At last what does he mean by this "NOTE: Signal counting only works properly if the application stays in memory." and how do I make it so and what happens when I turn of my computer.

The cute feature which provides text to you describing a count of signals of various flavors found the WU only works if the ap stays in memory from start to finish. But the returned result for the science (and credit) is OK regardless.

---

Is there an app_info floating around somewhere ? Windows

Thanks

I´m just looking for a good one too... I´m tired of :

09/05/2006 00:25:27|SETI@home|Started download of setiathome_5.12_windows_intelx86.exe
09/05/2006 00:25:27|SETI@home|MD5 check failed for 26fe99aa.5731.18848.540910.3.247
09/05/2006 00:25:27|SETI@home|expected 17da887445568f52d14589a27392b2cb, got b6cebddabde30cd57ba43af46953e101

Please someone, post the right app_info for us.

Best Regards,

Crunch3r has the file on his page that includes the 4.11 replacement for 4.18 and the 5.12.

---

Fred
BOINC Alpha, BOINC Beta, LHC Alpha, Einstein Alpha

Anyway I added the optimized Linux 5.12 apps today.

Thanks! I was waiting for those!

No problem.
Just make sure to take a look at the "README_APP"

I will check the dependancies shortly. Luckily I still have 4 days worth of 4.11 cache... :)

Greetings,
Sander

---

A cuople of questions.
I had a litle misshap and had to reinstall Boinc and the Seti client so I only have the enhanced on my computer, do I need the app_info.xml or can I just overrite the original 5.12 with the one from Crunch3r.

First thing I do when changing things is making a backup of the Boinc map.

Then I rename the file to be replaced and leave it in the map so I can easely have it back, for instance whateverthenameis(org).exe and then extract the new file to the requiered map.
Yes you do need the app_info.xml as provided by Cruncher on his page asuming you also uses his standard optimised app 4.11.

Should I use the one Windows 5.12_IPP_SSE2, I have checked that my Athlon64 supports SSE2 but not SSE3.

Yes

At last what does he mean by this "NOTE: Signal counting only works properly if the application stays in memory." and how do I make it so and what happens when I turn of my computer.

This means you can see something like this in your result details:

Spike count: 4
Pulse count: 0
Triplet count: 0
Gaussian count: 0

To keep the app in memory you have to go to general prefferences on the "Your account"page and set "Leave applications in memory while preempted? to yes.
Turning of the computer does'nt metter cos it's kept in the swapfile asuming you have one as most people do.

Thx Crunch3er - much appreciated ;-)

I have both the old 4.11 and the new optimized 5.12 (and the new app_info.xml) in the same /project/../ folder.
According to BOINC Manager the hosts have started to d/l WU's for the new 5.12 - but is still cruncing the WU's for 4.11 (4.18).

So I'll have to wait a few days before I actually will see if the new 5.12 will start crunching - due to the WU-cache is set to max ?

---

Greetings from Norway

Crunch3er & AK-V8 Inside

So I'll have to wait a few days befrore I actually will see if the new 5.12 will start crunching - due to the WU-cache is set to max ?

Yep. Quite some days as far as I can see.:-)

Yep. Quite some days as far as I can see.:-)

Hehe - those old Dual P3's are still hanging in there ;-)

---

Greetings from Norway

Crunch3er & AK-V8 Inside

Your probably right, Anyway, I cleaned it out and put the SSE2 version in since my Celeron processor supports it.
Seems to be doing fine.

I have installed these from Crunch3rs site,

setiathome_4.11_windows_intelx86.exe (Pentium3_SSE)
setiathome_5.12_windows_intelx86.exe (Enhanced_5.12_IPP_AR_FPOPS_SSE)
and the new app_info.xml

however I get this error on the non enhanced wu's...

5/10/2006 7:57:04 AM|SETI@home|WU refers to nonexistent app_version: setiathome 511

I don't think you got this using the app_info.xml posted currently (and since at least 16 hours ago) on crunch3r's site. It contains no reference to "511" at all. Perhaps you have a app_info.xml from another source in your ...\\projects\\setiathome.berkeley.edu directory which you failed to overwrite? For example, I posted an app_info.xml about May 7 which contained 511 references, which will not work with the current ap and ap name.

Also, why does say like this ?

<real_cpu_time>23425</real_cpu_time>
<corrected_cpu_time>15080</corrected_cpu_time>

That looks like a calibrating BOINC client at work: are you running TruXoft's or similar? It'll have no effect on your Enhanced credit, although it could influence the amount of work you get because of its effect on the DCF calculation.

---

The Windows 5.12_IPP_SSE2 was great, it improved my time on the AMD64 3000+ with about 50%

But unforutunatly the Windows 5.12_IPP_SSE was not good at all on my AMDXP 2600+

It actually increased the time from 5h.30min to 6h.35min for a WU with the exact same AR 0.426


Also, why does say like this ?

<real_cpu_time>23425</real_cpu_time>
<corrected_cpu_time>15080</corrected_cpu_time>

It could be possible that the standard app is faster on athonXP...we've seen that behaviour on previous versions (linux) of the "normal seti".

Anyhow you cannot even compare WUs that have the same AR. You need exactly the same to do "benchmarks". (Its a thing of how manny gaussians, pulses, triplets and spikes are in that wu)

I could publish a "Quick reference WU" which was done by Joe Segure to run standalone tests of the standard and optimized app.

That's the only way to see which ones faster....
Anything else does not work.

---

Join BOINC United now!

Crunch3R: Does this look right to you? Is your 5.12 Enhanced supposed to calibrate WU times?

Computer ID 1997635
Report deadline 1 Jun 2006 2:46:17 UTC
CPU time 15189.724901
stderr out
<core_client_version>5.3.12.tx36</core_client_version>
<real_cpu_time>16215</real_cpu_time>
<corrected_cpu_time>15190</corrected_cpu_time>
<corrected_Mfpops>4248.0</corrected_Mfpops>
<stderr_txt>

Windows optimized S@H Enhanced application by Crunch3r
Version info: Windows SSE2 V5.12 by Crunch3r

Work Unit Info:
...............
WU true angle range is : 0.4415

Flopcounter: 15750304114720.074000

Spike count: 7
Pulse count: 1
Triplet count: 0
Gaussian count: 1

</stderr_txt>
Validate state Valid
Claimed credit 61.2511826683558
Granted credit 60.904383711204
application version 5.12

BTW, Thanks for the teriffic work and efforts!

---

Crunch3R: Does this look right to you? Is your 5.12 Enhanced supposed to calibrate WU times?

Computer ID 1997635
Report deadline 1 Jun 2006 2:46:17 UTC
CPU time 15189.724901
stderr out
<core_client_version>5.3.12.tx36</core_client_version>
<real_cpu_time>16215</real_cpu_time>
<corrected_cpu_time>15190</corrected_cpu_time>
<corrected_Mfpops>4248.0</corrected_Mfpops>
<stderr_txt>

Windows optimized S@H Enhanced application by Crunch3r
Version info: Windows SSE2 V5.12 by Crunch3r

Work Unit Info:
...............
WU true angle range is : 0.4415

Flopcounter: 15750304114720.074000

Spike count: 7
Pulse count: 1
Triplet count: 0
Gaussian count: 1

</stderr_txt>
Validate state Valid
Claimed credit 61.2511826683558
Granted credit 60.904383711204
application version 5.12

BTW, Thanks for the teriffic work and efforts!

---

... Is your 5.12 Enhanced supposed to calibrate WU times?

...<core_client_version>5.3.12.tx36</core_client_version>

That's your BOINC client, nothing to do with the S@h app.

---

Crystallize, gecko_r7

I am coming in late but see two things... First has been mentioned that Trux's 5.2.13tx36 will try to calibrate Enhanced here... It will mess with how many workunits taht you want cached and a couple of things like incorrect times. I have two instances where restarting with that particular version trashed 5.12 (enhanced results)... I have been trying various words in the xml file to turn it off. Those verbs like "off or none" and can not find on trux's site where he tells how to turn it off... In both cases removed BOINC and reinstalled with the standard 5.2.13. These even impacted machine running Einstein and Seti and I removed "seti"...

Second I went and found your AMD XP 2600 and when I look at the ResultID's I see that you gained about 1,000 seconds over the standard 5.12 Enhanced Application for a AR 0.42xxx result... Generally before version 5.11 in Enhanced Beta the optimized app had a 50% advantage... After 5.11 it became between 5-20% depending on the Angle Range (AR) of the WU...

Pappa

The Windows 5.12_IPP_SSE2 was great, it improved my time on the AMD64 3000+ with about 50%

But unforutunatly the Windows 5.12_IPP_SSE was not good at all on my AMDXP 2600+

It actually increased the time from 5h.30min to 6h.35min for a WU with the exact same AR 0.426

Also, why does say like this ?

<real_cpu_time>23425</real_cpu_time> Real Time
<corrected_cpu_time>15080</corrected_cpu_time>

---

Please consider a Donation to the Seti Project.

Pappa, I ran into the same issue a few days ago when switching between 5.2.13tx36 & Crunch3r's Enhanced 5.11. When I edited truxoft_prefs.xml w/ "calibrate_credit = off" it indeed turned-off and 5.2.13.tx36 did not calibrate. WU times were as expected w/ no calibration and claimed credit immediately fell to 12-14 which is what I previously saw w/ WU times @ 3200secs for M-1.6.

I reactivated calibration "calibrate_credit = all" and it operated properly again.
Using Opt. Enhanced 5.12, I have left 5.2.13tx36 calibration on as I understood from the forum that is wasn't supposed to affect Enhanced. It does appear to be calibrating my Enhanced WU times, however.
I don't have too many Seti standard WU's left, so hopefully it won't matter much longer unless they keep sending standard units out. One thing's for certain, my claimed credit is falling like a stone w/ Enhanced.
Oh, well. We're in it for the science afterall.
Cheers!

---

Pappa, I ran into the same issue a few days ago when using Trux's 5.2.13tx36 client & Crunch3r's Enhanced 5.11 ap. When I edited truxoft_prefs.xml w/ "calibrate_credit = off" it indeed turned-off and 5.2.13.tx36 did not calibrate. WU times were as expected w/ no calibration and claimed credit immediately fell to 12-14 which is what I previously saw w/ WU times @ 3200secs for M-1.6.

I reactivated calibration "calibrate_credit = all" and it operated properly again.
Using Opt. Enhanced 5.12, I have left 5.2.13tx36 calibration on as I understood from the forum that is wasn't supposed to affect Enhanced. It does appear to be calibrating my Enhanced WU times, however.
I don't have too many Seti standard WU's left, so hopefully it won't matter much longer unless they keep sending standard units out. One thing's for certain, my claimed credit is falling like a stone w/ Enhanced.
Oh, well. We're in it for the science afterall.
Cheers!

---

gecko_r7

One thing that it definately does is add "confusion" to the "reported time."

I hate to bring up the Credit Soup issue again... One of the things that is happening is that it is now taking longer for everyone (or at least enough for the quorum) to return results... If you look "pending credit" builds until enough people are using 5.12 or 5.13 to see where "credits" will end up...

While I was running about 14 machines (top) in Enhanced Beta saw my pending credit go to about 28,0000 not enough people helping... So when 5.11 hit and people cleared their large caches to get 5.11 the daily credits jumped! So those with a 10 day cache that now has a 5.12 will not see that result return for over ten days... If i crashed the we could be waiting for it to expire before it is reissued... So that said "daily credits" or "RAC" is going to take it in the shorts for a bit! As the transition completes and everyone is running 5.12 or whatever the final version is, then people will see an increase as the pending Results get cleared.... That takes time as the transition is schedule for less than a month now... Note to Eric, add another splitter about 30-40% of the machines running Enhanced will help with this issue...

So what is necessary?
* A BOINC Core Client 5.2.13 or higher (5.4.9 is getting close) That is a MUST! to be honest, I setup Trux's 5.2.13tx36 for testing with Beta Enhanced... I am now spending time figuring out what went wrong and removing it... If you are running Seti and Einstein or another project that Trux supports then turn off Seti... a month from now you will have to do that anyways...

* Reduce your Cache size in preferences... With enhanced (until things are transitioned) a Large Chace will only complicat matters... I am running with .5 days.. and have not ran out of work...

Yes I know that Jim-R mentioned about return times and what happens if someone turns of the computer and goes on vacation for a week... He did a Great Job at pulling the highlights from Months of conversation in the Beta NC Forum...

Pappa

Pappa, I ran into the same issue a few days ago when using Trux's 5.2.13tx36 client & Crunch3r's Enhanced 5.11 ap. When I edited truxoft_prefs.xml w/ "calibrate_credit = off" it indeed turned-off and 5.2.13.tx36 did not calibrate. WU times were as expected w/ no calibration and claimed credit immediately fell to 12-14 which is what I previously saw w/ WU times @ 3200secs for M-1.6.

I reactivated calibration "calibrate_credit = all" and it operated properly again.
Using Opt. Enhanced 5.12, I have left 5.2.13tx36 calibration on as I understood from the forum that is wasn't supposed to affect Enhanced. It does appear to be calibrating my Enhanced WU times, however.
I don't have too many Seti standard WU's left, so hopefully it won't matter much longer unless they keep sending standard units out. One thing's for certain, my claimed credit is falling like a stone w/ Enhanced.
Oh, well. We're in it for the science afterall.
Cheers!

---

Please consider a Donation to the Seti Project.

Pappa, Gecko_r7, et al,

You now seeing the effect I indicated would happen 3 days ago. But nobody listens to me. Message forum_thread.php?id=30570#299712

Andy

Andy

I know believe me.... Grin... I spent over 20 minutes trying to recover.. Then trashed it... It is now running 5.4.6 BOINC Core Client... Which will upgrade to 5.4.9 which is the current "Alpha" release...

The mix of 5.2.13tx36 has some issues with Seti Enhanced...

Pappa, Gecko_r7, et al,

You now seeing the effect I indicated would happen 3 days ago. But nobody listens to me. Message forum_thread.php?id=30570#299712

Andy

Al

---

Please consider a Donation to the Seti Project.

Yes, Andy, I read your comments and your point is well taken. There have been differing opinions ("yes" it does, "no" it doesn't) and results, but not enough people running the same combos w/ clean installs, in the same threads at the same time. Looks like we're finally getting some clustering of issues w/ similar results. You indeed called it correctly.

5.4.6 core works w/ S. Enhanced w/o these issues?
Regards,
Ian

---