The Official BOINC news thread

The latest on BOINC:

The tentative schedule for releasing 2.28 is Monday. One of the developers said "Following that roll-out I�ll be dropping out of sight for two weeks while attempting to code up a storm and fix my M1 bugs." So BOINC won't release in March. After that the next round of releases will begin for the Alpha project driving toward a stable release for the Beta project. The next release, probably v2.29, will release in early April (at the earliest). This info changes the release date for BOINC final and Seti@home (on BOINC). I'm now estimating a Mid-April release at the earliest.

Here's an update on BOINC:

On Monday, BOINC v2.28 and SAH v2.28 will release for all platforms. At this time, one developer said they will fix all their bugs targetted for M1 in about 2 weeks. Not sure if all developers plan to do this, but if they are, that's great. The next release for Alpha will happen sometime after this. It'll take some time to alpha test this, fix more bugs possibly, release to beta sometime after. I now estimate a May release.

Some changes were made to the scheduler software running for the alpha test. If the changes pass, they will be included with the upcoming beta release. This may solve some credit problems.

Some fixes since last checkin update:

- Various changes to the way CPU benchmarking is done,
motivated by the need to make claimed credit constant across CPUs,
including "hyperthreaded" X86 CPUs
Old:
A single process runs the benchmark.
This gets 100% of the FPU on a HT CPU.
When 2 copies of an app are run,
they will each get 50% of the FPU,
so the claimed credit will be too high by a factor of 2.
New:
Run N copies of the benchmark in parallel,
where N is the number of CPUs reported by the OS
(e.g. twice the number of HT CPUs)
Average the results.
Revised interface:
start_cpu_benchmarks() starts up the N benchmark processes
cpu_benchmarks_poll() checks for completion
are_cpu_benchmarks_running() returns true iff running (doesn't poll)
abort_cpu_benchmarks() aborts processes
The set of running processes is maintained in an
array of BENCHMARK_DESC structures
Results are passed back in files in the Unix case,
where we use separate processes.
In the Windows case (where we use threads)
results are passed back in the BENCHMARK_DESC structure.

- check for result of status request parse. Should avoid problem of partial communication from client to server.

Other stuff:

I'll probably post a message here before everyone gets that email that tells them about the switch to BOINC.

Here are some useful BOINC links and their decriptions:

BOINC website-boinc.ssl.berkeley.edu...
BOINC Alpha test website-setiboinc.ssl.berkeley.edu...
BOINC Beta test website-setiboinc.ssl.berkeley.edu...
BOINC Taskbase-setiathome.ssl.berkeley.edu...
M1 blockers (release of BOINC)-
setiathome.ssl.berkeley.edu...
dme=&bycategory=*&byowner=*&bypriority=*&bystatus=*&bymilestone=M1+-+Public+Launch&find=find
M2 blockers (decommission of Seti@home)-
setiathome.ssl.berkeley.edu...
ndme=&bycategory=*&byowner=*&bypriority=*&bystatus=*&bymilestone=M2+-+Decommission+Old+SETI%40home&find=find
M3 blockers (other BOINC projects)-
setiathome.ssl.berkeley.edu...
ndme=&bycategory=*&byowner=*&bypriority=*&bystatus=*&bymilestone=M3+-+Other+BOINC+Projects&find=find
All bugs-
setiathome.ssl.berkeley.edu...*&byowner=*&bypriority=*&bystatus=*& bymilestone=*&find=find

These blockers include BOINC client, BOINC server, and BOINC web bugs.


[Edited on 21-3-2004 by xn]

More news on BOINC:

The roll-out plans is going to happen over the course of several weeks, if not months. During that time both the beta and alpha projects will still continue to operate.

BOINC v2.28 and SAH v2.28 have been released to the beta test. The minimum core client version has been set to 2.28, you'll need to upgrade to the latest client to receive new work-units.

SAH v2.29 (for Windows) has been released to the alpha test.

SAH v2.29 for Windows has been released (to beta). If you suffer from a corrupted glut32.dll you'll need to reset the project in order to pick up the uncorrupted version. Please note that the minimum core client version has been set to v2.28, you'll need to upgrade to the latest client to receive new workunits.

Thanx, xn, for all these posts!!

Sorry to interupt the thread but wanted you to know that we're following them my friend!!

Just joined BOINC guys. Hope to see you all there soon.

SAH v2.30 for Windows has been released to the Alpha test. SAH v2.29 for Solaris/Linux has also been released to the Alpha test.

New stuff in SAH v2.30:

This release is mainly to fix glut errors. Glut will no longer be an issue. This release should also speed up the amount of time required to display the graphics window.

- Unify all diagnostics functionality under the same banner, SETI_BOINC, BOINC_CLI, BOINC_GUI, upper_case now all initialize the BOINC Diagnostics Library which sets up stdout and stderr redirection, heap corruption detection, stackwalker for Windows, and stackalker for *nix can be plugged in and be automatically used by all applications.
- Taskbase#1131 -- closed
- Taskbase#1132 -- closed

- Fixed nightly tarball generation so that autoconf,configure,make will result in a successful build.
- SETI_BOINC Glut Free, Basically copied font code from glut and put it into the glut folder.
- Removed Glut libraries
- Removed Delay Load technologies
- Fix up project file so it'll build on a clean platform.

[Edited on 26-3-2004 by xn]

SAH v2.31 for Windows has been released to Alpha. SAH v2.30 for Linux/Solaris has also been released to Alpha.

Changes in v2.31-
- Release tags. Sources were tagged as seti_boinc_release_2_30 and then, after some build problems were fixed, were tagged as seti_boinc_release_2_30a.
The build from this latter tag was what was released as app version 2.30.
- Removed parseOutfile from the project.
- Fix a minor bug where after the screen saver shutdown BOINC would fire-up a new SAH Window
- Tag for a 2.31 release.

More BOINC news-
When v2.28 released for both the core clients and project applications, during the transfer process between the Alpha and Beta projects, the glut32.dll became corrupted somehow. This was fixed with the release of a new version of SAH to the Alpha project. There were no problems during the Alpha test, so it was sent to the Beta test. It turned out to have little effect, as Windows still didn't like the old glut32.dll file and refused to load it, which caused the 128 error. It was decided to dump glut32.dll as soon as possible. It was only being used to draw fonts on an OpenGL surface. Now the font technology has been incorporated into the SAH application. v2.30 has been released to the Alpha test, where a problem was discovered and fixed in v2.31, which is currently in Alpha testing. The failure rate because of this bug is at 53%. That's about 5300 failures per night. Prior to v2.28, the failure rates were between 10-12%. It was expected for the numbers to drop to sub 9% with the v2.28 release.

[Edited on 26-3-2004 by xn]

I finally got my first credits!

I was beginning to wonder if I would ever get anything done. I currently have 17.01 total credit with a 1.62 average credit. Does anyone know what these numbers mean?

boinc.ssl.berkeley.edu... This page talks about the credit system. The reason it took so long was because 2 other computers in the world had to process the same workunit. Some people don't like this idea, but I think it's a good idea for a number of reasons. I believe it's done for security, so nobody can cheat. It might take a while before you get credit, but that's with about 5000 users, just imagine about 5 million users, it'll be so much quicker.

Thanks for the info xn. That definitely makes sense with the two computers.

First points are in: 76.34 total credit with 7.36 average credit with 110 credit pending. Woot!

Hi meddled!!

Way to go!!

Thanx, too, to xn (as always!), for the explanation (now fully understood and taken on board!!) that to "get *your* credit", you're actually having to "wait" for the "third" person to complete the same unit on their machine (or, rather, that the pace is determined - in effect - by the "speed" of the slowest machine to return the result). Makes no difference to *you*, though because you're already crunching new units while you're "waiting". And,as you say xn, when 5M people are running BOINC, then the net result will be that it's *your* machine's "speed" that will determine the outcome effectively.

I *like* it - thanx for the enlightenment!!

SAH v2.31 for Windows has been released to the beta test, which should fix all the 128 errors you have been seeing because of the corrupted glut32.dll file. There is a known issue in this release where when the screensaver shuts down that BOINC starts up the SAH graphics in windowed mode.

Would it be possible to either sticky this post or make a new sticky post for BOINC? I know xn is doing a wonderful job submitting news and bumping the post up, but it might be a good idea to just sticky it. By the way, i moved up to #10 on our team!

Heap checking has been turned on which should be able to tell if there's a buffer-overrun somewhere. After every 1024 memory allocations, it runs through the memory linked list looking for some section of the memory guard bytes where it has been overwritten with some sequence it isn't expecting. This adds some time to the completion of a workunit.

For those who used SetiSpy for Seti@home and wishes to use something similar to this for BOINC, there's now BoincSpy. BoincSpy 0.8 has been released! Download it from here: www3.telus.net... Please note that before using BoincSpy you MUST have Microsoft .NET Framework 1.1 installed. This program has been tested on Windows XP, Windows 2000, Windows 98SE, and Windows ME (win98 should work). The executable can be located anywhere on the system. Application doesn't create any files on the system except for the AutoUpdate process, which is only supported on WinXP, Win2k, and WinServer2k3. To view the BOINC user's page you must have logged in with the Remember account key on this computer option clicked. Known bugs: BoincSpy will not work if boinc_cli is running as a service in Windows 2000/2003. Only occurs if you run BoincSpy under terminal services. Win98/WinME don't report the efficiency field correctly.

These are BOINC/SAH bugs. There are also server and web bugs which are required for the release of BOINC.

# of bugs blocking the release: 33 BOINC core client bugs

# of bugs blocking the release: 13 BOINC server bugs

# of bugs blocking the release: 1 BOINC web bug

Total # of bugs blocking the release: 47 BOINC bugs

Originally posted by meddled
Would it be possible to either sticky this post or make a new sticky post for BOINC? I know xn is doing a wonderful job submitting news and bumping the post up, but it might be a good idea to just sticky it. By the way, i moved up to #10 on our team!

Well done on the "promotion" to 10th place meddled!!

As for making "sticky's" - well, I can do this of course - and will do so, with pleasure!! BOINC is increasingly important and will continue to be so: xn'x thread rightfully deservves to be the "Number 1" source of guidance for "new" users", such as I will be when I convert to running it!!

Cheers meddled!!

EDIT: Just to say I've made this thread a sticky...

[Edited on 29-3-2004 by Genya]

i've completed all the work loads on most of the machines i'm running... now it won't download anymore wu's... what's wrong with it... or what do i have to do to get going again? (they all have the latest v2.28)