Rosetta is the ONLY project for BOINC, with which I NEVER get credited of any work.

There is unambiguously a severe bug in Rosetta, that causes to fail when it is suspenped (either because of CPU scheduling by BOINC, or bysuspending the workmanually or when logging out).

Here is an example log taken from BOINC:

18/02/2006 17:24:57|rosetta@home|Restarting result PRODUCTION_ABINITIO_QUADRUPLELONGRANGEANTIPARALLEL_1vls__311_31_1 using rosetta version 481
18/02/2006 17:29:57|rosetta@home|Pausing result PRODUCTION_ABINITIO_QUADRUPLELONGRANGEANTIPARALLEL_1vls__311_31_1 (removed from memory)
18/02/2006 17:29:58|rosetta@home|Unrecoverable error for result PRODUCTION_ABINITIO_QUADRUPLELONGRANGEANTIPARALLEL_1vls__311_31_1 ( - exit code -1073741819 (0xc0000005))
18/02/2006 17:29:58||request_reschedule_cpus: process exited
18/02/2006 17:29:58|rosetta@home|Computation for result PRODUCTION_ABINITIO_QUADRUPLELONGRANGEANTIPARALLEL_1vls__311_31_1 finished
18/02/2006 17:31:01|rosetta@home|Sending scheduler request to https://boinc.bakerlab.org/rosetta_cgi/cgi
18/02/2006 17:31:01|rosetta@home|Reason: To report results
18/02/2006 17:31:01|rosetta@home|Reporting 1 results
18/02/2006 17:31:06|rosetta@home|Scheduler request to https://boinc.bakerlab.org/rosetta_cgi/cgi succeeded

My PC is a notebook with AMD Turion 64ML-30, definitely not overclocked, and running perfectly for other projects.

The only reason why this happens is definitely a bug in rosetta's code that CAN'T handle properly job suspensions, a REQUIRED feature for any BOINC client.

Already more than 15 jobs performed, all with this "unrecoverable" error, and NO credits, despite many hours have been spent on the project. All I can do now is to retire from this project, and choose other more useful BOINC projects, because the way Rosetta works just wastes my CPU time.

Such bug implies lackof quality in the Rosetta software, and Ican't exposemy computer to such constant bugs because of a poor quality software I can't trust on my machine.

I have tried to run Rosetta on another PC, and it also fails (as a coincidence, it is also an AMD processor, an Athlon XP 1800+). I really think that the code that implements job suspension was optimized only for Intel processors, and never tested for AMD, which it incorrectly identifies in some Rosetta's internal assembly programming code.

Rosetta stinks... I doubt that it can even return usable and reliable results.

Note: the AMD Turion is the mobile version of the Athlon XP. At 1.6GHz, the Turion runs with the same performance as an Athlon XP at 3.2GHz, because it integrates two parallel execution pipelines with much lower energy used and less heat. This processor is then more expensive than the equivalent Athlon model, but is great for notebooks. (This is quite similar to the differences between Intel Pentium M and Pentium 4)

Are there people getting any success with Rosetta? Doother peopleget so many "client" errors on Intel processors too?

---

Currently there is a problem with running multiple projects with Rosetta. The following is copied from the Recommended System Requirements

If you are running multiple BOINC projects using the BOINC Manager, we recommend setting your general preferences to "Leave applications in memory while preempted." To do this, after creating an account, do the following:

Click the "Your account" link on the main home page under "Returning participants" or click the "Participants" link above.
Login if you haven't already.
Click the "View or edit general preferences" link on your account page.
Click "Edit preferences".
Select "yes" for the "Leave applications in memory while preempted?" option.

---

Join the Teddies@WCG

Many thousands of people are running Rosetta with complete success. While the application switching problem may be frustrating for you, it is a known condition of running rosetta. The latest release of the application (as of 18/2/06) contains a fix for this issue and others. But if you decide to stay with Rosetta@home, please take a few moments to read the FAQs that are posted in the number crunching forum, and the project requirements posted as a link on the home page. This will help you understand the limitations and requirements for running Rosetta.

---

Moderator9
ROSETTA@home FAQ
Moderator Contact

Many thousands of people are running Rosetta with complete success. While the application switching problem may be frustrating for you, it is a known condition of running rosetta. The latest release of the application (as of 18/2/06) contains a fix for this issue and others. But if you decide to stay with Rosetta@home, please take a few moments to read the FAQs that are posted in the number crunching forum, and the project requirements posted as a link on the home page. This will help you understand the limitations and requirements for running Rosetta.

Completely stupid. Such requierment is not only unacceptable, goes against all recommandatations for BOINC projects, and does not even solve the problem that Rosetta has a severe bug that just forbids someone of just suspending the project temporarily or shutting down his PC for whatever reason he likes (including to finish antivirus updates).

A BOINC project that can't be stopped at any time is not a BOINC project, just an unnecessary pollution (in all senses of this word). If you're not able to fix it in Rosetta (like it was done for al other projects), get out of BOINC, because your "recommanded" setting affects in fact ALL projects, not only Rosetta.

---

The are new versions of the ROSATTA application. But if you had set your system to keep application in memory as described in the FAQs and on the forums, during switchs you may not have had these errors. Rosetta is not the only project where this helps. Also if you set your time between application switchs to a longer value (120 MIN.) that would help too.

Ther are over 40,000 users of Rosetta. The vast majority are having sucess. Most of them are running intel processors.

---

Moderator9
ROSETTA@home FAQ
Moderator Contact

The are new versions of the ROSATTA application. But if you had set your system to keep application in memory as described in the FAQs and on the forums, during switchs you may not have had these errors. Rosetta is not the only project where this helps. Also if you set your time between application switchs to a longer value (120 MIN.) that would help too.

Ther are over 40,000 users of Rosetta. The vast majority are having sucess. Most of them are running intel processors.

Well, I have been successfully running ROSETTA for many months, up until last week. At that point, one of my computers began failing ALL ROSETTA WU with the same exit code 1073741819. I have reset the project and detached/re-attached the project with no success. Last night, a second machine began exhibiting the same errors. I am at a loss, but it appears that all WU are running about 7-8 hours now and then failing. They use to run about 2 hours to completion. Obviously, I have some heartburn with donating that much computer time to a project and having nothing come of it, no credit, no usable end result. I also noted that BOINC is downloading massive numbers of workunits for ROSETTA now, enough that BOINC switches to "earliest deadline" mode and pretty much stays there until the ROSETTA WU have all errored out.

Since the project was running successfully for many months on both machines, and I have made no changes to either one, it appears that "something" has changed in the ROSETTA application or WU. In scaning through the forums, I see that I am not the only person having the problem.

Sure would be nice to see some words on this posted on your main page under the news section and sent out via RSS feed. Not all of the folks donating resources to the project can spend time searching through forum postings trying to find help.

---

Well, I have been successfully running ROSETTA for many months, up until last week. At that point, one of my computers began failing ALL ROSETTA WU with the same exit code 1073741819. I have reset the project and detached/re-attached the project with no success. Last night, a second machine began exhibiting the same errors. I am at a loss, but it appears that all WU are running about 7-8 hours now and then failing. They use to run about 2 hours to completion. Obviously, I have some heartburn with donating that much computer time to a project and having nothing come of it, no credit, no usable end result. I also noted that BOINC is downloading massive numbers of workunits for ROSETTA now, enough that BOINC switches to "earliest deadline" mode and pretty much stays there until the ROSETTA WU have all errored out.

Since the project was running successfully for many months on both machines, and I have made no changes to either one, it appears that "something" has changed in the ROSETTA application or WU. In scaning through the forums, I see that I am not the only person having the problem.

Sure would be nice to see some words on this posted on your main page under the news section and sent out via RSS feed. Not all of the folks donating resources to the project can spend time searching through forum postings trying to find help.

The application has been updated twice since the date you mention. There are some words on your specific issues in the Rosetta FAQ thread. The three most important FAQs in fixing your issues would be the FAQ on having WUs that get stuck at 1%, the FAQ on adjusting the NEW Rosetta "Time" parameter in your preferences, and the temporary FAQ suggesting shorter run time settings. The Time parameter adjustment in combination with a large work queue setting, is what is causing your problem with EDF mode processing in BOINC and it is also responsible for downloading too much work. I would suggest you set your "time" to something like 2 or 4 hours. Adjust your connect time to something like .25 days (you do not actually have to connnect that often) then, Update the project (you don't have to reset, an update will do just fine). Let the system download some work. Then allow the system to run for a while. Initially it may go into EDF mode, but in a day or so of running it will settle out and should run fine from there. You can then either make small adjustments to the connection time to get the amount of work you actually want, or adjust the run time to run the WUs for the amount of time that will fill the time between connections. If you are on a modem, the second option would be best for your system. If you are on DSL or other full time connection, your system will run better with a small queue and shorter WU time settings.

While most projects keep the FAQ pages fairly static, I have been trying to update them to reflect changes in the project as they occur. So please check there if you have a surprise like this again. If you can't find an answer there, You should try posting wither in the number crunching forum, or on the Moderators contact thread in the Cafe. You will get a much faster response to questions that way.

I do check the Q&A section about once a day, time permitting, but I (and many others) tend to see posts to the regular forums much faster.

---

Moderator9
ROSETTA@home FAQ
Moderator Contact

The application has been updated twice since the date you mention. There are some words on your specific issues in the Rosetta FAQ thread. The three most important FAQs in fixing your issues would be the FAQ on having WUs that get stuck at 1%, the FAQ on adjusting the NEW Rosetta "Time" parameter in your preferences, and the temporary FAQ suggesting shorter run time settings. The Time parameter adjustment in combination with a large work queue setting, is what is causing your problem with EDF mode processing in BOINC and it is also responsible for downloading too much work. I would suggest you set your "time" to something like 2 or 4 hours. Adjust your connect time to something like .25 days (you do not actually have to connnect that often) then, Update the project (you don't have to reset, an update will do just fine). Let the system download some work. Then allow the system to run for a while. Initially it may go into EDF mode, but in a day or so of running it will settle out and should run fine from there. You can then either make small adjustments to the connection time to get the amount of work you actually want, or adjust the run time to run the WUs for the amount of time that will fill the time between connections. If you are on a modem, the second option would be best for your system. If you are on DSL or other full time connection, your system will run better with a small queue and shorter WU time settings.

While most projects keep the FAQ pages fairly static, I have been trying to update them to reflect changes in the project as they occur. So please check there if you have a surprise like this again. If you can't find an answer there, You should try posting wither in the number crunching forum, or on the Moderators contact thread in the Cafe. You will get a much faster response to questions that way.

I do check the Q&A section about once a day, time permitting, but I (and many others) tend to see posts to the regular forums much faster.[/color][/b]
[/quote]


I appreciate the response. I did indeed read the FAQ, but found that most of suggestions either did not apply (since I was already configured that way) or required a change to a global BOINC setting. I did adjust the time parameter as suggested and will see if that has any affect. I don't beleive that I should have to make global changes that affect all other projects just to make this one project work again. I would also suggest that the default setting for the time parameter be changed if that is causing the problem. Many folks donating time to this project will have left those parameters at the default, not understanding their purpose.

I will, in the future, take problems like this to the other board. I suppose I was fooled into thinking the forum named "QUESTIONS & PROBLEMS" would be the appropriate place to ask questions concerning problems. :-)

Thanks again for your help.

---

...I will, in the future, take problems like this to the other board. I suppose I was fooled into thinking the forum named "QUESTIONS & PROBLEMS" would be the appropriate place to ask questions concerning problems. :-)

Thanks again for your help.

I have to agree with you on that one. But this can be a busy place. When I have an issue (like this one) ongoing I try to check more often, But the reality is that the project team reads the forums, all day, and only gets to the Q&A section about once a week. In any case your point is well taken.

---

Moderator9
ROSETTA@home FAQ
Moderator Contact