S@H V5.17

[Dirk Sadowski]

Hello!

Of course, you noticed too, that at S@H Beta is now the V5.17 (Windows)...

When do you can do an optimized version?

Greetings!

 


PS. I need MMX and SSE2 

[mmciastro]

This is just my opinion, but I don't think users should be using any optimized applications in Beta unless Eric approves it.  Beta is to test the original application (Erics).  It would be like running water in your house while a plumber is checking the system.  It may be OK but it adds layers of complexity Eric might not want while he's trying to fix the app.

2C inserted.

[Dirk Sadowski]

Hello!

No, no, Iīm a "normal" (not Beta) S@H User... 

Greetings!

[Josef W. Segur]

This is just my opinion, but I don't think users should be using any optimized applications in Beta unless Eric approves it.  Beta is to test the original application (Erics).  It would be like running water in your house while a plumber is checking the system.  It may be OK but it adds layers of complexity Eric might not want while he's trying to fix the app.

2C inserted.

There have always been optimized versions in the Beta testing of setiathome_enhanced. Tetsuji made and distributed some in the early days, and continues to make his own personal builds. Crunch3r's limited group of testers also participated on Beta Test until enhanced went to main. Eric has never complained.

So long as it is a small percentage of those participating and the optimized apps identify clearly in stderr.txt it does no harm. Having more people working with the source and possibly finding bugs by compiling with different tools and/or flags could do a lot of good.
                                                                      Joe

[MikeK]

Would like to try a 5.17 with my new X2 4800+ @ 2,5 GHz.

Mike

[Dirk Sadowski]

Hello!
Of course, you noticed too, that at S@H Beta is now the V5.17 (Windows)...
When do you can do an optimized version?
Greetings!
 
PS. I need MMX and SSE2 

Hello Simon!

Or it is like this, that I can run the new coming out original S@H V5.17 WUs with your V5.15 or Cruch3rs V5.12?

Greetings!

[Josef W. Segur]

Hello!
Of course, you noticed too, that at S@H Beta is now the V5.17 (Windows)...
When do you can do an optimized version?
Greetings!
 
PS. I need MMX and SSE2 

Hello Simon!

Or it is like this, that I can run the new coming out original S@H V5.17 WUs with your V5.15 or Cruch3rs V5.12?

Greetings!

All work at the main setiathome project can be successfully crunched with 5.12 or better, yes. It's all produced by splitting data from the Arecibo 1420MHz Flat Feed. Sometime in the future, after a new version comes out of Beta, there will be work from the ALFA receiver which may require 5.17 or better.
                                                                    Joe

[Dirk Sadowski]

Hello!
Of course, you noticed too, that at S@H Beta is now the V5.17 (Windows)...
When do you can do an optimized version?
Greetings!
 
PS. I need MMX and SSE2 

Hello Simon!
Or it is like this, that I can run the new coming out original S@H V5.17 WUs with your V5.15 or Cruch3rs V5.12?
Greetings!

All work at the main setiathome project can be successfully crunched with 5.12 or better, yes. It's all produced by splitting data from the Arecibo 1420MHz Flat Feed. Sometime in the future, after a new version comes out of Beta, there will be work from the ALFA receiver which may require 5.17 or better.
                                                                    Joe

Hello!

Then itīs like this, that when "original" S@H V5.17 comes out (after testing in BETA), Simon "must" (when he would like, I hope he will do) do a new optimized version.
Thatīs correct?

You said, sometime in the future... (S@H V5.17 is now in S@H Beta), so in around 4 weeks itīs for the main S@H, or?

Greetings!

[Josef W. Segur]

Then itīs like this, that when "original" S@H V5.17 comes out (after testing in BETA), Simon "must" (when he would like, I hope he will do) a new optimized version.
Thatīs correct?

You said, sometime in the future... (S@H V5.17 is now in S@H Beta), so in around 4 weeks itīs for the main S@H, or?

Possibly 4 weeks, but probably not 5.17. The Windows 5.17 is unusable on Win9X, for the Mac 5.17 I saw results from one system which had been doing 5.15 with no problems but every 5.17 result errored out in 22 seconds. Eric hasn't produced a Linux 5.17 yet, says there's a BOINC side problem...

I'm not sure what the situation is with ALFA data. I presume the guys who installed the recorder brought back at least some data, and someone is probably working on splitter code. I suppose that could all come together in 4 weeks too.

One potential issue which shouldn't be forgotten is that 5.17+ will be using a 3.81 multiplier in reporting fpops_cumulative compared to 3.35 for 5.15. That 13.7% increase will get a lot of attention.
                                                                      Joe

[msattler]

When 5.17 is rolled out to mainstream Seti, will they go through a transition period (like the transition to enhanced) where the scheduler will still parse out 5.15 wu's to users who are not upgraded yet, or will they just force the Seti client to update to 5.17?  And if they force the update, will I come home from work some day to find all of my 7 crunchers merrily running pell mell through the wu's generating errors ('cuz they are all on Chicken apps)?  Or will Simon (I guess I don't know if he prefers to be referred to as Simon or the notorious Chicken) have his apps upgraded to 5.17-ready in advance to we can all upgrade before they start coming out?  I know, a lot of questions.

[Simon]

Actually,

Simon does just fine

I would imagine that there will not be a hard cut to 5.17 required. Based on application version (your BOINC client knows which you have in your app_info.xml) it should still send out crunchable WUs to 5.15 apps (meaning no ALFA data).

Still, as soon as there is a more formal announcement as to when 5.17 will come out of beta, I'll try and be ready with a version, as well.

Thing is, to my knowledge only the Mac code was recently changed (if you believe the changelog, that is) so the code changes I've heard about are still not incorporated.

Regards,
Simon.

[JavaPersona]

With all due respect, folks.  5.17 is the required version of SETI for Windows and has been for a few days!  I just reconnected my machines a few days ago (all were running Simon's versions for 5.15 and I had just gotten my Internet connection back after moving Home and Office) and the Windows machines have not been able to download work units at all.  This has been the case since I was able to get everything back and running this past Friday/Saturday PDT.  All machines were run dry of tasks before the move.

Linux boxes are getting 5.15 units, however.

[pepperammi]

With all due respect, folks.  5.17 is the required version of SETI for Windows and has been for a few days!  I just reconnected my machines a few days ago (all were running Simon's versions for 5.15 and I had just gotten my Internet connection back after moving Home and Office) and the Windows machines have not been able to download work units at all.  This has been the case since I was able to get everything back and running this past Friday/Saturday PDT.  All machines were run dry of tasks before the move.

Linux boxes are getting 5.15 units, however.

With all due respect again, I'm still getting plenty of 5.15 WUs to crunch. I think 5.17 is still beta and it would probly be very wise to avoid releasing any beta code into the wild. That partly got one person into a lot of hassle in the past. Better to stay on the safe side.
Any new app would still new to be fully tested not only in beta area but any newly compiled optimised versions need to be tested fully here too before they're sure to be ready for general release.

By the way. Last I heard 5.17 was/is slower than 5.15. Generally because the aamount of new code to handle the new ALFA reciever data.

[Simon]

JavaPersona,

respectfully, respect has nothing to do with the current issue you're experiencing

Rather, it's a problem various people experienced - when the 5.17 app release went onto Beta (BETA, not the main S@H project), anyone who is crunching both Beta and S@H with an optimized app ran into trouble (unless they already had an entry for 5.17 in their app_info.xml).

This is an unfortunate problem for which none of us are responsible, but that nonetheless is rather annoying.

The fix is quick and painless: change "514" to "517" in the app_info.xml, restart BOINC, done.

HTH,
Simon.

[JavaPersona]

Alright, fix applied. I am now crunching 5.17 tasks in SETI (thats what the WU's say they are )

It just seemed to me that the posts here were saying that 5.17 was not yet released to SETI.  My Windows machines were looking for the 5.17 version for SETI (BETA was already crunching those) and would not d/l any work becuase I had "the wrong" version of the app.  Now that the fix you mentioned is applied I have d/led some 5.17 task for SETI.

The impression I got from:

When 5.17 is rolled out to mainstream Seti, will they go through a transition period (like the transition to enhanced) where the scheduler will still parse out 5.15 wu's to users who are not upgraded yet, or will they just force the Seti client to update to 5.17?  And if they force the update, will I come home from work some day to find all of my 7 crunchers merrily running pell mell through the wu's generating errors ('cuz they are all on Chicken apps)?  Or will Simon (I guess I don't know if he prefers to be referred to as Simon or the notorious Chicken) have his apps upgraded to 5.17-ready in advance to we can all upgrade before they start coming out?  I know, a lot of questions.

Was that 5.17 had not yet been released to SETI.  It had been.  With Simon's (Chicken's app) there was no pell melling going on.  Just not d/ling at all.