A 100 workunits are now waiting to be finished.
I want the workunit time to be around 1 hour for an average machine.
Not too long in case something goes wrong, and not too short because of the overhead.
In the beginning workunits are longer than normal,
I estimate they are around 1-2 hours in the beginning.
The idea was to precalculate the first few workunits, but it just lasts too long, with boinc it really is much much faster. Yesterday I tried to calculate the first few and they took over 10 hours. So now I chopped them in smaller pieces and divide them out.
If you see any peculiar behaviour of these first workunits let me know.
Note: the very first workunit shows a wrong progress
indication probably due to a not properly defined integral.
EDIT: the first workunit might take a lot longer than indicated, it's harder
to predict than the weather. There are mathematical reasons for this.
/ 
I have delivered my first result and everything went smooth so far:
Result ID 59
This workunit was quite short.
Regards
Alex
P.S.: I have submitted a bug report concerning the website here.
____________
/ 
I have delivered my first result and everything went smooth so far:
Result ID 59
This workunit was quite short.
I was the other cruncher on this WU and granted credit is disappointing (to put it mildly). We both received less than claimed and both of us are using a standard issue BOINC client.
But the WU did complete successfully.
____________
Join the #1 Aussie Alliance on ABC@Home
/ I have delivered my first result and everything went smooth so far:
Result ID 59
This workunit was quite short.
I was the other cruncher on this WU and granted credit is disappointing (to put it mildly). We both received less than claimed and both of us are using a standard issue BOINC client.
But the WU did complete successfully.
I saw it. That's why I only generated a 100 at first. Everything will be adjusted
as we go along to get it ok.
/ I'll play :D if you will send me some work.:)
____________
208.55Ghz Home Owned and Operated Crunching Power
/ I saw it. That's why I only generated a 100 at first. Everything will be adjusted as we go along to get it ok.
And I'm reporting it for the same reason. My second WU has the same issue (but you've probably seen that one too)
____________
Join the #1 Aussie Alliance on ABC@Home
/ A 100 workunits are now waiting to be finished.
I want the workunit time to be around 1 hour for an average machine.
Not too long in case something goes wrong, and not too short because of the overhead.
In the beginning workunits are longer than normal,
I estimate they are around 1-2 hours in the beginning.
The idea was to precalculate the first few workunits, but it just lasts too long, with boinc it really is much much faster. Yesterday I tried to calculate the first few and they took over 10 hours. So now I chopped them in smaller pieces and divide them out.
If you see any peculiar behaviour of these first workunits let me know.
Note: the very first workunit shows a wrong progress
indication probably due to a not properly defined integral.
EDIT: the first workunit might take a lot longer than indicated, it's harder
to predict than the weather. There are mathematical reasons for this.
Hi!
Welcome to this "new" project!
My first wu (and this projects first; Oh Yeah!!)just finished at 4203.48 seconds, so the Your prediction is not bad, some other I have ended sooner,about half an hour or just some minutes duration!
Good Luck with this (ever)lasting project!!
With regards,
Hans Sveen
Oslo
Norway
/ Hi Hans, thanks, glad you're helping out.
Yoda, I adjusted the granted credits, it should now be much better.
Let me know how it is.
/ 
Resutl-ID 113 - running now since more than 1hr and stocked by 3.9% for 45 minutes.
/ Resutl-ID 113 - running now since more than 1hr and stocked by 3.9% for 45 minutes.
Can be, the first few workunits also behave irregular with progression.
I gamble it in total takes 1 hour and about 30 minutes, but who knows.
/ Progress report...
I have 5 CPUs on ABC at the moment and progress seems quite erratic. But perhaps there's a mathematical reason for it.
The first WU I completed took half an hour on the Athlon XP 3000+. The second WU on this same computer is at only 9% after 32:33 - quite a difference there. The other computers are all over the place in terms of progress:
- 2.8GHz Mobile Pentium 4 (no HT) is at 44.3% after 0:27:50
- Athlon 64 3700+ (at 2.6GHz) is at 32.8% after 0:43:24
- 2.4Ghz Pentium 4 (no HT) is at 66.1% after 1:23:25
- 3.4 GHz Pentium 4 (with HT) is at only 3.9% after 0:43:06 (EDIT: that matches Inais' report for the same WU)
EDIT: yes, quite irregular :D
/ Progress report...
I have 5 CPUs on ABC at the moment and progress seems quite erratic. But perhaps there's a mathematical reason for it.
The first WU I completed took half an hour on the Athlon XP 3000+. The second WU on this same computer is at only 9% after 32:33 - quite a difference there. The other computers are all over the place in terms of progress:
- 2.8GHz Mobile Pentium 4 (no HT) is at 44.3% after 0:27:50
- Athlon 64 3700+ (at 2.6GHz) is at 32.8% after 0:43:24
- 2.4Ghz Pentium 4 (no HT) is at 66.1% after 1:23:25
- 3.4 GHz Pentium 4 (with HT) is at only 3.9% after 0:43:06 (EDIT: that matches Inais' report for the same WU)
EDIT: yes, quite irregular :D
Thanks :) You can see a bit for yourself how regular they become:
abc_wu_x_y_z , the higher x gets, the more regular it becomes :)
/ Another progress report (last one from me for a while)
All 5 of my computers have now completed at least one WU and they did vary quite a bit in length.
Credit is still on the low side (so far they have all been below the claimed amount) and I understand you're looking into it. In my opinion, it needs to be at least as much as the median claim made with a standard BOINC client, perhaps a bit more (as happens at Einstein, Proteins@home, SETI and a number of other projects).
To make this project attractive for people who are in it for the competition without causing mass walkouts at other projects, I'd look at credit levels approximately 20-25% above the median level mentioned above.
But that's just my opinion :)
____________
Join the #1 Aussie Alliance on ABC@Home
/ Another progress report (last one from me for a while)
All 5 of my computers have now completed at least one WU and they did vary quite a bit in length.
Credit is still on the low side (so far they have all been below the claimed amount) and I understand you're looking into it. In my opinion, it needs to be at least as much as the median claim made with a standard BOINC client, perhaps a bit more (as happens at Einstein, Proteins@home, SETI and a number of other projects).
To make this project attractive for people who are in it for the competition without causing mass walkouts at other projects, I'd look at credit levels approximately 20-25% above the median level mentioned above.
But that's just my opinion :)
It seems you're right and they are still lower than before, I will raise them.
/ 
Hi, So far on abc_wu_1680000_10000, Im at 30.800% Done, With 15:45 (Min) Cpu Time increasing, And the To Completion also increasing with it, Its at 46:15.
Update:
Cpu Time = 28:00 Progress = 30.800 To Comp. = 54:00
Cpu Time = 1:01:10 Progress = 38.600 To Comp. = 1:05:00
Cpu Time = 1:11:11 Progress = 78.300 To Comp. = 18:58
G^R
____________
/ 
My 2 first came in on P4 3,2 HT: 917,02s, 1642,80s
/ Hi, So far on abc_wu_1680000_10000, Im at 30.800% Done, With 15:45 (Min) Cpu Time increasing, And the To Completion also increasing with it, Its at 46:15.
Update:
Cpu Time = 28:00 Progress = 30.800 To Comp. = 54:00
Cpu Time = 1:01:10 Progress = 38.600 To Comp. = 1:05:00
Cpu Time = 1:11:11 Progress = 78.300 To Comp. = 18:58
G^R
Cpu Time = 1:13:33 Progress = 100%
Really seemed to fly once it passed the 1:05:00 Cpu Time Mark (Or Abouts).
CPU time 4413.546875
stderr out <core_client_version>5.4.11</core_client_version>
<stderr_txt>
[17:05:45 ABC] starting abc client
[17:05:45 ABC] retrieved parameters a=1680000 b=1690000 l=1000000000000000000
[17:05:45 ABC] can't open abc_state to read header data
[17:05:45 ABC] starting from scratch
alpha^(1/3): 118
beta^(1/2): 1300
--
[18:24:13 ABC] finished with tot=3480887 hits=477
</stderr_txt>
Validate state Initial
Claimed credit 11.7477332638161
G^R
____________
/ I want the workunit time to be around 1 hour for an average machine.
Not too long in case something goes wrong, and not too short because of the overhead.
In the beginning workunits are longer than normal,
I estimate they are around 1-2 hours in the beginning.
Your "not too long in case something goes wrong" idea is good but compared to other beta projects your app seems very stable, just the progress indicators are a little irregular on some WUs which isn't a big problem.
I don't know where the impact of the overhead diminishes to "negligible" with respect to run length but considering how long it may take to get to 10^18 it seems wise to reduce the impact of overhead as much as possible. I think you would get very few complaints if WUs took 2 hours on an average machine. But what is "average machine"?
Now, I have a WU processing on an old 233 MHz machine and the predicted completion time is 7 hours. That's not too long for me and I think a lot of other crunchers would be OK with that too. Of course, what you want is to balance the impact of overhead against the impact of long run times on number of crunchers. If you could assign longer WUs to faster machines and shorter WUs to slower machines it would help. If you make it an option then most will select short WUs even if they have fast machines, because they won't consider the impact of overhead, so don't make it an option. Crunchers will accept your motive when the reason is explained.
/ Hello !
So far everything running smooth and nice, at least on my side; I am even the first one to hit total credit of one hundred {:>>)), not even a single error so far today!
Will more work be sent out/ made , the server status page says the abc_makework is not running. I guess You will wait for the running wu's to turn in, or start up open up the project after this weekend??
Back to abc-beta and other projects,
have a nice weekend every body. I am soon going back to bed, shivering from catched flu!!
____________
Hans Sveen
Oslo
Norway
/ WU = abc_wu_2530000_10000 Finished just over an hour (3,696.66 Sec.)
G^R
____________
/ 
This WU is running right now.
Ran nicely for the first 5 minutes and got to 21.9%. It's now sticking there for the last 30 minutes.
I'll let it run and see what happens. This might just be normal behavior.
Memory usage has dropped to 24112K from 50312K.
And I haven't seen a checkpoint since
1/13/2007 2:11:03 PM|ABC@home|[task_debug] result abc_wu_2510000_10000_0 checkpointed
It's now 2:55
/ This WU is running right now.
Ran nicely for the first 5 minutes and got to 21.9%. It's now sticking there for the last 30 minutes.
I'll let it run and see what happens. This might just be normal behavior.
Memory usage has dropped to 24112K from 50312K.
And I haven't seen a checkpoint since
1/13/2007 2:11:03 PM|ABC@home|[task_debug] result abc_wu_2510000_10000_0 checkpointed
It's now 2:55
Can be, this should after a 1000 or more workunits be not occuring anymore.
/ This WU is running right now.
Ran nicely for the first 5 minutes and got to 21.9%. It's now sticking there for the last 30 minutes.
I'll let it run and see what happens. This might just be normal behavior.
Memory usage has dropped to 24112K from 50312K.
And I haven't seen a checkpoint since
1/13/2007 2:11:03 PM|ABC@home|[task_debug] result abc_wu_2510000_10000_0 checkpointed
It's now 2:55
Can be, this should after a 1000 or more workunits be not occuring anymore.
It has jumped up to 57.5% and it wrote a checkpoint about 20 minutes after I posted. So all seems good. It's now waiting for it's next time slot to run.
/ WU abc_wu_3610000_10000 only ran around 11 minutes also (686.88 Sec).
G^R
____________
/ 
Hi Hendrick
Wat up today !?? Mi Quad core take wus around 2 hours of workunit, mi E6600 around 60 minutes and the old P4 1 hour 20, somes wus ares sended in 5 minutes and others in 2 hours... same before in alpha... i hope there gone to stabilise
Merci beaucoup
____________
/