Are workunits on individual numbers or ranges of numbers?
log in

Advanced search

Message boards : Science : Are workunits on individual numbers or ranges of numbers?

Author Message
Profile a2aaron
Send message
Joined: 22 Apr 13
Posts: 7
Credit: 684,328
RAC: 0
Message 21130 - Posted: 25 Aug 2015, 22:32:44 UTC

Even on large (30+ digit) numbers, my computer can easily solve an individual number in less than a second, leading me to belive that the WUs are over large ranges of numbers. Is this true? If so, how large are these ranges?

Dr Who Fan
Avatar
Send message
Joined: 27 May 14
Posts: 21
Credit: 4,562,054
RAC: 0
Message 21131 - Posted: 26 Aug 2015, 5:23:24 UTC - in response to Message 21130.

Some of your tasks are erroring out almost immediately (0.71 Seconds) with "255 (0xff) Unknown error number"
See this one as an example: http://boinc.thesonntags.com/collatz/result.php?resultid=21268952

BUT you are not the only one having that type of problem with those specific tasks. Seems to be am on-going Collatz PROGRAMMING ISSUE.
____________

Profile sosiris
Send message
Joined: 11 Dec 13
Posts: 123
Credit: 55,800,869
RAC: 0
Message 21132 - Posted: 26 Aug 2015, 5:24:24 UTC - in response to Message 21130.

About 2 ^ 40 numbers for a large WU, AFAIK.
____________
Sosiris, team BOINC@Taiwan

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 2
Message 21147 - Posted: 28 Aug 2015, 2:38:47 UTC - in response to Message 21131.

Some of your tasks are erroring out almost immediately (0.71 Seconds) with "255 (0xff) Unknown error number"
See this one as an example: http://boinc.thesonntags.com/collatz/result.php?resultid=21268952

BUT you are not the only one having that type of problem with those specific tasks. Seems to be am on-going Collatz PROGRAMMING ISSUE.


Fixing and testing a new 6.xx app when the sieve app is already being tested doesn't make much sense given that the sieve app does 70-250 times more science in the same time as the current apps. I haven't released the sieve app yet because, while reviewing the code for the hundredth time, I discovered a potential bug. I fixed that bug and have also made some changes that will make porting it to Linux and MACs easier allowing testing to occur on all three at the same time which, if the code is 99.9% the same, should work out the kinks before it goes into production. That's not to say that there won't still be issues. With 3 platforms with 8 apps each, a hundred or more types of GPUs, an equal number of driver versions, and hundreds of combinations of configuration settings, there are bound to be issues that will show up.

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 2
Message 21148 - Posted: 28 Aug 2015, 2:49:08 UTC - in response to Message 21132.

About 2 ^ 40 numbers for a large WU, AFAIK.


You can also tell the size by looking at the WU name. The first number is the starting number. The second is how many numbers it will check. For example:
solo_collatz_2398849770227473842176_1649267441664_2

means to start at 2398849770227473842176 and to check 1649267441664 numbers. The final number (2) means that two other people have been sent the WU but not completed it because it timed out, error out, they aborted it, etc.

In other words, the app doesn't need to read the work unit file. It knows exactly what to do just from the name of the work unit. Some day I'll check whether BOINC can send a WU with just the name but no file. For now, the WU file contains the start and stop numbers, but it really doesn't need them. It's all in the name.


Post to thread

Message boards : Science : Are workunits on individual numbers or ranges of numbers?


Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.