This is not a GPUUG Team Forum without one of this threads.

This is not a GPUUG Team Forum without one of this threads.

+1

---

This is not a GPUUG Team Forum without one of this threads.

+1

Or whatever 2, 3, whatever :)

---

Help, my tagline is missing..... Help, my tagline is......... Help, m........ Hel.....

Hey, I'm on the board. Got credit for my first task! Take that, COVID-19!

The first task took 28,500 sec for 330 points, but the second task was over 43,000 CPU seconds for 20 points! I thought S@H was stingy. Are these first tasks just trying to optimize what program I run?

If most are like that second one, I'll probably shift CPUs to E@H or WCG. Thoughts?

The first task took 28,500 sec for 330 points, but the second task was over 43,000 CPU seconds for 20 points! I thought S@H was stingy. Are these first tasks just trying to optimize what program I run?

If most are like that second one, I'll probably shift CPUs to E@H or WCG. Thoughts?

Keith recently has some issues with app being used.... I think it corrects itself after you run for bit and gets a better app for your setup.

---

This project has a 3 day hard deadline, so you do not want to buffer your jobs. I am still working out the best numbers but i think i am going back to default boinc buffer of 0.1 days + .1 days extra. I currently have more jobs hitting deadline and aborting than i have completing. Such a big change from seti which had a three month deadline......

---

Hey, I'm on the board. Got credit for my first task! Take that, COVID-19!

+1

---

I'm not sure if I'll let Rosetta give me another round. I wish these projects would go back to a standard metric for credit. 3 of 4 gave 20 credits and one had an error of some sort.

https://boinc.bakerlab.org/rosetta/results.php?userid=2134993&offset=0&show_names=0&state=4&appid=

I know WCG is getting ready to start a COVID-19 project called OpenPandemics and I believe they also focus on CPU jobs.

My 2nd RTX 2060 Super came smoking in here yesterday afternoon.
After giving it time to "rest"/stabilize I just installed it.

My UPS is reporting a 50~ watt drop in the power draw. And the other 2060 Super was running slightly faster than the p100-102 so the next step appears to be transplanting the system back into a regular ATX computer case.

The original plan was to get my big server case back from the sales/loan but the Coronavid-19 has put a cramp in that. I still have the "Cube" that I first ran my AMD 2990wx in. I may put the AMD 3950x into there.

Tom M

---

Help, my tagline is missing..... Help, my tagline is......... Help, m........ Hel.....

I'm not sure if I'll let Rosetta give me another round. I wish these projects would go back to a standard metric for credit. 3 of 4 gave 20 credits and one had an error of some sort.

https://boinc.bakerlab.org/rosetta/results.php?userid=2134993&offset=0&show_names=0&state=4&appid=

I know WCG is getting ready to start a COVID-19 project called OpenPandemics and I believe they also focus on CPU jobs.

I know that Keith has recommended this change because some cpu tasks are stalling which would produce the lack of credits you are describing.

From my cc_config.xml file:

<cc_config>
 <log_flags>
   <sched_op_debug>1</sched_op_debug>
 </log_flags>
 <options>
   <use_all_gpus>1</use_all_gpus>
   <save_stats_days>90</save_stats_days>
   <max_file_xfers>4</max_file_xfers>
   <max_file_xfers_per_project>2</max_file_xfers_per_project>
   <max_tasks_reported>50</max_tasks_reported>
   <no_alt_platform>1</no_alt_platform>
 </options>
</cc_config>

The no_alt_platform disables the download of the "other" version of the apps that appears to be stalling. Don't get me wrong, I thought the issue was specific to AMD cpus. But I could be wrong. And if I am, that will get rid of the stalling problem for you too once the cache is cleared of the "686" tasks.

The developer(s) are working on a fix but don't have a time to fix prediction.

Tom M

---

Help, my tagline is missing..... Help, my tagline is......... Help, m........ Hel.....

Has anyone got a good feel for where their RAC will end up?

---

Help, my tagline is missing..... Help, my tagline is......... Help, m........ Hel.....

If you take a look at the 20 credit task stderr.txt, you will see it basically spun its wheels for 8 hours or whatever your target time is set for and found 1 decoy. When you look at the same type of task run by another host you will see it find hundreds of decoys and was awarded 330 credits.

Your 20 credits was basically, "thanks for your contribution . . . now go somewhere else"

The problem is with the app stalling out and erroring or getting watchdogged. Try to eliminate any 32 bit application and only run the 64 bit apps.

Credit on this project sucks in general. I am running 40 cores full time on Linux machine and am getting a RAC of about 40k. My other three rigs add up to the total 45k that I am at now. I was getting better RAC when I set my computation time to the smaller time limit instead of the longer ones. I am not sure if that was a coincidence or not....

---

I'm not sure if I'll let Rosetta give me another round. I wish these projects would go back to a standard metric for credit. 3 of 4 gave 20 credits and one had an error of some sort.

https://boinc.bakerlab.org/rosetta/results.php?userid=2134993&offset=0&show_names=0&state=4&appid=

I know WCG is getting ready to start a COVID-19 project called OpenPandemics and I believe they also focus on CPU jobs.

I know that Keith has recommended this change because some cpu tasks are stalling which would produce the lack of credits you are describing.

From my cc_config.xml file:

<cc_config>
 <log_flags>
   <sched_op_debug>1</sched_op_debug>
 </log_flags>
 <options>
   <use_all_gpus>1</use_all_gpus>
   <save_stats_days>90</save_stats_days>
   <max_file_xfers>4</max_file_xfers>
   <max_file_xfers_per_project>2</max_file_xfers_per_project>
   <max_tasks_reported>50</max_tasks_reported>
  [b] <no_alt_platform>1</no_alt_platform>[/b]
 </options>
</cc_config>

The no_alt_platform disables the download of the "other" version of the apps that appears to be stalling. Don't get me wrong, I thought the issue was specific to AMD cpus. But I could be wrong. And if I am, that will get rid of the stalling problem for you too once the cache is cleared of the "686" tasks.

The developer(s) are working on a fix but don't have a time to fix prediction.

Tom M

Did setting no_alt_platform help Keith or Tom? Since this is in cc_config, does it affect any other projects?

---

I'm not sure if I'll let Rosetta give me another round. I wish these projects would go back to a standard metric for credit. 3 of 4 gave 20 credits and one had an error of some sort.

https://boinc.bakerlab.org/rosetta/results.php?userid=2134993&offset=0&show_names=0&state=4&appid=

I know WCG is getting ready to start a COVID-19 project called OpenPandemics and I believe they also focus on CPU jobs.

I know that Keith has recommended this change because some cpu tasks are stalling which would produce the lack of credits you are describing.

From my cc_config.xml file:

<cc_config>
 <log_flags>
   <sched_op_debug>1</sched_op_debug>
 </log_flags>
 <options>
   <use_all_gpus>1</use_all_gpus>
   <save_stats_days>90</save_stats_days>
   <max_file_xfers>4</max_file_xfers>
   <max_file_xfers_per_project>2</max_file_xfers_per_project>
   <max_tasks_reported>50</max_tasks_reported>
  [b] <no_alt_platform>1</no_alt_platform>[/b]
 </options>
</cc_config>

The no_alt_platform disables the download of the "other" version of the apps that appears to be stalling. Don't get me wrong, I thought the issue was specific to AMD cpus. But I could be wrong. And if I am, that will get rid of the stalling problem for you too once the cache is cleared of the "686" tasks.

The developer(s) are working on a fix but don't have a time to fix prediction.

Tom M

Did setting no_alt_platform help Keith or Tom? Since this is in cc_config, does it affect any other projects?

As far as I know anything in the cc_config.xml will effect all projects.
That said I put it in because of the reports that seemed to be AMD centred. They may not be AMD centred but strictly app version centred.

I am also running E@H (gpu/cpu), World Community Grid (cpu) and PrimeGrid (cpu). I am not noticing any difference except the possibly Rosetti has slowed down for some reason.

Tom

---

Help, my tagline is missing..... Help, my tagline is......... Help, m........ Hel.....

There is a new Pandora release in Seti group forum to address this issue at the project level instead of setting it globally. I am at work and cannot try it but Juan and Ville want the version tested.....any takers. It will be tomorrow by the time I can test it.

---

I'm not sure if I'll let Rosetta give me another round. I wish these projects would go back to a standard metric for credit. 3 of 4 gave 20 credits and one had an error of some sort.

https://boinc.bakerlab.org/rosetta/results.php?userid=2134993&offset=0&show_names=0&state=4&appid=

I know WCG is getting ready to start a COVID-19 project called OpenPandemics and I believe they also focus on CPU jobs.

I know that Keith has recommended this change because some cpu tasks are stalling which would produce the lack of credits you are describing.

From my cc_config.xml file:

<cc_config>
 <log_flags>
   <sched_op_debug>1</sched_op_debug>
 </log_flags>
 <options>
   <use_all_gpus>1</use_all_gpus>
   <save_stats_days>90</save_stats_days>
   <max_file_xfers>4</max_file_xfers>
   <max_file_xfers_per_project>2</max_file_xfers_per_project>
   <max_tasks_reported>50</max_tasks_reported>
  [b] <no_alt_platform>1</no_alt_platform>[/b]
 </options>
</cc_config>

The no_alt_platform disables the download of the "other" version of the apps that appears to be stalling. Don't get me wrong, I thought the issue was specific to AMD cpus. But I could be wrong. And if I am, that will get rid of the stalling problem for you too once the cache is cleared of the "686" tasks.

The developer(s) are working on a fix but don't have a time to fix prediction.

Tom M

Did setting no_alt_platform help Keith or Tom? Since this is in cc_config, does it affect any other projects?

As far as I know anything in the cc_config.xml will effect all projects.
That said I put it in because of the reports that seemed to be AMD centred. They may not be AMD centred but strictly app version centred.

I am also running E@H (gpu/cpu), World Community Grid (cpu) and PrimeGrid (cpu). I am not noticing any difference except the possibly Rosetti has slowed down for some reason.

Tom

I had to run my R@H cache down to 0 and reset the project before it would (apparenlty) stop sending me "686" apps. Someone suggested in on the "don't know where to post it thread".

Tom M

---

Help, my tagline is missing..... Help, my tagline is......... Help, m........ Hel.....

The new Pandora release allows you to set no_alt_platform at each project level. I have not been getting 686 apps but then again, i have been running Rosetta for a long time and it may have corrected itself over time.....

---

I think for accuracy we should be the "CPU Users Group" for this project. :^)

Most of my CPU cores were sitting idle with SETI@Home gone, so where better to fill them than here. Kudos to all of us for helping out where needed.