News archive

25 March 2012
We've encountered a problem with our feed of new work units, so there's no new work available right now. Sorry! We're working on fixing it.

10 September 2011
Due to a flood of spam, we've temporarily restricted who can post to the project forums.

9 April 2011
A quick update: A lot of spam profiles were removed in the last few days, do complain if we removed any legitimate ones by mistake! Also, preliminary data for all triples with c no more than 10^18 has been available from our data page for a while now, although it is (of course) still preliminary at this point. Many thanks to our dedicated users, and we've now moved onward to further reaches of the search space!

3 October 2010
ABC@home will be unavailable for some periods over the next few days.

24 August 2010
Binaries for all platforms have now been updated. There seem to be some issues with Windows 2000 hosts, which we'll try and fix in the next few days. Please report any other problems!

21 August 2010
We will be releasing new binaries over the next few days, with some speed improvements, fixes for some problematic regions, and as preparation for yet more updates. Please report any problems!

18 August 2010
The old applications have been released with a new code signing key as 2.01.

15 July 2010
We will hopefully be attempting some more upgrades over the next few days, probably resulting in a few hours of downtime.

2 July 2010
There might be downtime in the next few hours as the server is upgraded. Server was (hopefully) successfully upgraded, please report any problems in the forum.

23 November 2009
The new application has been running for a few months now, and it is much faster than the old one. Starting today we make results available as they come in.

5 June 2009
Our new application is ready for testing! See the forum for the full announcement.

9 January 2009
See the forum for news on the continuation of ABC@Home.

7 October 2008
There's a new developer at ABC@Home: S@NL - FilmFreak. I'll give the site a good update and add some pages about the abc conjecture.

16 July 2008
DNS for should be functioning properly again now.

15 July 2008
Our apologies for the current downtime caused by DNS problems. We are having some trouble contacting the people responsible for updating our DNS configuration.

14 April 2008
The server upgrade has been completed, and things seem to be running ok. If you encounter any problems, please report them on our forum.

13 April 2008
There will be some downtime on monday 14 April (in the evening, western European time) for a server software upgrade. Assuming all goes according to plan, it should be done within an hour.

21 November 2007
Powerpc client has been added. From now on version 1.03 is mandatory, and results from previous client versions are no longer accepted. (1.03 has been available for 4 weeks.)

22 October 2007
We're back on a gbit link to the net

20 October 2007
Tomorrow the server will be shutdown because monday it will be moved to a new network. On monday it should get back online

13 October 2007
I added additional workunits that somehow got skipped. Thanks a lot to Cori for the new logo!

13 October 2007
I fixed a bug in the stats daemon, stats are being regenerated. I also redesigned a part of the makework daemon, now before it is started it does a precheck over a large interval trying to predict the number of work to be done in a small interval for a workunit. It won't help any with these peak workunits over 30 hours, but it will prevent workunits that are too short, hopefully

11 October 2007
I doubled the workunit length for now, because most were really getting to short, and it's not good for the database to have tons of small workunits. I hope this doesn't cause inconvenience, today or tomorrow I will try to think and make something that will sort of predict the workunit length in order to have more or less constant workunit length

10 October 2007
Check out the images I've made about abc-triples from the project here

09 October 2007
Partial results from the project are available here for everybody to download and will be updated once every day. The binary file contains all abc-triples we've found so far. Probably later on we will release something to read the file with and search through this database of triples. If you don't need this, please don't download it to spare the server

09 October 2007
We have a new set of forum moderators to keep the forum a bit ordered when necessary. They are listed on project personnel linked from the mainpage. The 64bit windows version is going strong

08 October 2007
Both linux and windows 64bit version 103 are now in the stable project. You can expect the same amazing speed for the windows version as you're used from the linux 64bit one! The validator has also been adjusted to allow mixed results from version 102 and 103, if you encounter any problems, please report them as soon as possible. Happy calculating!

07 October 2007
The 64bit windows abc client is doing very well in the beta, so I will most likely put it into stable tomorrow

07 October 2007
The 64bit windows abc client is currently in the beta project waiting to be tested, together with other new clients except the mac one which yet needs to compiled. Please don't use applications from the beta with the stable project, it will generate all your results invalid

06 October 2007
Good news, the 64bit (truly 64bit) windows client is finally done and seems to work great! It will go into the beta project and then as soon as possible in the stable. Especially for this, there is a logo contest to express your great emotions about this, which can be found here

06 October 2007
The forums have been updated so it's unlikely to miss any user input. It seems that our mail server of the institute is dead, so it is impossible to receive mails from abc. Hopefully it will be fixed soon by somebody

06 October 2007
The delay bound has been increased to two weeks, let's see if that is large enough. The homepage layout has changed. The news from the beta project will from now on also be posted here, in order to simplify and let you see news faster

05 October 2007
In the beta project we are currently testing a new abc client, only for linux at the moment, which should make the code portable to more platforms including mac powerpc. It also contains better anti-cheat protection

01 July 2007
I added the mac intel version of the abc application into the stable project. Please let me know if there are any problems. Added a favicon made by Krunchin' Keith, thanks!

16 June 2007
Back. The workunit length has been slightly increased, the database has been considerably cleaned up, it should run a lot faster. Furthermore, the amount of workunits one can get has been increased from 5 to 20

13 May 2007
I'm going away to Italy for 3 weeks, my colleague Willem Jan will monitor the project. If you tested the mac client in beta, please let us know

04 April 2007
The internet gateway bailed out again. Nothing is wrong with the abc servers, don't worry they cannot handle it, abc can handle probably more than 50000 users with the new hardware

23 March 2007
Fixed the new stats and they are now being regenerated 'again' and are now appearing accordingly on All previously problems should be gone now with them

21 March 2007
The server status page has been 'fixed'. Some people got confused of the status page stating the upload/download server was disabled. However the upload download server was always on. On the status page it was turned off due to a boinc inconsistency, since some boinc server components did not recognize (and probably still don't) the uldl_pid and sched_pid xml tag.

20 March 2007
ABC@HOME beta is back, now running on a seperate server. The url changed, please report this to the stat sites: I hope we can soon start testing again a 64 bit windows binary

18 March 2007
I redesigned the abc specific statistics and they are currently already running. All statistics have been regenerated from when we started. All mentioned problems should disappear with this update. We're going to try to inject this into rekenmeemetabc tomorrow

17 March 2007
Everything is going great. Boinc stats are now getting updated again. Rekenmee now also runs on the new server without problems. The specific abc statistics of rekenmeemetabc will reappear shortly, they are partly being re-implemented and need to be regenerated. I need to re setup abc beta on the old server and then we will focus on releasing a windows 64bit and a mac client

15 March 2007
Today I moved the installation of the old server to the new one. The new one is far more powerful: 4 cpu's, 4GB of ram, gbit interfaces and very fast scsi disks should let everything run faster and smoother. Rekenmeemetabc will follow tomorrow. If you encounter any problems, report on the forums

11 March 2007
Down and up again, this time because the internet connection of the institute where the server resides was down. Now that I'm editing the news anyway, the new server is installed and seems to run really nice, so somewhere this week we will transfer. Further more, a new good abc-triple has been found belonging in the top 200 of the best abc-triples. The triple is a=13.651.536.370.466.816, b=875.615.065.424.903.125 and c=889.266.601.795.369.941 with quality 1,419559 and it has been found by the user Wiztar. Congratulations Wiztar!

04 March 2007
A small downtime occured because of mysql acting weird

04 March 2007
The workunit length has been doubled, almost all workunits were really too short. Tomorrow I'm beginning to install the new server and at the end of the week I think we will migrate, there might be a downtime of a few hours at most then

28 February 2007
The server has been out again for a while, now due to an assertion failed in ext3 code in the linux kernel

16 February 2007
The new server and the mac have arrived. Available work has been increased to 750 workunits, which is more than enough. A bug in the assimilator has been fixed

14 February 2007
The 64bit version for linux has been put into stable. It's about 3x or more faster than the 32bit version. A windows version still is giving some trouble, but should appear later on

10 February 2007
Credit granting has been adjusted in a positive sense, longer workunits are awarded a bit more to stimulate to finish them, you can read more about it on the forums. A macintosh has been ordered to begin developing and building a mac client

10 February 2007
We've been out for 12 hours due to network problems

07 February 2007
A new server has been ordered which will be more than sufficient to run everything smoothly. Further more I'm going to try before the end of the week to get a 64bit windows binary compiled, if it fails at least the linux 64bit version will then be put in the stable project. A few days the workunit length has been doubled, this should only influence the short workunits and not the ultra long ones

06 February 2007
It's been a while since the last news. Everything is running pretty good, and we're very satisfied. Currently there is a friendly exciting competition going on between Boinc Australia and all french speaking persons gathered in l'Alliance Francophone. Who will win is an interesting question, the French speaking guys have assembled an entire army, but the Australians are tough! Or will a third team arise and claim victory? Further more, today I rewrote part of the stats daemon, I hope it will be implemented tomorrow: stats can be updated more frequently and it contains a few bug fixes

29 January 2007
Yesterday the ABC server crashed due to a kernel oops. Everything should now be fine again

24 January 2007
Fixed a bug in the statistics program that caused some missing personal triples in your personal account view on Reken mee met ABC.

21 January 2007
Workunits have slightly been increased in overall length. Statistically, this should not influence the long workunits by making them even longer, but it will make all the shorter workunits more normal in length. If you might encounter a long workunit anyway, you should be fairly safe with version 1.01

20 January 2007
A great improvement has been made on both checkpointing and progress indication in the new just released version 1.01 . Progress is very very accurate and should update about every second. Checkpointing now totally relies on your boinc preference settings, there are no more large time gaps between when abc gives boinc a chance to checkpoint. I have to mention that this was made possible by all the good feedback that was given on the forums. The long workunits took us by surprise too, and by all the reporting we realized our old progress and checkpointing system was not sufficient for the huge bound 10^18 we scan to. The old system left users wondering how long it was going to last and left them in uncertainty. After we redesigned it excellent testing took place and bug reporting was done in the beta project, ensuring a reliable client in the stable. So when you see odd behaviour and you don't like it as a boinc user, report it, and keep reporting it, so that we together can make a rock solid client in the end. Note that workunits still can last long, but now you know exactly how long it is going to last, and you can stop it anytime and resume with almost no loss in computing time.

16 January 2007
On ABC@home beta testing of 64bit applications has begun. If all goes well after some good testing they will appear in this stable project soon, because they are very suitable for ABC@home and can be quite faster than their 32bit counterparts.

15 January 2007
ABC@home and "Reken mee met ABC" have been featured in several media, including You can view the article (in Dutch) here . Another article can be found here

15 January 2007
Some more news once again! As some of you might not know, the algorithm ABC@home uses to find abc-triples is brand new, and much much faster than any method before. Together with distributed computing power we are now able to scan where no man has scanned before. Also ABC@home has a national Dutch sister site called Reken mee met ABC. On this site counters are visible that show how much hits have been found and how many triples we scanned through already together! Statistics are also available, showing a top 100 each day of the best participants for example, who found the best abc-triple, who scanned the most triples and so on. You can also login with your BOINC account information (in the upper right corner) to view your own personal statistics. Furthermore there is a progress indication of the entire project. All this will probably become available in the near future for the international site too. Tomorrow this site will be officially opened and featured in Dutch media along with other information regarding ABC@home. The server might be busy, let's hope the load doesn't get too high. If it does and you cannot reach it anymore, please be patient with us!

15 January 2007
ABC's workunits seem to be coming more predictable, we're slowly getting to a point were they should have more or less a constant running time, we're not there yet, but we're closing in. On most hosts a workunit lasts now about 0 till 0.6 hours cpu time, with an occasional long workunit (which you should not abort, because it will finish correctly!). We strive to make them about 1 hour on most hosts, the last thing we want is irregular workunits jumping here and there with unpredictable behaviour and where you as a user do not know what is happening with it.

15 January 2007
Some people at the university are as it looks playing with the wrong wires, let's hope internet get's stable soon again

14 January 2007
Account creation has been opened and a mass mail has been sent out to our beta testers (we try to sent as little mail as possible). To all our beta testers: We are very gratefull for you and you enabled us to have come so far, without you we couldn't have made it. Now, let us calculate!

14 January 2007
Some workunits might last longer than indicated, but this does not mean it is a bad workunit. It will finish, and you will get credits for all that time. So far there hasn't been a single run-away workunit that kept on forever. These long workunits might just contain a lot of lucky numbers and calculations that were unpredictable causing the length of it.

14 January 2007
Again news from your local ABC news station! Work has restarted, everything is running smoothly except the duration of workunits. The further we come the more regular they are. Once all new workunits are predictable and have an average running time of about 1 hour, I can set the deadline and estimates properly
so that progress is indicicated properly.
After that account creation will be opened.
How fast we get there depends on the current testers, but I'd say pretty fast! Hopefully in a few hours

13 January 2007
Another 1000 workunits are generated.
Most of them should be done by tomorrow morning,
and then account creation will be openend.

13 January 2007
Statistics are now exported again in
Notify your local stats supplier. Don't forget ABC@home
is a different project than ABC@home beta, they are now independent.
Beta will be solely used for testing purposes.
Here your credits et al are all reset to zero, but not on the beta

13 January 2007
Workunits are waiting for the first diehards.
Please report any abnormalities or things you miss or are not working
on the forums so it gets fixed

13 January 2007
The immense search has started! One of the greatest mathematical searches
for numbers which might last for years.
In a few minutes the first workunits will be available for the participants from the beta.
Once everything is running smoothly the doors will be open for everyone

12 january 2007
The hardware upgrade is complete and succesful.
Plans about going live have changed slightly:
1. I will try to import all users from beta into the stable soon, maybe tonight
2. a 1000 results are waiting there for computation
3. if they are computed ok, and it is running smoothly, and everybody agrees that it is going ok, account creation for stable will open up and we will go live.
You will get the amount of credits from the stable you're used to from the beta, and the results you work off for the stable will be saved and used.
Work from the beta will slowly stop and when the stable is fully up and running work for beta will only be generated when there is something to test.

11 january 2007
Tomorrow a small hardware upgrade will take place before the stable starts. Another network interface will be put in and more memory is installed. In a few weeks a new server will be installed that is much more powerfull, and can handle many many users, workunits, traffic and so on.

11 january 2007
Version 5.40 lasts longer than previous versions. Because of this credits that were granted last night were lower than normal. The amount of granted credits should be back to normal now.

10 january 2007
If you are still attached to ABC@home and are actually doing work for ABC@home beta, please finish all your workunits you have today and either reattach to the beta or wait untill the stable is running.
Otherwise you will not be able to report workunits you got from ABC@home beta,
because tomorrow I will remove the reroute from ABC@home to ABC@home beta and ABC@home will then be used for private testing.
If everything goes well, ABC@home will go live on Saturday.

09 january 2007
Great news: the second test finished absolutely perfect, the amount of hits that should have been returned was 51689 and we exactly got this number, and the list of all triples was a perfect match too.
The second test was also meant as a stress test to see how things are handled with loads of wu's, validating, assimilating them fast. This went without any problems.
The first test is still running because some results haven't been turned in yet. This means the project will go live very soon.

07 january 2007
All work from the second test has been handed out, and almost complete.
It's amazing how much work has been done, we almost (still missing some results which will hopefully be turned in soon, but this might take another few days depending on passing deadlines etc) calculated all abc triples up to 10^10 5 times in hardly two days, and 10^8 is the biggest to be found on the net.
We still have to check and verify all these triples though, let's hope they are correct, so far it looks good.
Up to 10^12 is what is currently known and we hope to go up to 10^18 if there's enough enthusiasm, which means a 100 million more than the current test and a million times more than is currently known.
In other words, really go where no man has gone before!

07 january 2007
Test 1 is going ok but results are dropping in very slowly, about 65% is complete, and about 20% is pending.
Test 2, with the small wu's, is going as a train, I think we're halfway and that in less than 12 hours.

06 january 2007
Another test started with 10 times shorter workunits in comparison to the previous test. This is needed to test data harvesting while assimilating different data and with different parameters intermixed. This new test has a quorum of 1, with multiple target results to reach results faster.
The credits you get should be the same propertionally as you got with longer wu's.
Note that the previous test is also still going on, please don't cancel any workunits.
Once these tests are complete and finished properly, we can go live, so help along!

06 january 2007
The test is so far going good, I increased the delaybound to 4 days because the workunits last longer than anticipated.

06 january 2007
Today together we're going to calculate as a test all triples up to 10^10.
The duration of workunits can fluctuate for this, the total amount of workunits will be around 1000. During this testing period there might in some stages no work available. The deadline of these workunits will not be 3 days but 2 days, results are needed fast and the workunits should be easier than the ones sent out before.

05 january 2007
A new version is available, 5.36 , with some bugfixes.
We will probably go live next week or the week after.

31 december 2006
ABC@home wishes everybody a healthy and happy 2007!

29 december 2006
Project information about who we are and what this is all about is available

29 december 2006
Forum links should be fixed that referred to, result links do not work because they got cleaned and keep getting cleaned every so often.

29 december 2006
This project will start in January. You can already join the beta on Once the project starts you are automatically a member of ABC@home

27 december 2006
Staying attached to and being redirected by us through works, however boinc complains with an invalid url. The choice is up to you what to do. Forums in here have been closed, please use the ones of the beta for now. Beta will be now open for the public, i.e. account creation is open. Once the stable starts the beta userdatabase will be imported into the stable and credits will be reset to zero in the stable (not beta!). The stats for the beta are available through, please report this to your favourite stats supplier and notify the change of the project to permanent testing, beta development stage.

26 december 2006
The work daemon has been stopped and a large batch of 1000 workunits has been released with a quorum of 1. Everybody has more than 24 hours to work off any result they still might have or still get and report them. After that, I will dump the database and we will move to ABC@home beta: . There all your credits etc. will reappear, and the current project will become stable. Stable will go live around the beginning of January if everything goes as planned. All user accounts will remain on stable but credits will be reset to zero, they will not be reset in beta. Beta is a perm testing project and will continue with development, including new clients for amd64, mac, other archs and including screensaver code. After a certain life cycle new code will be moved to stable. Statistic sites can be updated once the movement is complete.

25 december 2006
The bug seems fixed. The current client and other code seem pretty stable, so it will probably be frozen now untill the project goes live for the public.

24 december 2006
We progressed about 1100 workunits today so far, and they all validated properly. I think it is almost safe to say the bug is gone. If no invalids reappear, the project will go live in/around the first week of January. A merry christmas to everybody and our appreciations of all the calculating that has been done so far.

24 december 2006
Last night we worked off 400 workunits together, and all workunits validated ok so far. I hope everybody can keep testing today so we can handle more workunits and hopefully they will also be all valid. Be sure though to use 5.34 and 5.35 and not any previous ones! Version 5.33 gave some errors which was during the rough testing period last night as stated in the forum. The credits granted are more than is claimed for almost everybody now, but I will look into that later. The most important question at this moment is, if the bug is fixed. Besides, it makes up for some of the lost hours some testers ran with invalid workunits.

23 december 2006
A lot was changed today. Version 5.34 should be used, any earlier version should be aborted. You can view some statistics about active users on . Maybe 5.34 solves the client bug. The rest that was changed is internally to make things work more safe and involves some additional checks on input.

19 december 2006
The bug has not yet been fixed, the generation of work is temporarily stopped because at this moment more testing is not necessary anymore and more a waste of cputime until the client bug is fixed. Once there's a new version to be tested, work generation will start again.

12 december 2006
We found most likely the bug that was responsible for reporting invalid hits. This means, in theory if everything is ok, there will be no more invalid workunits! The windows client has been rebuilt using new boinc code, previously it was built against some old so-called stable boinc code.

11 december 2006
Fixed a bug which was properly responsible for invalidating a bunch of workunits, now about the old 1% or even less of your workunits should be invalid (we hope). Increased validator credit granting by 10% overhead to hopefully make up for the invalid workunits. The bug that some triples are found as hits is still not solved.

11 december 2006
More and more workunits are invalid due to calculation errors. We suspect it has to do with the library we use and the larger the numbers are, the more likely the faults arise. It's not necessary to keep calculating on loads of workunits because of this, but please continue to do some workunits now and again so we can debug output from clients. This is really needed and we appreciate any help. If this gets fixed, all workunits should validate properly.

09 december 2006
A power outage occured, there was a downtime of about 15 hours.

08 december 2006
The pentium 3 code seemed to cause more invalid workunits so we're back to the old optimized code. Credits are increased now with 20%.

07 december 2006
A new client is being pushed out, 5.26, which should fix the bug of resetting the cpu time when the client is restarted. The windows binary should again be faster than yesterday, it might not work on pentium 2's or or some other cpu's like via's, let us know if it doesn't. A statistics program has been written to parse out statistics of our data, you will see the results of that soon (ie, best triple top100, a realtime triple counter and so on).

06 december 2006
There were/are some problems with the running time between the windows and linux binary, the linux executable run a lot faster. This caused linux machines to get more credits assigned per time unit. The new windows binary 5.25 should be 2-3 times faster than the old versions. Of course we want all binaries for whatever OS to run as fast as possible, and we are working towards that. Accordingly, the granted credit for work has been adjusted. It should be more fair between windows and linux now, if it's completely fair we still have to see.

02 december 2006
We scanned 38604947537 triples so far and found 366590 hits. About 0.1% of the workunits is invalid, this will be fixed soon. The credit system seems to work good. Statistics are now exported to several boinc stat pages.

29 november 2006
We worked a lot on the client today, from version 5.13 till 5.21. The client should now be a lot more error-proof. The update indication seems to work better. Tomorrow the stats for boincstats will be exported. Thanks everybody for still testing it even though we had quite some bumps on the road.

29 november 2006
Yesterday the result database was cleaned because timed error workunits kept regenerating. We didn't think this would cost cpu time since the wu's only lasted for a few seconds. Today a new client will be pushed out that should incorporate a better progress indication.

27 november 2006
The new credit system is now enabled. Try to see what you get and verify it's fair, you can even try to let your boinc client report more cputime or higher benchmarks and see if it influences anything. We also found what causes the bug of the 5% bad workunits, it should be solved soon.

26 november 2006
There is a bug in the client which makes about 5% of the workunits invalid. This will be fixed very soon. Tomorrow we are going to implement a new credits system, which is dynamic and independent of what your boinc core client reports, and which will give you very accurate credits for what you've actually done. The credits granted will be in the same order as other projects.

25 november 2006
The project runs again, the problem is fixed I hope. Please delete/abort any old workunits you're working on, and update to get new ones, otherwise they will be invalid.

24 november 2006
The project temporarily does not hand out or accept work because of a validation bug, making all the calculations useless. It will be fixed within 24 hours.

24 november 2006
Thanks to good feedback we have now set the max delay time to 5 minutes to request new work, set a minquorum of 2 for the moment, sharpened the verification of uploaded results, set a maximum claimed credit, set a delay bound of at least 3 days and raised the fpops limit. We also fixed the checkpointing system which contained a bug. The transition period between the old settings and these new ones might cause some problems. The duration of workunits we leave untouched for now and see where it goes. They should take between 2-5 hours. Registration is closed again, we have close to 300 participants which is more than enough for now. As a side node, you guys already scanned 24215398827 triples of which 332310 were actual hits! This is much faster than anticipated, so after the testing we will increase limits to search for. Fortunately, there are infinitely many abc hits so do not worry :)

23 november 2006
We've been testing a few days now. Some things are going well, some not. We're now working on improving progress indication, the checkpoint system, validate and credit grant system and better delay bounds. We would also like to have some (say 5) volunteers which are boinc die-hards that would like to be forum moderators and advisors of the boinc community to us, that know what is good and not from experience, and who can answer people's questions on the forums and relate to us. There's simply no time for us to both help everyone and maintain the project. If you would like to help out, you can apply on . You'll also get an mail alias if wanted.

21 november 2006
We are now open for testing. When we have enough users account creation will be closed again. The results which are found will most likely be deleted after the test, and some things will be reset except the userdatabase. We hope you can test along, and we appreciate the enthusiasm of the boinc community which we've encountered so far.

16 november 2006
Testing still hasn't started yet, our apologies. We are still doing some work on the client, to make it faster, more static and more reliable.

7 november 2006
We installed new boinc software.

2 november 2006
Migrated to new server

Return to ABC@home main page

Copyright © 2016 University of Leiden