're still showing "initial". I take it that these were processed w/ the opt ap Boog is working on?

http://setiathome.berkeley.edu/resul...ltid=326610319
http://setiathome.berkeley.edu/resul...ltid=325977756

Cheers!
Ian

lepetitmartien · May 16, 2006, 07:43 PM

Gecko I know for the app name, as I said in my 09:31 PM post (edited part).

All enhanced WU since the one reported the 15/05 at 7:10 UTC are done with Bogg's compile. I've switched back to the official client now until there's some now effort out.

note that this one
http://setiathome.berkeley.edu/worku...?wuid=78406404
I was the only one with no error… LOL