Message boards :
Number crunching :
How Many Splitters Are Used To Split For SETI Classic?
Message board moderation
Author | Message |
---|---|
Timcom99 Send message Joined: 30 Sep 04 Posts: 105 Credit: 8,927,290 RAC: 0 |
Just wondering how many Splitters are used to Split for SETI Classic? SETI BOINC uses 3 (Galileo, Milkyway, and Philmore). This is for when Classic does end and the Classic Crunchers move their Big Machines over to SETI BOINC. We probably will need 1 or 2 more Splitters at least to keep up with Demand for Work Units. |
Clyde C. Phillips, III Send message Joined: 2 Aug 00 Posts: 1851 Credit: 5,955,047 RAC: 0 |
Just as important as the number of splitters is their capacity and how much they're used. I understand that the splitters used by Seti have different capacities. I saw that somewhere. It's possible that there are only three that do all the work for both SetiClassic and SetiBoinc. Maybe that is the reason why I have seen, since about June 1, 2003, the flattening of the workunit production graph at Current Progress/ Collect Data/ Workunit Production at the SetiClassic Website. Maybe the same number of workunits have been split (or possibly fewer) but since June 1, 2003, some of those have been diverted to SetiBoinc which has resulted in the flattening. Maybe the splitting limitation is responsible for the paucity of data both at SetiBoinc and SetiClassic. |
Matt Lebofsky Send message Joined: 1 Mar 99 Posts: 1444 Credit: 957,058 RAC: 0 |
> It's possible that there are only three that > do all the work for both SetiClassic and SetiBoinc. Actually, this is exactly the case. The biggest bottleneck in splitting right now is CPU time. Adding CPU to the mix would be no big deal except for some nagging solaris/informix conflicts - the splitter machines need to running a specific older version of solaris to work. This will be fixed after the master database migration is complete (and the conflicts go away). > Maybe that is the reason > why I have seen, since about June 1, 2003, the flattening of the workunit > production graph at Current Progress/ Collect Data/ Workunit Production at the > SetiClassic Website. Actually, the reason for that this web page is broken and WAY out of date - one of the many things worked on originally by somebody else so it's a real pain to crack the mechanisms open and fix. At any rate, this page will be rendered obsolete once SETI classic goes away, so fixing it is loooow priority. > Maybe the splitting limitation > is responsible for the paucity of data both at SetiBoinc and SetiClassic. There is no paucity of data for either project - people aren't getting data because they are unable to get through to our data servers for one reason or another. Whole other issue. Over the last few weeks there's been a continual backlog of a half million results to send to users in BOINC (see server status page). - Matt -- BOINC/SETI@home network/web/science/development person -- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude |
Alan M. MacRobert Send message Joined: 10 Apr 99 Posts: 13 Credit: 4,385,900 RAC: 1 |
>> Maybe that is why I have seen, since about June 1, 2003, the flattening of >> the workunit production graph at Current Progress/ Collect Data/ >> Workunit Production at the SetiClassic Website. > > Actually, the reason for that this web page is broken and WAY out of date... Matt, Could you tell us what IS the total number of workunits split since the beginning of the project? And the total results received, Classic and BOINC? So that I can post the right numbers on Sky & Telescope's web site. The numbers now on the SETI@home site make it look like for nearly two years there has been gigantic redundancy and waste. Thanks very much. Alan MacRobert Senior Editor Sky & Telescope |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Matt, I hope you got to enjoy your vacation, You earned it. tony |
Matt Lebofsky Send message Joined: 1 Mar 99 Posts: 1444 Credit: 957,058 RAC: 0 |
> Could you tell us what IS the total number of workunits split since the > beginning of the project? And the total results received, Classic and BOINC? > So that I can post the right numbers on Sky & Telescope's web site. The > numbers now on the SETI@home site make it look like for nearly two years there > has been gigantic redundancy and waste. In the classic SETI@home project, as of RIGHT NOW, the numbers are: workunits generated: 222636185 results received: 1618010710 I know the web site is reporting 1.9 billion results - not sure why. Just noticed the discrepancy myself, and am looking into it. At any rate, the seemingly obvious result is that we have gotten 7.25 results per workunit on average. Bear in mind this key word here is average, and therefore is heavily skewed by several short periods of time where we were unable to create new work and so the redundancy levels were ridiculously high (sometimes as high as 20 results per workunit or more). But the mode is more like 4 to 5 results per workunit during the course of the project (it would be a sufficient drain on our busy database to calculate the exact number right now). In 1999, we were lucky to get 2 or 3 results per workunit, but thanks to the Moore's law and the unexpected large numbers of users motivated by the credit-gathering competition, we now have more CPU power than we need. Now, in 2005, if all systems are working at full speed, we get around 5 to 6 results per workunit - and that's *after* adding twice as much scientific computation to each workunit a couple of years ago. This was a known problem with the classic SETI@home model where users could get as much work as they wanted whenever they wanted, regardless of scientific merit. We never expected such an amazing surplus of CPU power (which is a wonderful thing). So we never built in a graceful method for turning users away. That's where BOINC comes in - redundancy has a ceiling, which is currently set at 4 results per workunit. If there's no work to be done, users can crunch on other projects. Basically, we're "sharing the wealth." In defense of SETI@home, handing out excess busywork to keep the crunchers happy has both pros and cons, not just cons. Anyway.. the current number of workunits in BOINC is: 2072191 and the results is.. well the count of results is taking too long to come back and I need to get on with other work. Just multiple the previous number by 4 (see logic above). - Matt -- BOINC/SETI@home network/web/science/development person -- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude |
Byron Leigh Hatch @ team Carl Sagan Send message Joined: 5 Jul 99 Posts: 4548 Credit: 35,667,570 RAC: 4 |
Hi Matt , thank you very much for your very good post ... enjoy your vacation ... as tony says ...you certainly have earned it. byron . |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.