Ooooooh - this is going to be verrrry interesting.

Just noticed that I have snagged a couple of AP WU's and the time to completion for them shown in Boinc Manager is varying between 8hrs 45mins and about 10hours as MB WU's are crunched. My system already goes into EDF panic when it hits a 100 cred MB WU, at which point it suspends everything (including the one CPDN that is crunching) and munches on shorties. I'm really looking forward to seeing what happens when a WU takes 6x longer to crunch than it expects rather than just 10% longer. However, because of my cache I'll have to wait a couple of days for the event.

F.

---

Hi, Fred - I haven't seen any AP units yet, so I am curious to know what they look like. I mean, what do you see to identify them as AP? Different WU name? Different app name? Or just longer crunch times? Thanks.
Whit

Hi, Fred - I haven't seen any AP units yet, so I am curious to know what they look like. I mean, what do you see to identify them as AP? Different WU name? Different app name? Or just longer crunch times? Thanks.
Whit

Well, from another message, it looks like AstroPulse WU's may be prefixed with an "ap_" Also, when looking in Boinc Manager, the "Application" listed should show the AstroPulse app rather than Setiathome_enhanced.

These are both just guesses on my part though, as my installation still has yet to grab an AstroPulse WU. :-( I want to run some AstroPulse to get my CPU temps lowered! :-)

Hi, Fred - I haven't seen any AP units yet, so I am curious to know what they look like. I mean, what do you see to identify them as AP? Different WU name? Different app name? Or just longer crunch times? Thanks.
Whit

Well, from another message, it looks like AstroPulse WU's may be prefixed with an "ap_" Also, when looking in Boinc Manager, the "Application" listed should show the AstroPulse app rather than Setiathome_enhanced.

These are both just guesses on my part though, as my installation still has yet to grab an AstroPulse WU. :-( I want to run some AstroPulse to get my CPU temps lowered! :-)

Very good guesses, though, because that is exactly what I see on my slowest (test) machine which picked up an astropulse unit yesterday ;-)

Regards,
John.

From the server performance recently, it looks like the community may be taking a few steps back for a leap forward (if it is a leap). Can anyone report how many AP units had actually been crunched by beta before they released the transition on the rest of us using the production servers/project? Or are expected to be happy guinea pigs.

Looks like they are giving you a month to crunch it as well.

---

From the server performance recently, it looks like the community may be taking a few steps back for a leap forward (if it is a leap). Can anyone report how many AP units had actually been crunched by beta before they released the transition on the rest of us using the production servers/project? Or are expected to be happy guinea pigs.

Why not visit the Astropulse section of the Beta message board and read for yourself? The app itself seems stable: what, by definition, can't be tested in advance is how well the server-side backend processes scale up to the throughput required to service the hungriest BOINC project on the planet.

For what it's worth, a single AP work generator seems to take about 2.5 hours to generate 400 WUs: nearly 40 seconds per WU, during which 8MB of data is written (probably over a network) to a storage location from which it can be downloaded. The BOINC database access will be trivial compared with the MB splitter stuff, but I guess the storage area may have to be fine-tuned for the new mix of roles. Something seems to be holding up the MB splitters, and that's the best suggestion I can make.

Hi, Fred - I haven't seen any AP units yet, so I am curious to know what they look like. I mean, what do you see to identify them as AP? Different WU name? Different app name? Or just longer crunch times? Thanks.
Whit

In Boinc Manager the Application is identified as "Astropulse 4.35" instead of "setiathome_.........."

F.

---

Something seems to be holding up the MB splitters, and that's the best suggestion I can make.

An additional suggestion may be that the 3 ap_splitters used to be MB splitters. The result creation rate is certainly jumping about though.

Back to topic. Congratulations Fred on getting your 2 AP WUs.

Hit by one AP!
After 30 minutes, progress bar shows ETA in about 13 hrs.
The best news is that app_info.xml and the project files are just fine - taken from here (I am using Alex Kan's optimized apps on Q6600). Big thank you, Richard!
Everything's under control so far. Shau'ma mal...

---

This is what I see when SETI is trying to send my MACs an AP WU (ratm I don't want one either ;) ) :

[error]State file error: missing application astropulse
[error]Can't handle task ap_1234.wu in scheduler reply
[error]State file error: missing task ap_1234.wu

and so on...

From the server performance recently, it looks like the community may be taking a few steps back for a leap forward (if it is a leap). Can anyone report how many AP units had actually been crunched by beta before they released the transition on the rest of us using the production servers/project? Or are expected to be happy guinea pigs.

Why not visit the Astropulse section of the Beta message board and read for yourself? The app itself seems stable: what, by definition, can't be tested in advance is how well the server-side backend processes scale up to the throughput required to service the hungriest BOINC project on the planet.

That is a rude reply to an innocent question, if you ask me.

Looking at your link I see a lot of complaining, so I don't think it is representative. But, from your response and not unexpectedly because SETI is a sort of hobby for Berkeley, I kind of doubt there is a proper plan for this transition. Too bad. Yet, I guess the servers will make it through the weekend and Einstein's repair job.

From the server performance recently, it looks like the community may be taking a few steps back for a leap forward (if it is a leap). Can anyone report how many AP units had actually been crunched by beta before they released the transition on the rest of us using the production servers/project? Or are expected to be happy guinea pigs.

Why not visit the Astropulse section of the Beta message board and read for yourself? The app itself seems stable: what, by definition, can't be tested in advance is how well the server-side backend processes scale up to the throughput required to service the hungriest BOINC project on the planet.

That is a rude reply to an innocent question, if you ask me.

Looking at your link I see a lot of complaining, so I don't think it is representative. But, from your response and not unexpectedly because SETI is a sort of hobby for Berkeley, I kind of doubt there is a proper plan for this transition. Too bad. Yet, I guess the servers will make it through the weekend and Einstein's repair job.

I cannot myself see anything rude in these replies. Simply a surgestion to see what we do over here

You are of course quite welcome to come and join us and help test them for yourself if you think we aren't up to your satifaction. After all it is only a hobby for most of us.

Michael.

But, from your response and not unexpectedly because SETI is a sort of hobby for Berkeley, I kind of doubt there is a proper plan for this transition. Too bad.

It's my hobby too, so who's complaining? Come and give it to uszz, our precious ;-)

just got an Astro WU on one of my machines... showing 237 hours till completion.. good grief =)

PhonAcg

Currently the population of Seti Beta Testers is less than 1% of those that run Seti. At it highest it was 1.5% when Enhanced and MultiBeam hit. The crossection of machine types to test on is less. So when people talk about problems that are or are expected, the idea is to solve as many as possible before it gets released into the wild.
So if at times it seems people are complaining it is because they are talking about problems that most would not like to see here.

Myself and Others as there are too many others to list (without embarassing someone) makes sure that we look to see what a user is talking about and feedback gets to Josh and Eric.

Regards

From the server performance recently, it looks like the community may be taking a few steps back for a leap forward (if it is a leap). Can anyone report how many AP units had actually been crunched by beta before they released the transition on the rest of us using the production servers/project? Or are expected to be happy guinea pigs.

Why not visit the Astropulse section of the Beta message board and read for yourself? The app itself seems stable: what, by definition, can't be tested in advance is how well the server-side backend processes scale up to the throughput required to service the hungriest BOINC project on the planet.

That is a rude reply to an innocent question, if you ask me.

Looking at your link I see a lot of complaining, so I don't think it is representative. But, from your response and not unexpectedly because SETI is a sort of hobby for Berkeley, I kind of doubt there is a proper plan for this transition. Too bad. Yet, I guess the servers will make it through the weekend and Einstein's repair job.

---

Please consider a Donation to the Seti Project.

I have one AP unit running on a AMD 9850 (4 processor running at 2.5G).
The time to complete started at 39 hours. After running 27 hours the total time now be over 70 hours.
Just how big is this file?

---

I have one AP unit running on a AMD 9850 (4 processor running at 2.5G).
The time to complete started at 39 hours. After running 27 hours the total time now be over 70 hours.
Just how big is this file?

Better question, howmuch calculation time or GFOLPS, would they take to complete.
It's completely different from S@H WU's.
I've 1 running on a QX9650 @ 3.25 GHz. Takes 20-30 hours to complete, at least, this one.
Keep On Crunch1ng

---

I have one AP unit running on a AMD 9850 (4 processor running at 2.5G).
The time to complete started at 39 hours. After running 27 hours the total time now be over 70 hours.
Just how big is this file?

Going from the performance of other AMD cpu's during testing before release here, 70 hrs would seem about right.

I noticed one of my 3 Ghz Pentium D boxes has got one. It's about 20 hours into it with about 80 hours left to go.

With Astropulse is Seti getting on Einstein@Home's turf? Maybe some sort of partnership should be formed.

---