Yes, optimized apps have been abandoned. At least, no one has not tell, will they ever make optimized SETI@home 5.21.

I am not going to use "Windows SSE3-P4/PD 32-bit based on S@H V5.15 'Noo? No - Ni!' Revision: R-2.2B|xP|FFT:IPP_SSE3|Ben-Joe" anymore, because it is much older than SETI@home 5.21. It will probably ruin the MultiBeam-WUs or start a credit war. Oh, I just hate those credit issues...

I just want to crunch scientifically correct results and take everything out of my CPU (optimize).

PS. I do not want to overclock my CPU.

---

Manned mission to Mars in 2019 Petition <-- Sign this, please.

Yes, optimized apps have been abandoned. At least, no one has not tell, will they ever make optimized SETI@home 5.21.

I am not going to use "Windows SSE3-P4/PD 32-bit based on S@H V5.15 'Noo? No - Ni!' Revision: R-2.2B|xP|FFT:IPP_SSE3|Ben-Joe" anymore, because it is much older than SETI@home 5.21. It will probably ruin the MultiBeam-WUs or start a credit war. Oh, I just hate those credit issues...

I just want to crunch scientifically correct results and take everything out of my CPU (optimize).

PS. I do not want to overclock my CPU.

Henri, the 5.21 apps ARE OPTIMIZED. They will be arriving in SETI main very soon, possibly in less than a week. The 5.20 and 5.21 apps now in SETI Beta are at least 20% quicker than the 5.17 MultiBeam app.

I am sure Joe or Simon or one of the other Optimizers can tell you more.

Regards, Keith

[EDIT]Changed probably to possibly[/edit]

Yes, optimized apps have been abandoned. At least, no one has not tell, will they ever make optimized SETI@home 5.21.

I am not going to use "Windows SSE3-P4/PD 32-bit based on S@H V5.15 'Noo? No - Ni!' Revision: R-2.2B|xP|FFT:IPP_SSE3|Ben-Joe" anymore, because it is much older than SETI@home 5.21. It will probably ruin the MultiBeam-WUs or start a credit war. Oh, I just hate those credit issues...

I just want to crunch scientifically correct results and take everything out of my CPU (optimize).

PS. I do not want to overclock my CPU.

Nobody has abandoned optimised apps.

You don't have to stop using Chicken 2.2B right now. When multibeam is released to Seti Main, yes, then you would have to stop using 2.2B.

When multibeam is released here, Simon et al will eventually release an optimised app here for multibeam.

I think you've misinterpreted Joe's post to you over on Beta.

It is good to be enthusiastic and to want to contribute as much as you can while keeping scientific integrity, however it is also important to have patience.

Everything doesn't happen overnight....;-)

---

If you use app_info.xml with V5.15, then you don't get the multibeam-WUs (V5.2x), or?

So my little 'trick', if you can't get new work, you must look to an other optimized app. :-)

That's correct, or I'm wrong?

---

If you use app_info.xml with V5.15, then you don't get the multibeam-WUs (V5.2x), or?

So my little 'trick', if you can't get new work, you must look to an other optimized app. :-)

That's correct, or I'm wrong?

When multibeam is released here I'm sure there will be an announcement on the front page.
Then you'll have to remove the app_info to get new multibeam work.

When the optimizers get to work and release cpu specific apps then you can go back to crunching multibeam with the new optimized app.

---

Er,

who told you anyone abandoned anything? Please be sure your sources are good before you post such ill-informed things.

Firstly, we haven't been idle. As we've told anyone who asked, as soon as Multibeam hits we'll have an app ready.

Well, we do. Have for some weeks, in fact. Even 2.2B can correctly process MB WUs, or, for that matter, old enhanced apps before 5.15.

This app has not been released because - along with Multibeam compatibility - the new 5.21 apps have a higher credit multiplier (3.81 vs. 3.35). This is the real reason we've not wanted to release them before the project moves to MB apps on S@H main.

Basically, I check a checkbox, and it's publicly accessible, simple as that.

Also, as others noted, 5.21 has several pieces of the Lunatics.at optimized apps in it, as well as some of Alex Kan's work. This manages to give it a respectable speedup - even when not using specialized compilers like the Lunatics.at apps but GCC.

HTH,
Simon.

P.S.: Renaming this thread would be nice, Henri - thanks.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Ok. Sorry. I just hate the idea that old versions can cause credit wars and ruined science.

I am trying to be patience...

(Maybe an moderator can change the title from "Optimized apps have been abandoned
" to "Have optimized apps been abandoned?".)
EDIT: Title changed by myself.

---

Manned mission to Mars in 2019 Petition <-- Sign this, please.

What happens at the very moment when they will put the stock 5.21 app online? Thousands of people will still continue using optimized 5.15 apps many hours/days/weeks. Is this OK?

---

Manned mission to Mars in 2019 Petition <-- Sign this, please.

Actually, you'd stop getting work when the last of the WU's split for straight E-SAH were used up, due to the change in version number (assuming you don't tweak the app_info file). If you were running stock you wouldn't notice anything from a work flow POV.

Alinator

And if you tweaked the app info file, you I believe you could continure with the 2.2b app, but without the increased credit muliplier.
The stock apps will auto update if I am not mistaken.
Chicken users will have to fetch the new pot of Chicken soup to put on the stove.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Actually, you'd stop getting work when the last of the WU's split for straight E-SAH were used up, due to the change in version number (assuming you don't tweak the app_info file). If you were running stock you wouldn't notice anything from a work flow POV.

Alinator

I wonder will it be a transition like when enhanced began where both type of wu were avaliable or will they run the cach dry and then turn on MB.
would be nice to know.

---

Actually, you'd stop getting work when the last of the WU's split for straight E-SAH were used up, due to the change in version number (assuming you don't tweak the app_info file). If you were running stock you wouldn't notice anything from a work flow POV.

Alinator

I wonder will it be a transition like when enhanced began where both type of wu were avaliable or will they run the cach dry and then turn on MB.
would be nice to know.

I would think that they should still be splitting some old WU for a short time. I'm certain that they still have several tapes (especially some of the very early ones) that could benefit from a closer second look with the enhanced app. But then that's a benefit vs cost (in complaints and time) decision that the devs will have to make.

---

Boinc V7.2.42
Win7 i5 3.33G 4GB, GTX470

My guess would be they would make a limited rollout of MB on the main project first while there was still plain vanilla E-SAH available, just in case there are problems which went undiscovered on Beta.

That way if it turned out the problem was a 'show stopper', it wouldn't bring the project to a screeching halt. That's a scary thought! <shudder> :-)

Alinator

I'll add a few notes to supplement what Simon posted.

Although the project could limit issue of MB WUs to 5.2x applications, I do not think they will. All 5.x setiathome_enhanced applications should provide scientifically correct results for both Line feed and Multibeam WUs, what the app does is controlled by parameters in the WU header.

The stock 5.2x application should claim about the same credit/time doing Multibeam work as the stock 5.15 app does while doing Line feed work. There's insufficient data from SETI Beta to know how close the match will be, but it seems to be in the right range to not push this project's overall credit/hour badly out of line with other projects during the transition. Because optimized apps will have less speed advantage compared to stock 5.2x, optimized credit/time will probably drop.

Those running 2.2B Chicken apps will probably continue to get work and report it with the old lower fpops multiplier. We hope most will be paying attention and upgrade promptly, but there will almost certainly be some who won't. The quorum of 2 will continue to make the credit grant match the lower claim, all we can do is attempt to encourage the transition.

In any case, there will certainly be a mix of Line feed and Multibeam results being reported while host queues drain, and lingering Line feed WUs which have to be reissued, etc. Reissues will probably be crunched with 5.2x and get higher claims, the mismatched claims will continue for a month or so.

                                                                 Joe

Wait a second...

Where I agree that theoretically the 5.15 and 5.2x should be able to process the MB data, won't the work have to be differentiated by version at split time?

I mean how else would BOINC know it should pick up the new app when running stock. Therefore, unless you have tweaked the app_info to reference 5.2x to the current Coop app you wouldn't DL the new MB work.

I suppose the proper way for running optimized would be to reference the E-SAH work to 2.2B, and MB work to whatever the name is for the new Coop app.

Or am I missing something here?

Alinator

Wait a second...

Where I agree that theoretically the 5.15 and 5.2x should be able to process the MB data, won't the work have to be differentiated by version at split time?

I mean how else would BOINC know it should pick up the new app when running stock. Therefore, unless you have tweaked the app_info to reference 5.2x to the current Coop app you wouldn't DL the new MB work.

Good question. That doesn't make much sense to me either.

---

Wait a second...

Where I agree that theoretically the 5.15 and 5.2x should be able to process the MB data, won't the work have to be differentiated by version at split time?

5.15 is able to process MB data.It's allready tested.

I mean how else would BOINC know it should pick up the new app when running stock. Therefore, unless you have tweaked the app_info to reference 5.2x to the current Coop app you wouldn't DL the new MB work.

Using the app_info.xml overrides any references to version numbers and forces your boinc client to use whatever the app is called in app_info.xml.

The "only" important part is to have "<name>setiathome_enhanced</name>" in the app_info.xml to get MB or line Feed WUs.

I suppose the proper way for running optimized would be to reference the E-SAH work to 2.2B, and MB work to whatever the name is for the new Coop app.

the name is 2.4 ... ;)

---

Join BOINC United now!

Wait a second...

Where I agree that theoretically the 5.15 and 5.2x should be able to process the MB data, won't the work have to be differentiated by version at split time?

I mean how else would BOINC know it should pick up the new app when running stock. Therefore, unless you have tweaked the app_info to reference 5.2x to the current Coop app you wouldn't DL the new MB work.

I suppose the proper way for running optimized would be to reference the E-SAH work to 2.2B, and MB work to whatever the name is for the new Coop app.

Or am I missing something here?

Alinator

When a host running stock requests work, the reply lists both work to download and the highest numbered app the Scheduler knows about for the platform and application name (setiathome_enhanced). If the host doesn't yet have that app version, the core client downloads it as well as the work. Once 5.2x is available here it will be used for any new work issued to hosts running stock, whether Multibeam or Line feed.

If multibeam processing had needed logic changes, the project would have named the app setiathome_multibeam or similar and the work would be tagged for that. But that wasn't needed, so it's just a version upgrade.

                                                                Joe

Ah OK, I was under the impression it was going to have a new app name.

Although I don't recall where I got that idea, probably just misinterpreted something I read in one of the fora. ;-)

Wait a second...

Where I agree that theoretically the 5.15 and 5.2x should be able to process the MB data, won't the work have to be differentiated by version at split time?

5.15 is able to process MB data.It's allready tested.

I mean how else would BOINC know it should pick up the new app when running stock. Therefore, unless you have tweaked the app_info to reference 5.2x to the current Coop app you wouldn't DL the new MB work.

Using the app_info.xml overrides any references to version numbers and forces your boinc client to use whatever the app is called in app_info.xml.

The "only" important part is to have "<name>setiathome_enhanced</name>" in the app_info.xml to get MB or line Feed WUs.

I suppose the proper way for running optimized would be to reference the E-SAH work to 2.2B, and MB work to whatever the name is for the new Coop app.

the name is 2.4 ... ;)

But don't you still have to add the (for example):

.
.
.
<version_num>521</version_num> <--- Whatever the final release number is.
.
.
.

entry for the new app?

Alinator