This new 4.02 application takes a long time to do it's job. I sure hope there were some significant changes made to justify the increase in crunching time. I used to crunch the v4.00 app in 3 hours 15 minutes. Tetsuji's optimized beta version only took 1 hour 30 minutes. I've been crunching with this new v4.02 for an hour and a half and I'm only at 5% complete with another 30 hours estimated to go!

What was changed to account for the drastic increase in crunch time?

Pentium M 1.8

---

Sounds like a full debug build, but I don't know for sure.

---

BOINC WIKI

BOINCing since 2002/12/8

I have 2 WU's of 4.02 that are taking >20 hours according to BOINC View. But are these really 4.02 or reissue 4.00 with increased measurements? Look here. http://setiweb.ssl.berkeley.edu/beta/workunit.php?wuid=49761 WU first issued 21 June as application 4.00, now we are getting them as 4.02?! How are these going to validate?

---

See this post: http://setiweb.ssl.berkeley.edu/beta/forum_thread.php?id=4#111

Per Eric Korpela, a codeing error caused v4.00 to skip some of the additional science. Run times will likely be somewhat longer, now that that bug is fixed in v4.02.

---

I finally completed a 4.02 wu : 123,185.25: Cpu (sec) Claimed credit: 323.01

32 hours to crunch, compared to 5+ hours with the 4.00 app.

Whew! That's one down Rampage! ;)

Either the new code to dynamically track and adjust the competion times for WU's on each host needs to get implemented quickly, or the estimated number of calculations for SETI@home WU's need to be significantly readjusted! This looks like a lot more than a 40% increase in standard crunch times. My client is downloading way too much beta work since it thinks it can complete it much faster than it really can.

I guess that the Classic crunchers aren't the only ones who will have some change to deal with. Can imagine the reaction from the general BOINC SETI@home community when this is rolled out to them?

---

Not sure why all your computers are taking longer. Mine took less time on a 4.02 unit: See here. Using standard 4.02, not TMR's version.

---

Jord, your WU took what is expected of the new 4.02. However if you look at the one Rampage finished and the one I listed, they are WU's that were previously distributed when there was the bug ~20 June. For some reason those WU's are taking much longer and I have suspended mine at the moment to download new ones and see how long they take to crunch.

If they are being compared to a WU that was returned before the bug fix, how are they going to validate against each other? 40 credit against ~320 for the same WU? I am not that concerned about credit, but that is extreme.

Here are my 2 WU's:

http://setiweb.ssl.berkeley.edu/beta/workunit.php?wuid=49761
http://setiweb.ssl.berkeley.edu/beta/workunit.php?wuid=49734

---

Checking my latest unit against what it did on other systems, it doesn't look good. Unless it only borks out on ancient Intels (due to the missing SSE instructions?). Fingers crossed.

But the ~20th of June bug was an interesting one. I managed to run all those units without a hitch.

---

No it does not, JM7 and the other both using 4.70 client aa are you, your other is still using 4.19 client and has returned all the 4.02 with noisy ~400 seconds WU's.

I have downloaded new WU's and they are crunching normal about 4 hours on standard app. I may abort the 30 hours WU's, will wait and see if Rampage's WU validates.

---

No it does not, JM7 and the other both using 4.70 client aa are you, your other is still using 4.19 client and has returned all the 4.02 with noisy ~400 seconds WU's.

Interesting...

I've completed two WUs successfully using the 4.70 CC, yet running Tetsuji's optimized 4.02. Is there something in the public code that isn't in Tetsuji's that's causing the problem?

<a href="http://setiweb.ssl.berkeley.edu/beta/workunit.php?wuid=99618">Another interesting WU</a>, mine machine ran 4,627.92 seconds and completed successfully, yet the other two machines (one running 4.68, the other 4.45) reported back a -9 overflow. And we all were validated? How does that happen?

Jord, your WU took what is expected of the new 4.02. However if you look at the one Rampage finished and the one I listed, they are WU's that were previously distributed when there was the bug ~20 June. For some reason those WU's are taking much longer and I have suspended mine at the moment to download new ones and see how long they take to crunch.

I think you're on to something, Brian. I aborted my WU's that were initially sent out June 21 and were showing an abnormally long estimated time to complete. When I moved on to a fresh v4.02 WU from July 6, the time to complete dropped back down to a level similiar to what I was crunching before, perhaps a little bit longer.

---

I've completed two WUs successfully using the 4.70 CC, yet running Tetsuji's optimized 4.02. Is there something in the public code that isn't in Tetsuji's that's causing the problem?

Look at <a href="http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=604">this host</a>. Every WU they have tried to crunch today using 4.70 has errored out, all 241 of them. Same for JM7. They are also using the standard app.

The WU's I downloaded with normal time to crunch, crunched fine in about 3.8 hours with standard app. I will be aborting the 30 hour WU's and running Tetsuji's optimized app. Already using it on Seti II.

Using CC4.70 has an extra downside, I found. Seti has put me on a 1 unit/day diet. Beta here has now put me on a 2 unit/day diet.

I'll keep running it for today, if only to help figure out if it's a continuing problem, or that the daily quota will go up again.

---

Using CC4.70 has an extra downside, I found. Seti has put me on a 1 unit/day diet. Beta here has now put me on a 2 unit/day diet.

Another interesting phenomena!

I just checked my quota here at beta, and found I'm at 1 WU a day. I've only had a single WU error out, so it's not due to the normal reasons. I checked over at Seti, and I still have 100 a day over there.

Odd...

Can you post that behaviour on the Boinc Developer's email list, just so they don't think I'm going (too) crazy? ;)

---

Can you post that behaviour on the Boinc Developer's email list, just so they don't think I'm going (too) crazy? ;)

It's way to late to change that! ;)

---

Can you post that behaviour on the Boinc Developer's email list, just so they don't think I'm going (too) crazy? ;)

Was just going to do that, but noticed you had taken care of it. :)

Running fine for me with the standard 4.02 (not the optimized version that i will try later)
- faster than the 4.00 :
before 14xxx ; now 107xx to 112xx
- Cpu : Barton 2500+
- CC : 4.32

---