Note: Not Requesting new work or reporting results.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5845
Credit: 109946460069
RAC: 31481698

RE: Thanks. I never

Message 19175 in response to message 19173

Quote:
Thanks. I never interpreted anything you said as critical, only helpful..

OK, that's great!!

Quote:
.... So how do I do it manually? I see the settings in the client_state.xml file for both SETI and EINSTEIN but an attempt to change to "0" is ineffective - the old values pop right back up. So what do I do?

There is both a client_state.xml and a client_state_prev.xml. Here is what I have done with older versions of BOINC which has had precisely the desired effect. I think it might solve your problem:-

  • * Shut down BOINC and confirm through task manager that nothing is running.
    * Carefully edit with a text editor like notepad

both the above named .xml files.
* Make sure you find all LTDs (both pos and neg) and edit them to read 0.000000
* Save both files. I'd save the prev file first then the normal file just in case something checks dates as part of some sort of integrity check.
* Restart BOINC.

As a matter of interest, how do you fire up BOINC? If you double click BOINC Manager, it will start and it will also start the daemon, boinc.exe. If you shut down BOINC Manager under these circumstances, even with the red X, the daemon will be stopped as well. However, if a daemon is already running and you invoke BOINC Manager, the Manager will notice and will not try to start another daemon. Then if you shut down the Manager, it will NOT shut down the daemon, simply because it didn't start the daemon in the first place. So, as Bill points out, you do have to be careful to check that nothing is still running even after you have stopped the Manager.

Quote:
... I don't know if I did any good - the SETI @HOME app stills says 4.18 in BOINC. Whether it is 4.18 for real, or the 4.11 in the YAOSCW-N-r8.1.zip file, I don't know. It took 2 hours and 11 minutes for the last SETI@home to run to completion, if this is any indication.

Congratulations!! You've done really well. The tricks are in the app_info.xml file that was installed along with the optimised seti executable. Basically this .xml file tells BOINC that it is OK to run 4.11 even although the one that is supposed to run is 4.18.

You are indeed running the optimised Seti app. Go to your account page and view your computers and bring up the results list and you will see that your previous 16,000 seconds times are now around 7,000 seconds and you have already had results validated. Enjoy your new found speed :).

There is one more thing you can do to help those who crunch the same WUs as you. Notice how your claim is now quite low? That is because the science app is now much faster than the benchmarks say it should be. You need an optimised BOINC client to get you back to claiming the correct amount. (Do you notice on the new results how your fellow crunchers are dragging your low claim up?)

The optimised BOINC client from here will do the job. All you do is download the one for SSE2 and do the following:

  • * Completely stop BOINC
    * Delete boinc.exe from your BOINC folder - perhaps save it elsewhere
    * Insert the new optimised boinc.exe
    * Restart BOINC

BOINC will notice the version change and will redo the benchmarks so don't have anything else running at that time. You should get a nice increase in the total of both scores compared to the old total. It's the total that's important, not so much each individual one.

Cheers,
Gary.

Fred Plett
Fred Plett
Joined: 1 Sep 05
Posts: 4
Credit: 37104849
RAC: 12459

Thanks. I successfully (and

Thanks. I successfully (and manually) reset the long term debt. I successfully upgraded BOINC as you suggested. I can see from the messages that the benchmarks were indeed reset since I am now running Version 5.3.1 of BOINC (I think the last one posted was 5.2.7 so I must have a jump on everybody else now). I don't know how the LTD for Einstein got so out of kilter, but it is now zero.

Stick
Stick
Joined: 24 Feb 05
Posts: 790
Credit: 31200434
RAC: 262

RE: Thanks. I successfully

Message 19177 in response to message 19176

Quote:
Thanks. I successfully (and manually) reset the long term debt. I successfully upgraded BOINC as you suggested. I can see from the messages that the benchmarks were indeed reset since I am now running Version 5.3.1 of BOINC (I think the last one posted was 5.2.7 so I must have a jump on everybody else now). I don't know how the LTD for Einstein got so out of kilter, but it is now zero.

Fred,

Glad to hear you got things straightened out. I hadn't seen anybody with a problem like yours before so I was interested in reading the advice you got from Gary and Bill. As to your BOINC v5.3.1, I am also using it (I installed it a couple of days ago). However, it is not newer than v5.2.7. It is really an optimized v5.2.6. If you go to the BOINC Manager "Help" menu and select "About BOINC Manager", you'll see that is says v5.2.6. (I don't know why they chose to call it 5.3.1 - seems very confusing.) But, my guess is they will be offering an optimized 5.2.7 version soon, so stay tuned.

Stick

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5845
Credit: 109946470069
RAC: 31482632

RE: Thanks. I successfully

Message 19178 in response to message 19176

Quote:
Thanks. I successfully (and manually) reset the long term debt. I successfully upgraded BOINC as you suggested. I can see from the messages that the benchmarks were indeed reset since I am now running Version 5.3.1 of BOINC (I think the last one posted was 5.2.7 so I must have a jump on everybody else now). I don't know how the LTD for Einstein got so out of kilter, but it is now zero.

Good to hear that it's all sorted. I've just had a look at your results list over at Seti and all seems well. You are still doing results in around 7,400 seconds but your credit claim has now been raised to a more appropriate value. You will now not be dragging anyone else down. I also see you have new work at EAH as well. It must feel good to have everything running smoothly at last :).

All you have left to do now (over at Seti) is merge the two older CPUIDs (1140448 and 1299453) so that all your stats are back together under your one actual ID (1350264).

Good luck!!

Cheers,
Gary.

Tern
Tern
Joined: 27 Jul 05
Posts: 309
Credit: 93405878
RAC: 137

RE: As to your BOINC

Message 19179 in response to message 19177

Quote:
As to your BOINC v5.3.1, I am also using it (I installed it a couple of days ago). However, it is not newer than v5.2.7. It is really an optimized v5.2.6. If you go to the BOINC Manager "Help" menu and select "About BOINC Manager", you'll see that is says v5.2.6. (I don't know why they chose to call it 5.3.1 - seems very confusing.)

The V5.3.1 optimized client is actually an optimized "daemon" (boinc.exe) - and not the Manager itself (boincmgr.exe), which remains unchanged at whatever version you had when you put the optimized client in there... So you can be running the V5.3.1 optimized daemon under BOINC Manager V5.2.4 or V5.2.6 or V5.2.7...

There is no reason to offer another optimized client because of a change that only affects the manager version. That only needs to happen if the change affects the daemon. So, I doubt there will be another optimized version, at least not right now, as 5.3.1 daemon works fine with 5.2.7 manager.

As for the confusing numbers; anything with an ODD middle digit comes from the "test" software streams. So when Trux went to create an optimized client, he got the "latest" version of the software at the time, which was 5.3.1, which could be older than 5.2.6 or newer. I _think_ it dates to about 5.2.4. I guarantee it's older than 5.2.7 because 5.2.7 just came out. All 5.3 versions are intended eventually to become the first 'release' 5.4 version - although bug fixes found and fixed in 5.3 versions may wind up "back" in 5.2 releases, not every change in a 5.3 will be put in ANY 5.2. Bottom line is, if you want to run only "released" software, never run anything with an odd middle digit. If you are willing to run "beta" code, or optimized versions of the code, you're on your own as far as keeping it up to date, and the version numbers don't help much. Optimizers have to number their code _something_, and there isn't an official number available for them to use - so 'stealing' the test-stream number it's based on is as good as anything.

Stick
Stick
Joined: 24 Feb 05
Posts: 790
Credit: 31200434
RAC: 262

Bill, Thank you for the

Message 19180 in response to message 19179

Bill,

Thank you for the explanation! I understand a lot better now (at least I think so).

Stick

Quote:
Quote:
As to your BOINC v5.3.1, I am also using it (I installed it a couple of days ago). However, it is not newer than v5.2.7. It is really an optimized v5.2.6. If you go to the BOINC Manager "Help" menu and select "About BOINC Manager", you'll see that is says v5.2.6. (I don't know why they chose to call it 5.3.1 - seems very confusing.)

The V5.3.1 optimized client is actually an optimized "daemon" (boinc.exe) - and not the Manager itself (boincmgr.exe), which remains unchanged at whatever version you had when you put the optimized client in there... So you can be running the V5.3.1 optimized daemon under BOINC Manager V5.2.4 or V5.2.6 or V5.2.7...

There is no reason to offer another optimized client because of a change that only affects the manager version. That only needs to happen if the change affects the daemon. So, I doubt there will be another optimized version, at least not right now, as 5.3.1 daemon works fine with 5.2.7 manager.

As for the confusing numbers; anything with an ODD middle digit comes from the "test" software streams. So when Trux went to create an optimized client, he got the "latest" version of the software at the time, which was 5.3.1, which could be older than 5.2.6 or newer. I _think_ it dates to about 5.2.4. I guarantee it's older than 5.2.7 because 5.2.7 just came out. All 5.3 versions are intended eventually to become the first 'release' 5.4 version - although bug fixes found and fixed in 5.3 versions may wind up "back" in 5.2 releases, not every change in a 5.3 will be put in ANY 5.2. Bottom line is, if you want to run only "released" software, never run anything with an odd middle digit. If you are willing to run "beta" code, or optimized versions of the code, you're on your own as far as keeping it up to date, and the version numbers don't help much. Optimizers have to number their code _something_, and there isn't an official number available for them to use - so 'stealing' the test-stream number it's based on is as good as anything.


Tern
Tern
Joined: 27 Jul 05
Posts: 309
Credit: 93405878
RAC: 137

RE: Thank you for the

Message 19181 in response to message 19180

Quote:
Thank you for the explanation! I understand a lot better now (at least I think so).

I checked the dates after posting - the optimized V5.3.1 is dated 10/23 and V5.2.6 is dated 10/31. I don't have V5.2.4 to double-check, but this leads me to believe that 5.3.1 is going to most closely match 5.2.4...

I'm about to remove 5.3.1, as I'm taking SETI off my Windows box (leaving it on the Macs) and the whole purpose for it is to better "match" the optimized SETI app on credit claims. I think it's over-claiming for my other projects (especially Rosetta). I'll leave an optimized BOINC client on the two G4 Macs that are "mostly" doing SETI and using the Altivec-optimized app, and Einstein (also Altivec-optimized); unfortunately, that still means V4.72 as that's the latest optimized client for Mac. Isn't all this fun? :-/

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.