I live in central Florida, the thunderstorm capital of the United States and do not have an UPS. As a result when Progress Energy has a problem (becoming much more frequent) when the PC shuts down, the work unit currently running is invariably corrupted. I hsve changed the number of times per day that my PC attempts to contact Seti@home, but still get WUs of about the same size (100+ hrs.). As of late I have had to abort more tham I have submitted and eould really like to get some work done. I figure that with smaller WUs I stand a better chance of finishing them. Any help will be deeply appreciated.

---

Hmm.. that's a little odd.

To answer your question, no there is no way to request specific size WUs. They are (mostly) randomly choosen and sent to clients to crunch and return. If there were a way to only choose the smaller ones, those that like to increase their RAC or their credit would only choose the smaller ones so that they could "chew" through more of them at a time.

Also, the workunits should not become corrupted, even after a power outage. I've had a few recently here that has brought my entire mini-farm down and once power is restored they all pick up where they left off.

See, every minute (or so, depending on preferences, but the default it a minute) data is written to a result file and the data is continued to be crunched on. Worse case scenario, you only lose up to one minute of crunch (since the last update) and your CPU would have to re-crunch that minute. Otherwise, it shouldn't hurt the workunit at all.

Are you sure it's starting over every time?

---

I would also update from the BOINC 4.45 version that you are using. The new 5.4.9 has a built in security that checks the client_state.xml file upon each restart. If found corrupt, it picks the backup file, thereby making sure your results continue to be crunched. 4.45 doesn't have this option, so if at that time of the power-outage you were writing to client_state.xml, you're bound to get a corrupt CS file and thus corrupt results.

---

Well I've let a number of them continue to run until finished in the normal manner and they appear to be OK until you upload them. Then I've gotten a "computational error" message and no credit. This is just as if you had aborted the WU. When I first started back with the program again several monthes ago, the WUs were small 40 - 70 hrs. Since then however I have changed virtually all of the performance related parameters in preferences to get the most out of this machine, which is ancient a Pentium II, and is used little except for emails etc., so I'm not trying to really go after the credits I would do that at work were my workstation has duel xenon pocessors and really screams. I guess maybe my next step should be to try to restore these to the defaults and see if that helps. Is there a way to get in touch with the system administrator? Maybe he/she/they will have an answer. This is really becoming annoying too, I had to abort 4 yesterday and being that its Florida there is no real way to predict when this might happen. For example this morning we had a brown-out as a result of a storm at least 8 - 10 miles away. It was just enough of a drop to get the PC and a TV that was on - nothing else though.

---

When I first started back with the program again several monthes ago, the WUs were small 40 - 70 hrs.

Yes, that was before SETI@Home Enhanced was released. Most computers were running 5x faster, but not looking nearly as hard for the right info in each workunit. Since then, SETI@Home Enhanced was released, which allows CPUs to look much deeper into each workunit, which in turn increases our chances of finding E.T.I. but the downside is that it takes much longer to complete a workunit.

No mannor of tweaking your system (short of download an optimized client) will speed this up, of course, short of a system upgrade.

Is there a way to get in touch with the system administrator? Maybe he/she/they will have an answer.

What answer? They will not give you a different one that we have. There is no way to specify what size workunit you receive, nor will this change due to people out there that would love to gain any advantage they could just to pad their results, regardless of others (such as yourself) good intentions. I'm sure you're not in it for credits, but if it became an option, it would be abused by those that are only in it for credits.

---

Well I've let a number of them continue to run until finished in the normal manner and they appear to be OK until you upload them. Then I've gotten a "computational error" message and no credit. This is just as if you had aborted the WU. When I first started back with the program again several monthes ago, the WUs were small 40 - 70 hrs. Since then however I have changed virtually all of the performance related parameters in preferences to get the most out of this machine, which is ancient a Pentium II, and is used little except for emails etc., so I'm not trying to really go after the credits I would do that at work were my workstation has duel xenon pocessors and really screams. I guess maybe my next step should be to try to restore these to the defaults and see if that helps. Is there a way to get in touch with the system administrator? Maybe he/she/they will have an answer. This is really becoming annoying too, I had to abort 4 yesterday and being that its Florida there is no real way to predict when this might happen. For example this morning we had a brown-out as a result of a storm at least 8 - 10 miles away. It was just enough of a drop to get the PC and a TV that was on - nothing else though.

The problem is you are crunching on a 200mhz computer and Seti is deeply analyzing the units. The combination of those two factors makes for long units for everyone. I used to be able to do units on my computer is less than an hour, now I am up to around 2 3/4 hours per unit. I am using a 3.2ghz computer, or said another way a 3200mhz computer.
You might want to consider getting a UPS, Uninteruptable Power Supply, basically a battery pack you plug your comput into and then it into the wall. Then when the power goes out you will have time to either weather it out, depending on how long your power is out and the size of the UPS you buy, or at least time to turn the pc off so you don't lose what it is doing. You can UPS's at the local CompUSA, Best Buy, Circuit City, WalMart, etc. Just don't get too small of a one.

---

Hmm.. that's a little odd.

To answer your question, no there is no way to request specific size WUs. They are (mostly) randomly choosen and sent to clients to crunch and return. If there were a way to only choose the smaller ones, those that like to increase their RAC or their credit would only choose the smaller ones so that they could "chew" through more of them at a time.

Also, the workunits should not become corrupted, even after a power outage. I've had a few recently here that has brought my entire mini-farm down and once power is restored they all pick up where they left off.

See, every minute (or so, depending on preferences, but the default it a minute) data is written to a result file and the data is continued to be crunched on. Worse case scenario, you only lose up to one minute of crunch (since the last update) and your CPU would have to re-crunch that minute. Otherwise, it shouldn't hurt the workunit at all.

Are you sure it's starting over every time?

But what is when the shutdown, comes exactly in the moment where BOINC trys to write the result file, i mean when BOINC had allready started to writing but had not finished it until the shutdown? I think it could be possilble that the file is corrupted then, or do you not think so?

---

* Hi! I'm a signature virus! Copy me into your ~/.signature to help me spread! *

But what is when the shutdown, comes exactly in the moment where BOINC trys to write the result file, i mean when BOINC had allready started to writing but had not finished it until the shutdown? I think it could be possilble that the file is corrupted then, or do you not think so?

With BOINC 5.4.x onwards, BOINC has a safe-guard against this: If the client_state.xml file is corrupt due to whatever reason, the backup file (client_state_prev.xml) is used. This ensures that you have a complete and readable file at all times.

---

But what is when the shutdown, comes exactly in the moment where BOINC trys to write the result file, i mean when BOINC had allready started to writing but had not finished it until the shutdown? I think it could be possilble that the file is corrupted then, or do you not think so?

If you run the software that is supplied with the UPS, it will do a clean shutdown on your system rather than a "hard" shutdown (which is basically like pulling the power cord.).