Posts by Purple Rabbit

21) Message boards : Number crunching : Optimized BOINC 4.72 core (Windows/Linux) (Message 173902)
Posted 3 Oct 2005 by Profile Purple Rabbit
Post:
When I try to disable work requests, the text in the button doesn't change, but it does seem to disable work requests. However, I can't reenable except by restarting the BOINC manager. Is this perhaps what you're seeing?


Yes this is what is happening. The difference was that the Linux v4.72 gave the correct status. I used BoincView for the commands as well as the Boinc Manager.

Ah well, I guess the Linux and Windows v4.72 clients work differently in this regard.
22) Message boards : Number crunching : Optimized BOINC 4.72 core (Windows/Linux) (Message 173671)
Posted 2 Oct 2005 by Profile Purple Rabbit
Post:
I see an interesting side effect with Maverick's v4.72 for Windows.

I can't disable work requests with Maverick's optimized version for Windows. Otherwise it works just fine. I can suspend work so the commands are getting through. I didn't try the more destructive commands :-) I put v4.45 back and I could disable work requests. Reinstall Maverick's v4.72 and I couldn't disable work requests.

I have been running Metod's client for v4.72 on Linux for almost two months with no problems.

Is this an artifact of the optimization or am I missing something? I presume that since I can disable work requests with Metod's Linux v4.72 that the BOINC version level isn't the problem? Maybe it's some Windows thingy?
23) Message boards : Number crunching : Paul is Looking for message log entries (again) (Message 119344)
Posted 5 Jun 2005 by Profile Purple Rabbit
Post:
Error -108

2005-05-31 08:00:14 [ProteinPredictorAtHome] Giving up on download of h0008A_1_49618.ini: file was not found on server
2005-05-31 08:00:14 [ProteinPredictorAtHome] Giving up on download of h0008A_1_49618.inp: file was not found on server
md5_file: No such file or directory
md5_file: can't open projectspredictor.scripps.eduh0008A_1_49618.ini
2005-05-31 08:00:14 [ProteinPredictorAtHome] MD5 computation error for h0008A_1_49618.ini: -108
md5_file: No such file or directory
2005-05-31 08:00:14 [ProteinPredictorAtHome] Checksum or signature error for h0008A_1_49618.ini
md5_file: can't open projectspredictor.scripps.eduh0008A_1_49618.inp
2005-05-31 08:00:14 [ProteinPredictorAtHome] MD5 computation error for h0008A_1_49618.inp: -108
2005-05-31 08:00:14 [ProteinPredictorAtHome] Checksum or signature error for h0008A_1_49618.inp
2005-05-31 08:00:14 [ProteinPredictorAtHome] Unrecoverable error for result h0008A_1_49618_9 (WU download error: couldn't get input files:

...XML stuff deleted-it was interesting, but it didn't transfer to the forum nicely :-)

2005-05-31 08:00:15 [ProteinPredictorAtHome] Giving up on download of h0008A_1_49618.seq: file was not found on server
md5_file: No such file or directory

24) Message boards : Number crunching : Paul is Looking for message log entries (again) (Message 119317)
Posted 5 Jun 2005 by Profile Purple Rabbit
Post:
Error -107. I think this is new. I checked Paul's list and didn't find it. Two WU failed at the transition time (non-HT machine). Predictor WU
just completed and SETI WU just starting. Recovery required BOINC V4.4 restart (ultimately a computer reboot).

SETI WU:

2005-06-03 14:36:05 [SETI@home] CreateProcess() failed -
2005-06-03 14:36:08 [SETI@home] Unrecoverable error for result 03dc03ab.22340.16977.984630.78_0 (CreateProcess() failed - )
2005-06-03 14:37:09 [SETI@home] scheduler init_op_project to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi failed, error -107

Predictor WU:

2005-06-03 14:36:09 [ProteinPredictorAtHome] Couldn't start upload of h0009A_1_59166_7_0
2005-06-03 14:36:09 [ProteinPredictorAtHome] URL http://predictor.scripps.edu/predictor_cgi/file_upload_handler: error -107

...etc for the pieces

2005-06-03 14:36:09 [ProteinPredictorAtHome] CreateProcess() failed -
2005-06-03 14:36:12 [ProteinPredictorAtHome] Unrecoverable error for result h0009A_1_60234_7 (CreateProcess() failed - )
25) Message boards : Number crunching : Why I like the 4.4x scheduler (Message 118332)
Posted 4 Jun 2005 by Profile Purple Rabbit
Post:
The new scheduler takes a little time to get used to. I used to have to manually manage an over commitment of resources. I didn't have to do this very often because I had tuned the cache, but occasionally a project would dump a bunch of WUs on me.

After I got the new scheduler stable (adjust time to connect, etc.) everything worked just fine. I think a lot of the hollering is really about change and not functionality (modem guys/guyettes excepted).

The bugs are still present, but the idea is right.

Now, who has an optimized V4.43 Linux BOINC CC? I'm stuck at 4.30. I'm too new to Linux to do it myself, but I'll figure it out if I can't leech from someone else :-)
26) Message boards : Number crunching : "Legal' optimized client for Windows (Message 118302)
Posted 3 Jun 2005 by Profile Purple Rabbit
Post:

The difference is Tetsuji's great work. Kind of a gas isn't it.


It sure is! Thanks to Tetsuji for the improvement.
27) Message boards : Number crunching : "Legal' optimized client for Windows (Message 118281)
Posted 3 Jun 2005 by Profile Purple Rabbit
Post:
I'm really impressed with the difference in processing time between these optimized clients and the standard one.

My sample size is small for the optimized client (2 days), but I checked the other crunching hosts to make sure I didn't have a "short" unit. My numbers are eyeballed from a list, but should be close enough for my point. I've received credit for each host so the optimized clients seem to be working.

Performance (1 SETI WU):

866MHz P3
old: about 6.2 hours new: about 5.5 hours

1.3GHz P3 Celeron
old: about 6.5 hours new: about 5.5 hours

3.4GHz P4 HT
old: about 3.0 hours new: about 1.25 hours

One of my hosts scares me tho :-) On the two slower computers the gain is about 1.2x. The 3.4GHz computer shows a 2.4x gain. WOW! It has done 11 WU consistently at this rate. I don't think it's an HT thing since my numbers are per WU from BOINCView.

I'm just providing my observations on my early experience with the optimized client. However, I am curious if the P4 HT's gain is typical. Could this difference be because of SSE3?
28) Message boards : Number crunching : Report your ~6 month old dead WU here (Message 118248)
Posted 3 Jun 2005 by Profile Purple Rabbit
Post:
June 28, 2004: http://setiweb.ssl.berkeley.edu/workunit.php?wuid=167120

I thought I "won", but Ageless beat me to it :-)

It's the only one I have so I can't complain.
29) Message boards : Number crunching : POLL-New 4.4X scheduler-Rate it yourself (Message 118208)
Posted 3 Jun 2005 by Profile Purple Rabbit
Post:
Broadband 2
30) Message boards : Number crunching : 4.43 Problems (Message 114187)
Posted 23 May 2005 by Profile Purple Rabbit
Post:
I have just submitted a fix that may fix these problems.


Thanks John. That will help a lot. I won't upgrade any more of my computers until the release becomes more stable (I've only done one).

Hopefully this will all work out.

Rick
31) Message boards : Number crunching : 4.43 Problems (Message 114177)
Posted 23 May 2005 by Profile Purple Rabbit
Post:
I haven't seen the problems previously noted. However, I did get an additional CPDN WU when I upgraded one of my computers. This should not have happened, but no big deal. At least this was on one of my faster machines.

I'm not sure if 4.43 is ready for prime time yet.

32) Message boards : Number crunching : 4.25 has been replaced by 4.43 as the "recommended" Core Client. (Message 113692)
Posted 22 May 2005 by Profile Purple Rabbit
Post:
I think I've started something bigger than I expected.

My original question related to why the BOINC team would revert back to a version earlier than what they had officially released. Ageless pointed out a reason and I agree. I think both 4.19 and 4.25 should be publicly available without recourse to "obscure" server links. Both versions have their advantages.

I'm sorry I asked :-) but thanks everyone.
33) Message boards : Number crunching : 4.25 has been replaced by 4.43 as the "recommended" Core Client. (Message 113672)
Posted 22 May 2005 by Profile Purple Rabbit
Post:
That's why you save every optional version to your harddrive, or ask team members who have done so through email for such a version, Pink Hare. ;)


But I didn't save them, silly me :-)I also thought this might be a wider issue than just our team. I certainly would ask the team if I needed help.

Bottom Line: I think BOINC version 4.25 should be officially available.

PINK? You insult me :-) I'm a royal purple rodent!
34) Message boards : Number crunching : 4.25 has been replaced by 4.43 as the "recommended" Core Client. (Message 113666)
Posted 22 May 2005 by Profile Purple Rabbit
Post:
Because 4.19 can still show all graphics for all projects that have them, where as 4.25 - 4.4x have problems with ATI cards.


I forgot about that. I don't have any ATI cards and I don't look at the graphics very much. I think the idea of two "old" versions might fit here. The backward compatibility of 4.19 and the features of 4.25.

I'm still not sold on 4.43 so having access to "old" golden versions would be helpful for me.
35) Message boards : Number crunching : 4.25 has been replaced by 4.43 as the "recommended" Core Client. (Message 113647)
Posted 22 May 2005 by Profile Purple Rabbit
Post:
I'm curious why the "old" version isn't 4.25. Perhaps this is a website edit error (although they seem quite specific about this)?

I've been using 4.25 since they declared it golden with no problems. I've held off using the later versions because they had some issues (development versions with no immediate benefit to me). I think 4.25 should still be made available to those who want it no matter what the official "old" version is.

IMHO the features of 4.25 seem to be much better than 4.19. Why would they backtrack on the old version number?
36) Message boards : Number crunching : so... what's your slowest machine??? (Message 91983)
Posted 28 Mar 2005 by Profile Purple Rabbit
Post:
I ran my 486-DX66 on SETI/BOINC for 1 WU. It's been "upgraded" with an AMD 586 chip (Pentium 58 equivalent) and 96 MB RAM. It took 12 days to complete. It was an interesting experiment.

For lack of finding anything else for it to do, I put it back on SETI/Classic where it does (you guessed it) about 1 WU in 12 days.

I know this falls WAY below your idea of a slow cruncher, but it is a cruncher and it can work if you're really patient. It's really not worth it tho. I just can't let go of a working machine :-)
37) Message boards : Number crunching : Uploads failing (Message 87659)
Posted 19 Mar 2005 by Profile Purple Rabbit
Post:
Same problem here.
38) Message boards : Number crunching : Deferring constantly. Server prob? (Message 79520)
Posted 15 Feb 2005 by Profile Purple Rabbit
Post:
I wonder if there might also be a Berkeley bandwidth issue along with the server load. The app and the pdb files are about 5MB. Mulitiply that by everyone who wants them NOW and the weakest link will bottleneck data transfers.

Of course I also wonder what is different with the new version :-)
39) Message boards : Number crunching : Deferring constantly. Server prob? (Message 79490)
Posted 15 Feb 2005 by Profile Purple Rabbit
Post:
I just got a new SETI Client version 4.09 at 0100 UTC. Maybe there is a large server load as it hands out the new client software?

I am also seeing deferred uploads.
40) Message boards : Number crunching : Alpha Client Release---Inverse POLL (Message 74420)
Posted 27 Jan 2005 by Profile Purple Rabbit
Post:
I also know the risks. I "accidentally" updated to 4.14 during the brief period it was shown as released. I quickly reverted to 4.13 when I found it wasn't really released. My paranoia was confirmed with the later versions. My CPU time investment in my other projects makes this gamble not acceptable.

It's a matter of what you are testing. Alpha and Beta BOINC code testers are needed and wanted. They should be out of the mainstream crunchers. For those who test the code, great. For the rest of us, crunch on!

I'm involved with a beta site that is just coming up. I don't think injecting unproven BOINC code is a good test for the site.

I just upgraded to 4.18 and then a few minutes later to 4.19 :-) These were released versions. 4.19 has a proxy access problem, but my system doesn't care about that. I've had no problems with 4.19.

Unless there is a problem, "don't fix what ain't broke". This doesn't apply to the testers. If you are really a tester, you know who you are.


Previous 20 · Next 20


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.