First

Hi Joe ,
if you see any Technical errors I have Made in my post - could you please make the neccesary corection by makeing a post , thanks in advance :-)


Second

Hi Eric ,
I hope you had a Good Christmas :-)

Third

Eric ,
Josef W. Segur has been using three of my multi-CPU WinXP hosts to investigate multi-CPU WinXP systems, including the occasional unexplained errors. -- Joe had said he might be able to figure out something about those errors if they do show in -- 517STO2 -- It's a matter of being able to correlate the address of the errors with the source code

Eric ,
Joe has had tremendous success ... after a lot of hard work and mental effort -- Since 22 Dec 2006 21:34:33 UTC --- until now ---27 Dec 2006 15:01:44 UTC --- my three multi-CPU WinXP hosts have in beta testing produced - and successfully completed -- 295 -- tasks -- wu -- with one only computational error -- which Joe says -- "he had an error on a unit which was -9 overflow on other systems. That error comes after crunching has started, but is somewhat similar since it also involves exception handling."

Eric
I have consistently been getting - four computational errors for - every - twenty tasks - or - wu - completed - I beta testing --- now 295 - successfully completed and only one computational error :-)

In message 11054, BYRON LEIGH HATCH wrote:

...
Josef W. Segur has built a excellent SETI@home Enhanced Application called _5.17SB1_build by Josef W. Segur_ that runs beautifully on operating System Windows 95 , 98 or ME
...

Byron, I'm going to suggest another use for 5.17SB1. It's almost identical to the stock build, so should behave the same as stock on your multi-CPU WinXP systems, including the occasional unexplained errors. I might be able to figure out something about those errors if they do show in SB1. It's a matter of being able to correlate the address of the errors with the source code, which I can't do for Eric's builds. Eric hasn't been able to pin down the problems, but he has limited time to spend on such efforts. There's just a chance I might have better luck either because of a different perspective or because I'm retired and can spend as much time on the effort as I want.

I don't have a multi-CPU system so can't test directly. Anyone else with multi-CPU systems could also try SB1 if they want, there seem to be a number of different errors which crop up for some but not others.

Joe

Consider 5.17SB1 for unexplained error test

After a day and a half using ,517STO2 , the startup errors haven't showed on Byron's systems. 86 WUs completed with no errors, 1 had an error on a unit which was -9 overflow on other systems. That error comes after crunching has started, but is somewhat similar since it also involves exception handling.

Although the indications are good, more time and testing is needed. Byron is running Seti Beta only so always gets the "...starting fresh" path. If there's someone running with multiple projects who is seeing occasional errors at restarts of S@H I'd appreciate some testing of 5.17ST02 with that host.

Joe

just a friendly Little reminder and Bump :)

Hi Byron,

Sorry to have taken so long to reply.

I guess my question is , how does the progress look to you in the crunching of - the data - from - Multibeam - using app 5.17 ? :-)

and is there anything new on the horizon , that you could tell us about ? :-)

Josef W. Segur has spent many long hours , as an unpaid Volunteer , helping , Dr. Eric J Korpela , Project scientist , meticulously read through code to find , any bugs or problems in app 5.17 running Multibeam data

Hi Byron,

Sorry to have taken so long to reply.

I guess my question is , how does the progress look to you in the crunching of - the data - from - Multibeam - using app 5.17 ? :-)

So far things look far better than I expected. The error rate doesn't see to be significantly changed, but at least it didn't increase. I'm getting close to having enough data returned so that I can adjust the thresholds, but that doesn't require a new application release. But I would like to get a new release out there that fixes the Windows 98 problems.

and is there anything new on the horizon , that you could tell us about ? :-)

There has been a lot of work done by volunteers (Alex Kan, Joe Segur, Ben Herndon, Simon, Tetsuji, etc) to optimize the code. I would like to get that optimized code into our stock application, but I haven't had the time to complete that project yet. Other (unavoidable) things keep getting in my way. I do need to compile a new application for the main project before we can start doing multibeam work there.

Astropulse is actually getting close. We have a working Astropulse splitter again, and Josh is doing test runs of the application. I hope to get it into beta early in the new year.

Josef W. Segur has spent many long hours , as an unpaid Volunteer , helping , Dr. Eric J Korpela , Project scientist , meticulously read through code to find , any bugs or problems in app 5.17 running Multibeam data

Yes he has, and he should be thanked again and again for his efforts. I'm going to be corresponding with him regarding the new builds to make sure all the fixes have been implemented.

And I should thank you for your generous support, and I don't think I've thanked you personally for the DVD and other gifts you sent in May. Thank you!

Eric

Hi Eric,

thanks for letting us know about progress :o)
A happy new year to you and everyone else actively working on and for the project.

And yes, Joe Segur really should be commended for his tireless efforts - though he does like to keep a low profile, he really deserves a lot of credit.

Regards,
Simon.

I'm getting close to having enough data returned so that I can adjust the thresholds, but that doesn't require a new application release. But I would like to get a new release out there that fixes the Windows 98 problems.