Odd thing happening with times

Questions and Answers : Macintosh : Odd thing happening with times
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Mad Moggies
Avatar

Send message
Joined: 27 Mar 04
Posts: 92
Credit: 14,408
RAC: 0
United Kingdom
Message 150927 - Posted: 13 Aug 2005, 11:29:35 UTC
Last modified: 13 Aug 2005, 11:31:15 UTC

I'm using BOINC Manager version 4.43 with SETI@home version 4.18, on a 450mHz G4 Powermac with 768MB RAM and running Panther 10.3.9. I was recently running Einstein too but currently have it suspended as an Eistein WU takes ages on this Mac and I kept missing the shorter deadlines.

When a SETI WU is waiting to be processed, it gives what looks like a very long time as an estimate 'to completion'. As soon as the unit starts to be processed, the 'to completion' time drops to something incredibly short till 'Progress' reaches 25%. Then, the 'to completion' number increases till 'Progress' reaches 50%, when it starts decreasing as one would expect though not at the same rate as the 'CPU time' is increasing.

I find this very odd. The time a WU takes usually seems to be around half the 'to completion' time given at download.

Has anyone else seen this odd behaviour?
ID: 150927 · Report as offensive
Profile Tern
Volunteer tester
Avatar

Send message
Joined: 4 Dec 03
Posts: 1122
Credit: 13,376,822
RAC: 44
United States
Message 151721 - Posted: 15 Aug 2005, 5:27:48 UTC

I have not seen this behavior, but may be able to shed some light on it. Your computers are hidden, so I can't look at your configurations...

The estimate is first set by the server. The WIKI may give an explanation of how this is done, I haven't looked, as for me it's usually fairly close. This "very long time" for you may be because of your other computers?

Once it's being processed, the BOINC app looks at the CPU Time used and the Percent Complete and calculates a more accurate estimate - the further along it is, the more accurate the estimate gets, but on my Macs, it's usually pretty close by the time 1% is done. If this is dropping to "incredibly short", then something is wrong with the BOINC app. As much as I hate to see lost WUs, the best thing for you to do may be to toss the whole thing (especially the Library/Application Support/BOINC folder) and reinstall.

Re Einstein units taking too long - your G4 actually SHOULD be able to complete them just fine, as long as a) you run the Optimized app available from MacNN (which uses the Altivec unit in the G4, cutting the time approximately in half) AND b) you have your "connect to" set in the web preferences to something very low, like 0.1 days, to avoid downloading more units than you can get to before the deadline.
ID: 151721 · Report as offensive
Profile Mad Moggies
Avatar

Send message
Joined: 27 Mar 04
Posts: 92
Credit: 14,408
RAC: 0
United Kingdom
Message 151758 - Posted: 15 Aug 2005, 8:37:03 UTC - in response to Message 151721.  
Last modified: 15 Aug 2005, 8:44:09 UTC

I've only got the one computer, the Mac as described fully in my question!

I have 'connect to' set at 0.25 days as I didn't seem to be getting any work from Einstein when it was set at 0.1. I usually have one SETI unit running and one waiting. Ditto with Einstein if I let it but because of the one week deadline (and I don't have my Mac running 100% of the time) I now just let the one unit in at a time. I have SETI set for a 100 share and Einstein set at 200 because of the shorter deadline.

With Einstein, I'm running the beta version 0.12 now (the previous two beta versions kept hanging, which was the main problem in getting a WU processed in time). The time estimates with Einstein seem pretty accurate, so I don't think it can be a BOINC problem - rather a SETI one but I've downloaded BOINC version 4.44 from MacNN and will try that once the current WUs have finished.

ID: 151758 · Report as offensive
C

Send message
Joined: 3 Apr 99
Posts: 240
Credit: 7,716,977
RAC: 0
United States
Message 151992 - Posted: 15 Aug 2005, 22:18:18 UTC

Mad Moggies:
You can swap the new BOINC 4.44 in without harming any WU that are currently being calculated. Just QUIT boinc, then Control-Click on the BOINC Manager app. Click on "show package contents", then open the package and find the file "boinc". Remove it, and replace with the 4.44 version from Team ManNN. Rename the 4.44 version to "boinc" (without the quotes) so the computer thinks it's running the old one. Close up everything and restart BOINC Manager. It'll redo your benchmarks, but then resume running the WU.

C

Join Team MacNN
ID: 151992 · Report as offensive
Profile Shaktai
Volunteer tester
Avatar

Send message
Joined: 16 Jun 99
Posts: 211
Credit: 259,752
RAC: 0
United States
Message 152038 - Posted: 16 Aug 2005, 0:56:32 UTC

Well for what it is worth, barring any last day glitches, it looks like Einstein's "beta app for Mac" may become the mainstream very soon. That will drop crunch time for G4 and G5 computers in half on Einstein. It has some very agressive Altivec enhancements and is created by the Einstein team. You won't need to do anything except be attached to Einstein to get it.

http://members.dslextreme.com/~readerforum/forum_team/boinc.html

The team MacNN BOINC client will reduce the time estimates to a more realistic level. If you haven't tried it, you may also want to check out JavaLizard's optimized SETI app at the bottom of the same page. It will reduce the times for the SETI work units as well. The combination should let you run both projects without problems. Might need to increase your cache slightly to get Einstein work. Maybe .3 or .35. As long as you are running at least 40% Einstein on resource share you should be okay timewise. Keep watching the Einstein home page for an annoucement. But don't drop SETI either. There are some exciting possibilities being worked on for Macs running SETI by a couple of folks out Berkeley way.


Team MacNN - The best Macintosh team ever.
ID: 152038 · Report as offensive
Profile Mad Moggies
Avatar

Send message
Joined: 27 Mar 04
Posts: 92
Credit: 14,408
RAC: 0
United Kingdom
Message 152160 - Posted: 16 Aug 2005, 8:00:11 UTC - in response to Message 151992.  


You can swap the new BOINC 4.44 in without harming any WU that are currently being calculated.
C


Thanks, C, I'll do that then.
ID: 152160 · Report as offensive
Profile Mad Moggies
Avatar

Send message
Joined: 27 Mar 04
Posts: 92
Credit: 14,408
RAC: 0
United Kingdom
Message 152161 - Posted: 16 Aug 2005, 8:02:04 UTC - in response to Message 152038.  
Last modified: 16 Aug 2005, 8:07:34 UTC

Well for what it is worth, barring any last day glitches, it looks like Einstein's "beta app for Mac" may become the mainstream very soon. That will drop crunch time for G4 and G5 computers in half on Einstein. It has some very agressive Altivec enhancements and is created by the Einstein team. You won't need to do anything except be attached to Einstein to get it.

http://members.dslextreme.com/~readerforum/forum_team/boinc.html

The team MacNN BOINC client will reduce the time estimates to a more realistic level. If you haven't tried it, you may also want to check out JavaLizard's optimized SETI app at the bottom of the same page. It will reduce the times for the SETI work units as well. The combination should let you run both projects without problems. Might need to increase your cache slightly to get Einstein work. Maybe .3 or .35. As long as you are running at least 40% Einstein on resource share you should be okay timewise. Keep watching the Einstein home page for an annoucement. But don't drop SETI either. There are some exciting possibilities being worked on for Macs running SETI by a couple of folks out Berkeley way.



Thanks, Shaktai. I'm using Einstein 0.12 at the moment and think that's the version you are talking about.

As for the optimised version of SETI, I saw it but wasn't sure what the 7400/7410 and 7450 meant, so didn't download it and install it in case it messed things up, this Mac being a 5 year old G4!
ID: 152161 · Report as offensive

Questions and Answers : Macintosh : Odd thing happening with times


 
©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.