I get an insufficient memory notice:

Rosetta@home: Notice from server
Rosetta needs 3337.86 MB RAM but only 2879.42 MB is available for use.
1/30/2022 5:47:45 AM

My mobo is constrained to 4gb of RAM, I will never have more, even if I wanted. I know it is an underpowered machine.

Should I withdraw from Rosetta@home?

For info this is my setup:
Operating System Windows 10 Home 64-bit
CPU Intel Pentium E6600 @ 3.06GHz
RAM 4.00GB Single-Channel DDR2 @ 400MHz (6-5-5-17)
Motherboard eMachines EMCP73VT-PM (CPU 1) 33 °C

Thanks

It's tight, but if you go into Boinc Computing Preferences and on the Disk and Memory tab allow 95% for RAM, it should just about fit in the tasks asking for 3337.86Mb
Don't worry that it sounds a lot to give to Boinc. We've found the tasks generally don't use anything like that full amount in practise. It just allows you to get over this limit at the time of task launch.

---

[quote]It's tight, but if you go into Boinc Computing Preferences and on the Disk and Memory tab allow 95% for RAM, it should just about fit in the tasks asking for 3337.86Mb
Don't worry that it sounds a lot to give to Boinc. We've found the tasks generally don't use anything like that full amount in practise. It just allows you to get over this limit at the time of task launch.

I've put mine to 100%. It's only a rough guide anyway. Numberfields or something like that goes over what I ask. Most projects never reach it.

I have set crunching time to one day and 12 hours in settings.

Eh? It takes as long as it takes.

There's a setting on Rosetta preferences that allows you to specify for how long you want each Rosetta 4.2 task to run.
It's called "Target CPU run time" and you can set that from 2 CPU hour to 36 CPU hours.
Rosetta 4.2 tasks will try to respect that setting. You can change the setting at any time and a project update will change the runtime of any running tasks.

To avoid Rosetta 4.2 tasks from running for an "unlimited" time (say, on a "stuck" model), there is a built-in feature called watchdog that will kill any Rosetta task once it passes the Target CPU runtime setting + 10 CPU hours. Eg. CPU target is 8 hours, watchdog will kill the task at 18 hours of CPU time.

I have heard of this option, but I don't understand the point of it. Presumably the scientists have already chosen the best time. Why adjust it?

hurry up....28 4.2 tasks waiting LMAO

So, since I could not get tasks for my regular BOINC (Rosetta) client, I went ahead and removed the old BOINC install, and reluctantly installed the bloated one with Virtual Box instead. This one uses an inordinate amount of resources even before crunching any numbers. On top of that, tasks run for something between 20 seconds and a few minutes and then fail. Not one single task actually went past 7 minutes. 100% of tasks failed with calculation errors. Right now, the computer sits idle with that big fat thing doing nothing, showing 5 failed tasks - I am used to seeing 8 tasks happily being crunched.

Now, this is a Core I7 with 32 GB ram running on a fully patched Windows 10 and 2 TB of available disk space. ( I have several computers using Windows or Linux Mint running BOINC - This is the guinea pig computer trying out the new client) Nothing else ever failed on it. I am thoroughly not impressed right now (i feel a need to confirm my state of mind in case the tone so far does not indicate it - hopefully this is not going to get me locked out). I tried to find settings that I could perhaps tweak, to no avail. I am seriously considering dumping BOINC and Rosetta altogether. But in desperation, I am looking for a glimmer of hope here. Is there some way to get this to work properly?

Oh, getting that off my chest does feel good... So, anyone with tips on making this contraption work, I owe a debt of gratitude. ;-)

For instant failures every time, i.e. VBox not detected or doesn't start at all, see #4 here:
https://boinc.bakerlab.org/rosetta/forum_thread.php?id=14886&postid=104574

For: CPU load drops to 0 within 1 minute, see the rest of that post above.

For occasional errors, typically after a few hours, see:
https://boinc.bakerlab.org/rosetta/forum_thread.php?id=14897

Sounds like the middle one is affecting your machine. You can open the VirtualBox manager and see the screen output for a running VM to see any errors.

---

So, since I could not get tasks for my regular BOINC (Rosetta) client, I went ahead and removed the old BOINC install, and reluctantly installed the bloated one with Virtual Box instead. This one uses an inordinate amount of resources even before crunching any numbers. On top of that, tasks run for something between 20 seconds and a few minutes and then fail. Not one single task actually went past 7 minutes. 100% of tasks failed with calculation errors. Right now, the computer sits idle with that big fat thing doing nothing, showing 5 failed tasks - I am used to seeing 8 tasks happily being crunched.

Now, this is a Core I7 with 32 GB ram running on a fully patched Windows 10 and 2 TB of available disk space. ( I have several computers using Windows or Linux Mint running BOINC - This is the guinea pig computer trying out the new client) Nothing else ever failed on it. I am thoroughly not impressed right now (i feel a need to confirm my state of mind in case the tone so far does not indicate it - hopefully this is not going to get me locked out). I tried to find settings that I could perhaps tweak, to no avail. I am seriously considering dumping BOINC and Rosetta altogether. But in desperation, I am looking for a glimmer of hope here. Is there some way to get this to work properly?

Oh, getting that off my chest does feel good... So, anyone with tips on making this contraption work, I owe a debt of gratitude. ;-)

Give up and attach to World Community Grid, Sidock, etc.

LouisD you have infected your computed with VirtualPox ,
its not a virus its just a new way of going nowhere fast
It took many months for the Rosetta team to develop the a custom build of Rosetta gon Boinc to use maximal resources [they have trimmed it a bit]
{Though they can run things in a virtual box that they cant do outside of one }
The error messages that the VB python work can produce are long , short and sometimes `far out man`
Do have a go at getting it to work , you my win , or in the end give up on it.
Some computers run it with no problems ,
others can run anything but the Python work units.
You may find that some of your other computers will run it without problems ,
it is impossible to know until you try and there will be no good reason why some work and others dont

Thank you for the insights. As it turns out, virtualization was not enabled in the bios for that computer. I would have sworn on my first born head that it was... That is now fixed. But tasks are not available anymore on the server. Oh well...

In the mean time, I signed up for World Community Grid. They have tasks available.

Maybe server decided to disable virtualbox tasks for your computer?
https://boinc.bakerlab.org/rosetta/forum_thread.php?id=14850#103842
https://imgur.com/KhnsqGU

Thank you for the insights. As it turns out, virtualization was not enabled in the bios for that computer. I would have sworn on my first born head that it was... That is now fixed. But tasks are not available anymore on the server. Oh well...

In the mean time, I signed up for World Community Grid. They have tasks available.

Cancer and Covid. They used to have loads of African Rainfall, TB, Childhood Cancer, all sorts. Covid sometimes (very occasionally) gives you GPU work, so make sure that's enabled. In the meantime they're ploughing through the cancer work really fast since the covid is fairly rare (the scientists can't keep up with our computers) and Rainfall is having technical problems with the programming.

Why aren't you getting pythons now? They're always available.

Virtual Box is not necessarily bad. But, in the Windows world I prefer Hyper-V (I run a few Windows Server 2019 Hyper-V servers) and in the Linux world, I like KVM better than Virtual Box. But I don't currently run productive KVM servers. Only test stuff. It is the plan that when Win Hyper-V server 2019 becomes unsupported, I will move VM's to a KVM server.

I feel that creating Virtual box VM's for BOINC is overkill and gross abuse of volunteered resources. Perhaps someone can explain, I might change mi mind if I am missing something here.

Anyway, thanks for your post. It was entertaining. :-)

Too many errors returned.
Server decided to disable python wus for the computer. you have to press allow at your host at
https://boinc.bakerlab.org/rosetta/hosts_user.php
i have left a screenshot several posts back

Why aren't you getting pythons now? They're always available.

I am just the guy offering my computer. Not the guy who designed the system... I don't know what to answer here. If it becomes too complicated to help, help may just be driven away. That is my answer, but it is not really an answer, is it?

ah. This explains that. So I allowed in the computer details. But they lost some of my compute time now. This is what happens when helping becomes too complicated.

Virtual Box is not necessarily bad. But, in the Windows world I prefer Hyper-V (I run a few Windows Server 2019 Hyper-V servers) and in the Linux world, I like KVM better than Virtual Box. But I don't currently run productive KVM servers. Only test stuff. It is the plan that when Win Hyper-V server 2019 becomes unsupported, I will move VM's to a KVM server.

I feel that creating Virtual box VM's for BOINC is overkill and gross abuse of volunteered resources. Perhaps someone can explain, I might change mi mind if I am missing something here.

Anyway, thanks for your post. It was entertaining. :-)

Ask in LHC - I'm sure they said it's the only way they can get consistent reproduceable comparable results from more than one user. Must depend on the kind of work you're doing. Kryptos, which is now closed started with VB and tried to make native version, which is apparently harder.

ah. This explains that. So I allowed in the computer details. But they lost some of my compute time now. This is what happens when helping becomes too complicated.

To be fair they're probably having just as many difficulties at their end. It's a new thing for Rosetta, this VB stuff. But.... communication with the users could help both ends!

There are three things that they can optimise for (not mutually exclusive):
1. Computer resources
2. Ease of use
3. Researcher's time

I think so far they're optimising #3 with VBox. I'd argue that the research is about as cutting edge as you can get (Nature's breakthrough and method of the year 2021) so I'm willing to put the effort in, but the reality is it could be a lot easier to run and that would lead to a lot more resources available.

And it wouldn't be difficult to improve communications, even is it's just once a week.

---

Take a look here https://www.manualslib.com/manual/1435811/Msi-X470-Gaming-Pro.html?page=32#manual and read the text to the side of the graphic.
I am in slot 2_1

As for slot 2_2 I am not sure what I have set up for SATA ports. I rarely go in there.
But I don't have much for SATA gear. DVD ROM and HDD and SDD I think.
So I could probably move them to other ports if 1 is not available.

Shit, I have the same board and wanted an M2 some day. Although mine has "plus max" on the end instead of "pro", not sure what that means.

there were additional issues overnight. I decided this morning that I would uninstall Virtual Box from my computer. As of now, my most capable computers are enlisted on World community grid, others will follow as time allows. If and when regular Rosetta tasks are available, they will process them. For now, I don't need the Virtual Box aggravation.