Are you working on this with any of the linux optimizers?

My source code was always public, so of course, everyone who wants to compile it for other platforms, can do it. Most of my modifications are platform independent. Only the CPU affinity and process priority may need modifications for Unix systems. Credit calibration is platform independent. Myself, I plan to compile some clients and applications for FreeBSD, but currently I do not find time enough. Hopefully later.

---

trux
BOINC software
Freediving Team
Czech Republic

...Most of my modifications are platform independent. Only the CPU affinity and process priority may need modifications for Unix systems...

I don't know about the other *nix systems but the Linux kernel has had processor affinity for processes available for some time now.

Regards,
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

I don't know about the other *nix systems but the Linux kernel has had processor affinity for processes available for some time now.

I do not think than anyone doubts that :) The problem is that different approach needs to be used under different OS or hardware platforms.

---

trux
BOINC software
Freediving Team
Czech Republic

Are you working on this with any of the linux optimizers?

My source code was always public, so of course, everyone who wants to compile it for other platforms, can do it. Most of my modifications are platform independent. Only the CPU affinity and process priority may need modifications for Unix systems. Credit calibration is platform independent. Myself, I plan to compile some clients and applications for FreeBSD, but currently I do not find time enough. Hopefully later.

I'll hope for a timely and successful completion of your tests and release of your final modified version. I compile my own slightly optimized version of BOINC and look forward to implementing and testing these changes locally.

---

Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws.
Douglas Adams (1952 - 2001)

I'll hope for a timely and successful completion of your tests and release of your final modified version. I compile my own slightly optimized version of BOINC and look forward to implementing and testing these changes locally.

Thanks :) Mr. Pernod (I hope he does not mind mentioning him here) found already a small bug in configuration in case of using venues, but otherwise I have no negative reports. I've gotten some positive reports, but more of them are missing though too, so it is hard to tell how far I am. I am just fixing the venue issue and will keep you updated.

There is still room for another beta tester or two, so if you are interested, please register in our forum, and send me a Private Message request there.

---

trux
BOINC software
Freediving Team
Czech Republic

I don't mind being mentioned ;)

Well, so far the results do not look too bad, and I believe it works well as expected. Some minor bugs in the credit calibration and in the priority_project feature were fixed thanks to the beta testers, but I need more feedback and more testers. Unfortunately, I have ony 2-3 active users who give any feedback (although there were more people downloading the client), so I do not feel yet quite comfortable with releasing it publicly.

If you are interested in testing the credit calibration client and want to help me testing it, please register in the CNT forum and send me a Private Message there. Should you have troubles with it, please post here.

---

trux
BOINC software
Freediving Team
Czech Republic

Well, so far the results do not look too bad, and I believe it works well as expected.

I know you're concentrating on the new client, but thought I'd pass this on in case it's of any use to you. I just completed a 5.02 work unit on a P3/1000 (256M, Win2K) using your 5.3.6.tx12 (SSE) client, and it claimed 190.71 credits for 284515.91 seconds of work, which looks pretty close to "standard". Result information is HERE.

As I mentioned, S@H Enhanced uses different crediting system - it counts operations instead of time & benchmarks. Please read further below the discussion about S@H Enhanced.

---

trux
BOINC software
Freediving Team
Czech Republic

Some longer testing would be better, but so far it looks fine, some glitches were fixed, and especially I'd like that the client can serve its purpose before the S@H Enhanced comes. With S@H Enhanced, the calibration will not be needed so urgently (at least not initially - as we discussed below). Still, it will be fairer to calibrate credits at other projects, to avoid over-claiming there. Currently, on most computers, it is only S@H that needs the positive calibration.

Well, I am making the client public now. Besides the discussed credit calibration, it has a bunch of new features, that you may find useful even if you decide not to use the calibration:

• configurable RPC port
  
• network masks / IP blocks in remote_hosts.cfg
  
• return_results_immediately
  
• configurable CPU affinity
  
• configurable process priority
  
• automatic credit self-calibration
  
• priority projects - backup projects will be started only when there is no work for the priority project(s)
  
• automatic deleting of uncompleted overdue WU's
  
• manual resetting of all long term and short term debts
  

You can download the last modified BOINC core client release 5.3.11.tx31 here:

http://boinc.truxoft.com/core-cal.htm

Hope it works well for all. In case of troubles, please be sure to have read the instructions, and also verify messages in BOINC manager - they contain a lot of additional information that can help tracking down the problems.

Please read the description before installing, and please follow the installation and configuration instructions.

The source code of the modified files is also available to download, so if you are a Volunteer Developer and think some of the features may be interesting for the dev team, feel free to submit it to them. I am not familiar with their diff / CVS system, so would be thankful if you can do it for me. I tried to make all changes as clean as I could. They are all marked with // TX and well commented.

Special thanks to all beta testers, and to members of our team for the help

---

trux
BOINC software
Freediving Team
Czech Republic

Just installed your version on this box, after upgrading from 4.45, and if it pans out OK, it will go on the other one too.

I saw you mentuioned an option to set project(s) as a backup,

priority projects - backup projects will be started only when there is no work for the priority project(s)

So it will only get work if the main project is dry, but I can't find the relevent setting to achieve this.


Also, When I suspend network activities, they seem to turn back on, on their own, almost instantly...




Edited for link.

---

Foamy is "Lord and Master".
(Oh, + some Classic WUs too.)

look in your BOINC directory for a file called truxoft_prefs.xml (also included in the download) and add the line <priority_projects>SETI@Home</priority_projects> to make SETI the main project.
For more information see the section Sample Configuration on the download page.

look in your BOINC directory for a file called truxoft_prefs.xml (also included in the download) and add the line <priority_projects>SETI@Home</priority_projects> to make SETI the main project.
For more information see the section Sample Configuration on the download page.

Thanks, will do.

Re the network thing, very odd, but Boinc just popped up a message, "16/01/2006 18:43:11||Suspending network activity - user request"...Odd...I turn it off, it turns back on, then a few mins later turns it'self back off..Odd.. :-/

---

Foamy is "Lord and Master".
(Oh, + some Classic WUs too.)

Also, When I suspend network activities, they seem to turn back on, on their own, almost instantly...

I upgraded to this Boinc level a while back and the stock client has this problem. I returned my boxes to the recomended Boinc version for this reason.

---

Boinc....Boinc....Boinc....Boinc....

Yes, there were some changes in the later official BOINC releases in the network access, and I did not really study what their purpose was, but it is true that people report these problems since around the version 5.3.8 (?), maybe even 5.3.6. I saw there was another change in the official 5.3.11, and thought it was a fix. It is a pity nobody mentioned it during the beta test - although it is apparently problem in the official source, I could look at it and fix it. Well, I'll try looking at it anyway and let you know when there is a new version.

---

trux
BOINC software
Freediving Team
Czech Republic

Running trux's client with no problems to report. I prefer the seperate prefs.xml idea, much better to keep me out of things i can break =).

The priority projects option is a good enough reason for the client on its own let alone the calibration system.

---

I gave the new client a try and got this:

1/16/2006 8:26:51 PM||truXoft add-on: calibrate_credit = SETI@home
1/16/2006 8:26:51 PM|seti@home|Resuming computation for result 03se03aa.7540.6624.634652.1.202_0 using setiathome version 411
1/16/2006 8:26:51 PM||using earliest-deadline-first scheduling because computer is overcommitted.
1/16/2006 8:26:51 PM|seti@home|CC calibration failed: credit_correction_factor checksum failed!

What is going on in the last line?

---

This message can come initially, when the coefficients are not established yet. Otherways it also happens when you try to manually mess with the correction factor. Or when the status file is incomplete or corrupted. Anyway, unless you get it at each completed WU, it is rahter unimportant, since the calibrating simply restarts again, in such case. It is a protection against corrupted data, that could result in unreasonable results, or against cheaters trying to influence the results manually.

Even if the claibration fails, nothing bad happens, and you still claim rather fair credit (the default uncalibrated benchmarks are still relatively high).

---

trux
BOINC software
Freediving Team
Czech Republic

Im getting several of these:

2006-01-17 05:19:19 [seti@home] Started upload of 05oc03ab.24335.3042.379824.1.86_3_0
2006-01-17 05:19:26 [seti@home] Finished upload of 05oc03ab.24335.3042.379824.1.86_3_0
2006-01-17 05:19:26 [seti@home] Throughput 4830 bytes/sec
2006-01-17 06:38:52 [---] Rescheduling CPU: application exited
2006-01-17 06:38:52 [seti@home] Computation for result 18se04ab.27430.1170.811074.1.9_0 finished
2006-01-17 06:38:52 [seti@home] Starting result 05oc03ab.24335.3042.379824.1.75_3 using setiathome version 411
2006-01-17 06:38:52 [seti@home] CC calibration: blocked [negative calibration limit] 17.64 >> 8.15 (time: 4753s >> 2132s / Gfpops: 1.29 >> 1.48)

then the last one was like this:

2006-01-17 09:21:16 [seti@home] Computation for result 05oc03ab.24335.3042.379824.1.92_1 finished
2006-01-17 09:21:16 [seti@home] Starting result 13dc03aa.20895.13312.253418.1.192_4 using setiathome version 411
2006-01-17 09:21:16 [seti@home] CC calibration: 17.47 >> 26.80 (time: 4706s >> 4610s / Gfpops: 1.29 >> 4.92)

What do this mean?

Thanks
Roberto

---

Im getting several of these:

2006-01-17 06:38:52 [seti@home] CC calibration: blocked [negative calibration limit] 17.64 >> 8.15 (time: 4753s >> 2132s / Gfpops: 1.29 >> 1.48)

then the last one was like this:

2006-01-17 09:21:16 [seti@home] CC calibration: 17.47 >> 26.80 (time: 4706s >> 4610s / Gfpops: 1.29 >> 4.92)

What do this mean?
Thanks
Roberto

It means that the aplication needed some wu's tot calibrate as said by Trux.
Mine looks like:


16/01/2006 22:26:27|seti@home|CC calibration: blocked [negative calibration limit] 16.05 >> 11.79 (time: 2868s >>

1849s / Gfpops: 2.20 >> 3.55)
16/01/2006 23:42:09|seti@home|CC calibration: blocked [negative calibration limit] 16.06 >> 11.80 (time: 2870s >>

1850s / Gfpops: 2.20 >> 3.55)
17/01/2006 00:34:52|seti@home|CC calibration: blocked [negative calibration limit] 15.93 >> 12.97 (time: 2847s >>

1966s / Gfpops: 2.20 >> 3.94)
17/01/2006 01:06:12|seti@home|CC calibration: blocked [negative calibration limit] 15.66 >> 14.04 (time: 2800s >>

2056s / Gfpops: 2.20 >> 4.33)
17/01/2006 01:53:24|seti@home|CC calibration: blocked [negative calibration limit] 15.80 >> 15.47 (time: 2824s >>

2191s / Gfpops: 2.20 >> 4.73)
17/01/2006 02:40:26|seti@home|CC calibration: 15.74 >> 16.72 (time: 2814s >> 2292s / Gfpops: 2.20 >> 5.13)
17/01/2006 08:33:07|seti@home|CC calibration: 15.64 >> 17.89 (time: 2795s >> 2378s / Gfpops: 2.20 >> 5.53)
17/01/2006 09:21:35|seti@home|CC calibration: 15.93 >> 19.50 (time: 2847s >> 2517s / Gfpops: 2.20 >> 5.92)
17/01/2006 10:16:45|seti@home|CC calibration: 15.91 >> 20.69 (time: 2843s >> 2599s / Gfpops: 2.20 >> 6.29)
17/01/2006 11:16:30|seti@home|CC calibration: 16.21 >> 22.26 (time: 2897s >> 2726s / Gfpops: 2.20 >> 6.64)
17/01/2006 12:11:14|seti@home|CC calibration: 16.21 >> 23.37 (time: 2898s >> 2797s / Gfpops: 2.20 >> 6.97)
17/01/2006 13:03:28|seti@home|CC calibration: 16.21 >> 24.38 (time: 2897s >> 2858s / Gfpops: 2.20 >> 7.27)

From start till now. The figure between >> and ( is getting bigger until fully calibrated I think.

The actual claimed credits apear some more than 1 higher as the figure indicates, with my computer that is. Exept for where te cc calibration is blocked because callibrated figure was lower than initially claimed by the aplication.