So what is the problem here?
I keep getting credit and my user total keeps going up, but my user average keeps diving to the bottom of the graph every time a WU completes and credit is granted.

Anyone got any ideas?

---

I have the same "observation", more or less, though it's been around since the new credit system went into effect. I intend to make my opinion known when I get around to it, but not just yet. I will say this: my RAC here at Rosetta is slightly lower than my RAC at Einstein even though my resource share at Rosetta is DOUBLE that of Einstein, 2:1.

See here for more discussion on this problem. My post and their reply seem to explain it.

---

See here for more discussion on this problem. My post and their reply seem to explain it.

My contention is that awarding of credit here at Rosetta is low, that's all. The more resources I dedicate to Rosetta the lower my ranking becomes on the world stage.

That may be, but then we are not talking about the world stage, only what Rosetta shows as our average and stats related to Rosetta not other projects.
I only run Rosetta, thats all I am interested in doing.

See here for more discussion on this problem. My post and their reply seem to explain it.

My contention is that awarding of credit here at Rosetta is low, that's all. The more resources I dedicate to Rosetta the lower my ranking becomes on the world stage.

---

The recent problems were due to the filling of the rosetta@home queue with the memory intensive "HINGE" workunits. With these work units we were building models for the current CAPRI protein-protein structure prediction challenge http://capri.ebi.ac.uk/, for which predictions must be submitted by Sunday, and so the jobs were targeted as high priority. The problem was that while we specified a minimum of 512MB of memory for these jobs, this was only sufficient for many machines if a single CPU was being used. We apologize for the problems which the combination of high memory and high priority caused, and will definitely avoid this in the future.

---

The recent problems were due to the filling of the rosetta@home queue with the memory intensive "HINGE" workunits. With these work units we were building models for the current CAPRI protein-protein structure prediction challenge http://capri.ebi.ac.uk/, for which predictions must be submitted by Sunday, and so the jobs were targeted as high priority. The problem was that while we specified a minimum of 512MB of memory for these jobs, this was only sufficient for many machines if a single CPU was being used. We apologize for the problems which the combination of high memory and high priority caused, and will definitely avoid this in the future.

I've reattached nine machines due to your response.

I've reattached nine machines due to your response.

I kept all of my computers attached during this non-issue.

I've reattached nine machines due to your response.

I kept all of my computers attached during this non-issue.

It isn't a non-issue to me when Rosetta crashes my remotely located dual CPU machines.

How much memory is needed to run 2 "HINGE" workunits on a 2 CPU

computer without problems?

I have 2 computers with 2 cpus.

A 3,0 P4 HT with 2048MB memory = no problem even when used for work.

A C2D 2.16GHz with 1024MB memory = no problem to run when not in use.
(I did not get to test it in use I'm out of HINGE-s)

Did anybody else notice any thing?

Anders n

---

How much memory is needed to run 2 "HINGE" workunits on a 2 CPU
computer without problems?

My observations would say BOINC would have to be allowed to use about 750-800MB of memory to keep two HINGE tasks running at the same time.

==============
I've created a new thread to discuss memory related issues. RAC, the topic of this thread is now discussed there. I hope you will agree after reviewing the information there, that things are working properly with regard to memory and issuing large tasks to appropriate systems.
==============
j2satx, I'm not clear how your remote machines "crashed". Perhaps you mean they were unable to download work for a period of time, or that they are unable to keep both CPUs active. I've attempted to address these points in the above referenced thread. Please post there if you have further questions or insight based on your observations.

---

Rosetta Moderator: Mod.Sense

How much memory is needed to run 2 "HINGE" workunits on a 2 CPU computer without problems?
A 3,0 P4 HT with 2048MB memory = no problem even when used for work.
Did anybody else notice any thing?
Anders n

I'm no expert at interpreting windows task manager. I was running 2 HINGE tasks simultaneously (2 Einstein tasks started but in wait status) with peak mem usage of about 360MB each. My total computer-wide commit charge was 1.2GB and windows generated system cache of around 0.7-0.9 GB. Those numbers are the highest I've ever seen on my P4/HT machine. I have 2.5GB of installed memory which is probably why I never experience the problems that so many often write about.

==============
I've created a new thread to discuss memory related issues. RAC, the topic of this thread is now discussed there. I hope you will agree after reviewing the information there, that things are working properly with regard to memory and issuing large tasks to appropriate systems.
==============

Hi Mod.Sense

The reason I asked the question is not that I think it does not work as it should.

It is more so we learn the limits and maybe can "tweak" Boinc so it works better in the future.

Anders n

---

The reason I asked the question...

Yes, I was trying to respond to three ideas going on in this thread. The rest of my reply wasn't meant to be directed at your comments.

I'm all for trying to better understand and tweak BOINC. I hope the info. I put together helps to do that.

---

Rosetta Moderator: Mod.Sense