In this period I tried to return folding for this project, but it is too demanding, even for some high-end computer:
- Some WUs take up to 3 gb of RAM each
- A lot of writes to hard disk (many GBs per day), and if you have an SSD, it will age very fast
- Tricky BOINC settings configuration for a typical user (for ex. settings to leave app in memory, app change every 120 mins, number of cores balance per RAM available, cache, etc.)
- You may lose hours of computations on some WUs if you turn off your computer in the wrong moment, due to how rosetta handle some checkpoints
- Shorts deadlines "not for everyone"
- Sudden and subtle project updates (which can impact heavily on your machine) that you have to search on the forums to find them (for ex. "from now on WU will take more time to complete", or "will consume more RAM", etc.).

Comparing to other projects (f@h, wcg, etc.), this project seems too niche for the generic audience. For example, for f@h, I can tell people "download here", "install this", "right click and select light" ... and that's all. I'm sure they will not have memory or performance problems, short deadlines, ssd problems, etc. for a lot of time.
I can understand every project has its needs, but from the homepage of this project it seems "for everyone", which is not. Even for "aware & conscious users", with a good hardware, can be a problem, and you have to follow it over time to avoid finding problems on your machine in the future. In other words, it's not a set and forget project.
I know it's a scientific project with the main focus on the science part, but why after so many years there was no improvement in this regard? If devs don't want/don't have resources to sort out these things, at least putting an warning message somewhere, a really brief description on what to do and what to expect, can be a good thing (don't know, for ex. in the homepage, in the project description, or as a BOINC alert, and so on) ... a way to be honest to who is supporting you donating resource for your cause. Yes, "some" (and only some) things are written in the FAQ, but ... they are too long and detailed to read for a typical user, too hidden in the main website to find them, and, most importantly, a typical user tends to not read them anyway.

p.s.
For typical user I'm referring to users that don't know what a CPU or RAM is, make confusion between facebook and internet, don't know what a forum is, but can click the big orange button "Join Rosetta@home" in the homepage and can install software programs.

---

Comparing to other projects (f@h, wcg, etc.), this project seems too niche for the generic audience. For example, for f@h, I can tell people "download here", "install this", "right click and select light" ... and that's all. I'm sure they will not have memory or performance problems, short deadlines, ssd problems, etc. for a lot of time.

Points of view.
I think F@H is more "obscure" for volunteers (there is not a user profile to see errors, no button to force download/upload, web interface is crappy, etc)

---

- A lot of writes to hard disk (many GBs per day), and if you have an SSD, it will age very fast

A non-issue actually, as The Tech Report SSD Endurance Experiment showed, many years ago. As much as Rosetta does work storage, it is bugger all over all for even a consumer SSD. And If you are running some heavy database queries on your system, i would hope you're using a more appropriate SSD, and Rosetta's load would be barely a drop in the ocean.

- Tricky BOINC settings configuration for a typical user (for ex. settings to leave app in memory, app change every 120 mins, number of cores balance per RAM available, cache, etc.)

There's nothing tricky about it- you click on Edit, make the changes you want, update the changes. But like anything- it takes time to understand the terminology and determine the meaning.
Which is why there are sticky threads & other quick references for those that have a look around before asking questions.

People complain when they can't configure things just the way they want them. People complain when there are too many options available to them. People complain.

- You may lose hours of computations on some WUs if you turn off your computer in the wrong moment, due to how rosetta handle some checkpoints

This is certainly a problem that does need to be addressed, urgently as there are Tasks that will take even longer to complete.
Lose a few minutes of work (as with a checkpoint), no problem. Lose 2, 8, 20 hours of work. Big problem.

- Shorts deadlines "not for everyone"

*shrug*
Long deadlines are only of use if you run just the one project, and the results aren't time sensitive (or you have a system that is rarely on or rarely connected to the internet- then most certainly, Rosetta isn't for you). For more than one project (even when results aren't time sensitive) short deadlines are best in order for your Resource share settings to be honoured sooner rather than much, much, much later.
And guess what- the reason most of the present users are here, is because there is work here that is time sensitive (you know that virus thing that's doing the rounds..) and the present deadlines reflect that fact.
If doing work that isn't important isn't of interest to you, then Rosetta most certainly is not the project for you.

- Sudden and subtle project updates (which can impact heavily on your machine) that you have to search on the forums to find them (for ex. "from now on WU will take more time to complete", or "will consume more RAM", etc.).

I agree it would be much better if these things were posted as News for the Main page (not than many people bother), and so it comes up on their BOINC Manager's message tab (not that people bother to read it), before such updates are implemented.
People will still complain (because they don't bother paying attention to the announcements), but at least they can be told- we did inform you of these pending changes.
Much better than explaining why people's systems are now doing strange things compared to what they were doing before (but that will still be necessary because people don't bother paying attention to the things they should).
Such is life.

Comparing to other projects (f@h, wcg, etc.), this project seems too niche for the generic audience. For example, for f@h, I can tell people "download here", "install this", "right click and select light" ... and that's all. I'm sure they will not have memory or performance problems, short deadlines, ssd problems, etc. for a lot of time.

One of the advantages of a Project that doesn't let any others make use of their volunteer computing resources..
The idea of BOINC is that any project can set themselves up & make of the computing resources available. Of course, different projects have different requirements, and different people have different ideas as to which project will get their support. And if they support multiple projects some projects will get more support than others. And so on.

I can understand every project has its needs, but from the homepage of this project it seems "for everyone", which is not. Even for "aware & conscious users", with a good hardware, can be a problem, and you have to follow it over time to avoid finding problems on your machine in the future. In other words, it's not a set and forget project.

It depends on how you set it up, and that's the thing- It doesn't stop you from doing dumb things. It will try and honour your decisions a best as it can, no matter how ridiculous they might be.
Rosetta is one project of many, and each project has it's own requirements. You can set things up with the defaults, and generally it will work, even for people that attach to all sorts of projects. But the requirements for Rosetta will cause issues for some systems, and minimum requirements should be put on the front page, or at least the the download link (not that people will read them).

Yes, "some" (and only some) things are written in the FAQ, but ... they are too long and detailed to read for a typical user, too hidden in the main website to find them, and, most importantly, a typical user tends to not read them anyway.

Just like they don't read all the other warnings you're suggesting Rosetta should put up.
I agree, there should bee a minimum system requirement easy to find for those looking to join.
But most people won't bother to read it. No need, they can ask here for help.

p.s.
For typical user I'm referring to users that don't know what a CPU or RAM is, make confusion between facebook and internet, don't know what a forum is, but can click the big orange button "Join Rosetta@home" in the homepage and can install software programs.

And those typical users often need help with even those tasks.

---

Grant
Darwin NT

Points of view.
I think F@H is more "obscure" for volunteers (there is not a user profile to see errors, no button to force download/upload, web interface is crappy, etc)

Other projects may have other problems, but some of them are very easy to set up and run, and I think this is one of the most important thing for a project (this way it's easier to spread out).

A non-issue actually, as The Tech Report SSD Endurance Experiment showed, many years ago. As much as Rosetta does work storage, it is bugger all over all for even a consumer SSD. And If you are running some heavy database queries on your system, i would hope you're using a more appropriate SSD, and Rosetta's load would be barely a drop in the ocean.

Understandable, but Rosetta shouldn't take that decision for me. This is why even a simple warning message would be very useful. For example "WARNING: this project may write and delete many GBs of data per day on your memory. Some experiments says this isn't harmful."

There's nothing tricky about it- you click on Edit, make the changes you want, update the changes. But like anything- it takes time to understand the terminology and determine the meaning.
Which is why there are sticky threads & other quick references for those that have a look around before asking questions.

People complain when they can't configure things just the way they want them. People complain when there are too many options available to them. People complain.

It would be so complicated to tell a typical user to click edit, and to modify certain options. Example: I can't say in a "tweet" how to properly get running Rosetta on a generic computer, while I can do it for Folding@home without having to specify anything in detail (for ex. "download here, double click to install, click next until finished, select "light" profile when prompted, and you are set"). Btw, I'm not here to praise f@h, that project have other kind of problems too, it's just to explain the concept I want to express :)

Long deadlines are only of use if you run just the one project, and the results aren't time sensitive (or you have a system that is rarely on or rarely connected to the internet- then most certainly, Rosetta isn't for you). For more than one project (even when results aren't time sensitive) short deadlines are best in order for your Resource share settings to be honoured sooner rather than much, much, much later.
And guess what- the reason most of the present users are here, is because there is work here that is time sensitive (you know that virus thing that's doing the rounds..) and the present deadlines reflect that fact.
If doing work that isn't important isn't of interest to you, then Rosetta most certainly is not the project for you.

I accept your opinion, so this is why a simple message would clarify things to the end user, to express the point of view of Rosetta scientists. For example: "Rosetta tasks are time sensitive, you are encouraged to leave your PC on and connected."

One of the advantages of a Project that doesn't let any others make use of their volunteer computing resources..
The idea of BOINC is that any project can set themselves up & make of the computing resources available. Of course, different projects have different requirements, and different people have different ideas as to which project will get their support. And if they support multiple projects some projects will get more support than others. And so on.

I 100% agree with you!

It depends on how you set it up, and that's the thing- It doesn't stop you from doing dumb things. It will try and honour your decisions a best as it can, no matter how ridiculous they might be.
Rosetta is one project of many, and each project has it's own requirements. You can set things up with the defaults, and generally it will work, even for people that attach to all sorts of projects. But the requirements for Rosetta will cause issues for some systems, and minimum requirements should be put on the front page, or at least the the download link (not that people will read them).

I agree


[quote]Just like they don't read all the other warnings you're suggesting Rosetta should put up.
I ag

A non-issue actually, as The Tech Report SSD Endurance Experiment showed, many years ago. As much as Rosetta does work storage, it is bugger all over all for even a consumer SSD. And If you are running some heavy database queries on your system, i would hope you're using a more appropriate SSD, and Rosetta's load would be barely a drop in the ocean.

Understandable, but Rosetta shouldn't take that decision for me. This is why even a simple warning message would be very useful. For example "WARNING: this project may write and delete many GBs of data per day on your memory. Some experiments says this isn't harmful."

Along with a warning Rosetta will use your CPU more than other programmes, Rosetta will use your Memory more than other programmes, Rosetta will use your Network more than other programmes, Rosetta will use your computer power supply more than other programmes, etc, etc.
These would be required warnings for any programme you install on a system. Some will have bigger impacts on different parts of a system than others, but they will all have impacts.

But unless any of these impacts actually poses a risk to the continued function of one of those parts, a warning is unnecessary. And that is the case with Rosetta and SSDs- It's disk usage, while more than for many other programmes, is not sufficient to result in any significant reduction in the life expectancy of even the most basic of consumer SSDs.

---

Grant
Darwin NT

I guess I look at it a bit differently. While I do agree that Rosetta is not for everyone, I think I can point someone to it and have them install with the default settings and run just fine. They can run for many years without ever knowing they are only using 2 of their 4 CPUs because the default settings are conservative, especially on memory. But they were running along and because of the conservative default settings, they never noticed any sluggish response on their machine. They were blissfully unaware that a WU didn't take a checkpoint and they lost hours of work when they powered down last night.

One way to look at a failed WU, a missed deadline or lost crunching time is to try and fix it (from the user's side). Another way is to presume that the Project Team sees my outcomes, learns over time, and makes changes on their end to help better utilize the population of unattended machines that are out there. If one of the search protocols is seeing frequent termination due to the watchdog, it is going to take a number of machines, across various environments to study that problem and recode to address it. The "failure" is contributing to the next iteration of the science, and that is what the project is all about.

---

Rosetta Moderator: Mod.Sense

In this period I tried to return folding for this project, but it is too demanding, even for some high-end computer:
- Some WUs take up to 3 gb of RAM each
- A lot of writes to hard disk (many GBs per day), and if you have an SSD, it will age very fast
- Tricky BOINC settings configuration for a typical user (for ex. settings to leave app in memory, app change every 120 mins, number of cores balance per RAM available, cache, etc.)
- You may lose hours of computations on some WUs if you turn off your computer in the wrong moment, due to how rosetta handle some checkpoints
- Shorts deadlines "not for everyone"
- Sudden and subtle project updates (which can impact heavily on your machine) that you have to search on the forums to find them (for ex. "from now on WU will take more time to complete", or "will consume more RAM", etc.).

Comparing to other projects (f@h, wcg, etc.), this project seems too niche for the generic audience. For example, for f@h, I can tell people "download here", "install this", "right click and select light" ... and that's all. I'm sure they will not have memory or performance problems, short deadlines, ssd problems, etc. for a lot of time.
I can understand every project has its needs, but from the homepage of this project it seems "for everyone", which is not. Even for "aware & conscious users", with a good hardware, can be a problem, and you have to follow it over time to avoid finding problems on your machine in the future. In other words, it's not a set and forget project.
I know it's a scientific project with the main focus on the science part, but why after so many years there was no improvement in this regard? If devs don't want/don't have resources to sort out these things, at least putting an warning message somewhere, a really brief description on what to do and what to expect, can be a good thing (don't know, for ex. in the homepage, in the project description, or as a BOINC alert, and so on) ... a way to be honest to who is supporting you donating resource for your cause. Yes, "some" (and only some) things are written in the FAQ, but ... they are too long and detailed to read for a typical user, too hidden in the main website to find them, and, most importantly, a typical user tends to not read them anyway.

p.s.
For typical user I'm referring to users that don't know what a CPU or RAM is, make confusion between facebook and internet, don't know what a forum is, but can click the big orange button "Join Rosetta@home" in the homepage and can install software programs.

I'm not sure at all what you're even on about. All i did was "go here, click this" to install the software. On startup I put in my info and off I went. Quite frankly I think the rest of your logic is nonsensical at best, based entirely on your definition of typical users being (and no I'm not calling anybody one) a total idiot. If you really know that little about computers/software perhaps don't run around installing stuff on your system?

I guess I look at it a bit differently. While I do agree that Rosetta is not for everyone, I think I can point someone to it and have them install with the default settings and run just fine. They can run for many years without ever knowing they are only using 2 of their 4 CPUs because the default settings are conservative, especially on memory. But they were running along and because of the conservative default settings, they never noticed any sluggish response on their machine. They were blissfully unaware that a WU didn't take a checkpoint and they lost hours of work when they powered down last night.

One way to look at a failed WU, a missed deadline or lost crunching time is to try and fix it (from the user's side). Another way is to presume that the Project Team sees my outcomes, learns over time, and makes changes on their end to help better utilize the population of unattended machines that are out there. If one of the search protocols is seeing frequent termination due to the watchdog, it is going to take a number of machines, across various environments to study that problem and recode to address it. The "failure" is contributing to the next iteration of the science, and that is what the project is all about.

Strange, now the default settings are 100% CPU, half day of cache, use half of RAM, leave app in memory option set to off. Try yourself ... they might worsen it?
Anyway, even with 2 task for a quad core, if those people had 8 GB of RAM, considering peak usage of 3 GB per task, would be 3/4 of their RAM wasted. Also considering the "% of total RAM in use" option is at its default (50%), and considering each task has a variable RAM usage, both task would be continuously terminated by BOINC if exceeding the 50% of RAM. This way elaborations risk to not respect the deadline, or risk to go into a fault (because we all know Rosetta doesn't like too much being shutting down and up too fast).

I'm not sure at all what you're even on about. All i did was "go here, click this" to install the software. On startup I put in my info and off I went. Quite frankly I think the rest of your logic is nonsensical at best, based entirely on your definition of typical users being (and no I'm not calling anybody one) a total idiot. If you really know that little about computers/software perhaps don't run around installing stuff on your system?

They are not idiots, but most of them are ignorant, and it's not their fault: simply it's not their job nor their passion. Sadly I work in this sector, and everyday I find this kind of people. I'll tell you more: I find these people even when it's their job fixing computers or developing software.
For the Rosetta project, if you run it at its defaults, then there is an high chance to have problems or slowdowns with your computer, and/or create problems to Rosetta project returning faulty elaborations or exceeding the expiration time without knowing it (wasting electricity on your side and precious time on Rosetta side).

---

my guess is one way is to setup the boinc-directory say on an usb flash drive, or an sd-card.
io performance would deteriorate but it would be about as fast (or slow) as perhaps a Raspberry pi where io goes.
and as flash drives or sd cards these days are rather low costs and easily comes in sizes of 16 GB - 32 GB, it would be less of a concern that those age. alternately, if there is a spare sata slot on the pc, it may help to setup the boinc directory on a separate small ssd drive, that drive could be used for various 'scratch' purposes as well since the space is reclaimed when boinc isn't running and the results uploaded. that could probably work well for r@h

i tend to crunch on and off, it is my main pc. so i'd only run r@h jobs only at night when i'm not using it.
i'd let the jobs run overnight if need be. but some jobs tend to be very long.
r@h is rather demanding on disk and memory, but i've seen more extreme ones like atlas@home from cern. some of those have both higher demand on cpu, storage and memory and some use virtual machines which requires a lot of allocated memory. by comparison r@h is rather memory and storage efficient in that sense

as i'm using linux, i'd limit the max cpu speeds so that they run a little less than the max ghz speeds.
and by doing that i'm able to get the cpu fan to temper down at some lower rpm than full speeds which lets it run rather quiet and yet not losing much performance

There is no problem with SSD....
You can use the SLC SSD that is very long life.Don't use MLC or TLC SSD,it's fragile.

Recommended SLC SSD is Micron P300 100/200G(SATA600) 、Micron P320H 350/700G（PCI-E 8X）.
Micron P300 100/200G is very cheap,just $50/100…… in China it is……

There is no problem with SSD....
You can use the SLC SSD that is very long life.Don't use MLC or TLC SSD,it's fragile.

MLC & TLC technologies so have less write cycles before they fail compared to SLC.
However with over provisioning & good garbage collection & wear levelling, write amplification is reduced to the point that current TLC drives have similar TBW (Terra Bytes Written) per day/DWPD (Drive Writes Per Day) to MLC drives.

---

Grant
Darwin NT