Can't wait to try out the fixed _5nvx_ tasks.

BOINC insists on getting WCG tasks despite Rosetta having twice the resource share.

WCG Resource share is different to all other BOINC projects in the way it is set (and probably as a result how it is honoured).
Best to reduce the share for WCG (you have to do it at their site from memory) than increase the share for Rosetta.

From a quick search

"at WCG, Resource Share" is called "Project Weight", and is set under Device Manager > Device Profiles."

If Rosetta is set to 200, WCG needs to be 100 for Rosetta to have double.
And given the number of projects you run, the smaller your cache, the sooner your Resource share settings will be honoured. 0 Cache would be best- 0 days + 0.01 additional days.

---

Grant
Darwin NT

That's exactly my setting, 100 for WCG, 200 for Rosetta, and a ridiculously small cache. It does not seem to work well since my RAC for WCG is always over twice of that of Rosetta.

I think the problem is that WCG has GPU tasks and those massively boost my RAC.

That's exactly my setting, 100 for WCG, 200 for Rosetta, and a ridiculously small cache. It does not seem to work well since my RAC for WCG is always over twice of that of Rosetta.

I think the problem is that WCG has GPU tasks and those massively boost my RAC.

Your cache isn't ridiculously small. For the number of projects you've signed up to, it's huge (even with only two projects, it's still a large cache).
The idea of the cache is to give you enough work till you're able to connect to the internet again. As most people are able to connect as required, it makes a cache unnecessary. If you run only one project, then you may use a cache to get through project outages. But if you have more than 1 project, then there's no need for a cache.

If you had no cache, then as soon as Rosetta started producing work again, you'd have gotten some. As it is the fact that your cache is over a day, and WCG is actually a whole bunch of projects, and there has been no work from Rosetta for several days (and this after not having any work for several days and only having work for a week or so before running out again) means your system will have loaded up on WCG Tasks, and any of your other Projects if they had work available.
Until most of them have been cleared you won't get any more Rosetta Tasks. Then BOINC should do Rosetta to meet your Resource share settings, then do the odd WCG Task once the debt to Rosetta has been repaid (but if Rosetta runs out of work again (although that shouldn't happen as quickly this time), then it would load up on Tasks for other projects due to your large cache setting, delaying it doing Rosetta work when it became available again).




Edit- oh, and the Scheduler uses REC (Recent Estimated Credit) for determining scheduling, not the actual Credit granted by a a project. So a project that pays well above the definition of a Cobblestone will show more Credit than one that pays by the book (or underpays), even if they have actually done work in accordance with their Resource Share settings.

---

Grant
Darwin NT

Wait, is 0.1 + 0 days a huge cache? That's the setting I have locally (BTW, what makes you think my cache settings are over a day? Are my actual settings listed somewhere? If my cache setting is actually 1 day, that could explain some things.). I believe local cache settings take precedence over web settings (I have left web cache settings at default), is that correct?

The only WCG tasks I have are the ones that are currently running, there isn't any in the queue. There are Rosetta tasks fresh in the queue though, since the WCG tasks (ARP, 12 hours per task on my PC) are all nearly done.

My problem is that BOINC seems to really like downloading WCG tasks over Rosetta tasks, regardless of RAC. Your explanation of REC makes a lot of sense. Thanks.

Hi,

One of my computer has not got tasks for a week. I resetted the project, than reinstall the whole BOINC client to the newest, but nothing happened. Other computers get jobs well. This computer has only Windows 10 32 bit - does this matter?
Thanks in advance,

fkmaster

---

I'm not sure of there are take that will run on the 32-bit version, but if your can't upgrade toa64 but OS you could run Rosetta in a 64 bit Linux VM, e.g WSL.

---

Wait, is 0.1 + 0 days a huge cache?

If you've got more than 1 project, yes. And given that the deadlines for Rosetta are 3 days, having any cache is a very significant portion of the deadline limit. Where as if the deadline on a Task is 2 months, and you have a 7 day cache, it's an insignificant portion of the deadline period.
So you would be much better off with it as 0 & 0.01

That's the setting I have locally (BTW, what makes you think my cache settings are over a day? Are my actual settings listed somewhere? If my cache setting is actually 1 day, that could explain some things.).

Your cache is over a day.
If you look at your Computer list, and click on Details for the Ryzen 5 3600, it shows your Average turnaround time is 1.32 days. Click on "Application details- Show" and the turnaround time for Rosetta 4.20 (the current application) is 1.71 days.
That's a massive portion of the deadline period.

I believe local cache settings take precedence over web settings.

Yep.
Local settings override any web based settings.
It allows those with heaps of systems to set up different settings for multiple computers using the web based options & locations (School, Work & Home), and if there are one or two computers that are used regularly for other things that are impacted in some y by BOINC crunching they can use the Local settings on those & not affect any of the other systems.

(I have left web cache settings at default), is that correct?

I've no way of knowing- only you can see your account pages & the values there.

The only WCG tasks I have are the ones that are currently running, there isn't any in the queue. There are Rosetta tasks fresh in the queue though, since the WCG tasks (ARP, 12 hours per task on my PC) are all nearly done.

So once those WCG Tasks are done, and the already downloaded Rosetta ones start, then the next Tasks to be downloaded should be more Rosetta.

---

Grant
Darwin NT

One of my computer has not got tasks for a week. I resetted the project, than reinstall the whole BOINC client to the newest, but nothing happened. Other computers get jobs well. This computer has only Windows 10 32 bit - does this matter?

Only one of your systems has Rosetta work- the Ryzen 5 3600.
None of your other systems have any Rosetta work at present. This is most likely due to their low core/thread counts, the fact that Rosetta is your least important project behind the other 2, your large cache settings, and Rosetta having been out of work for a while.

If you reduce your cache to zero, then once those system have completed their present work for the other projects, they should start getting some Rosetta work. And with the smaller cache your Resource share settings will be meet much sooner.

Your account, Preferences, When and how BOINC uses your computer, click on Computing preferences, Other

           Store at least 0.0  days of work
Store up to an additional 0.01 days of work

Save the changes.
The next time a system contacts the Rosetta Scheduler it will get the new settings & they will take effect (or click on Update in the BOINC Manager on each system).




Edit- and it would be worth running the BOINC Benchmarks on your i3 system as it is using the default values & Rosetta uses the benchmarks for determining Credit for work done.

---

Grant
Darwin NT

One of my computer has not got tasks for a week. I resetted the project, than reinstall the whole BOINC client to the newest, but nothing happened. Other computers get jobs well. This computer has only Windows 10 32 bit - does this matter?

Only one of your systems has Rosetta work- the Ryzen 5 3600.
None of your other systems have any Rosetta work at present. This is most likely due to their low core/thread counts, the fact that Rosetta is your least important project behind the other 2, your large cache settings, and Rosetta having been out of work for a while.

Rosetta has the highest resource share (200 vs 100) out of my projects, and I'm currently only running WCG alongside it on my devices. The reason why only one of my systems has Rosetta work is that I only have 2 working devices right now, the other device that is active is an Android phone (the Asus ROG 5 which has a 0 cache setting) which refuses to download Rosetta tasks. It completes a WCG task then insists on downloading another WCG task instead of a Rosetta task.

BTW, all of my devices have either a 0 cache setting or a 0.1 + 0 cache setting set locally, depending on how fast the GPU tasks (if available) run.

BTW, all of my devices have either a 0 cache setting or a 0.1 + 0 cache setting set locally, depending on how fast the GPU tasks (if available) run.

That is too short for Rosetta. The work units run longer than that, and they won't download at all. (That will happen on CPDN too, with their long work units.)
Try 0.1 + 0.5 days.

---

Rosetta has the highest resource share (200 vs 100) out of my projects, and I'm currently only running WCG alongside it on my devices. The reason why only one of my systems has Rosetta work is that I only have 2 working devices right now, the other device that is active is an Android phone (the Asus ROG 5 which has a 0 cache setting) which refuses to download Rosetta tasks. It completes a WCG task then insists on downloading another WCG task instead of a Rosetta task.

BTW, all of my devices have either a 0 cache setting or a 0.1 + 0 cache setting set locally, depending on how fast the GPU tasks (if available) run.

If you look at what i posted there, it's all all in response to dcdc.

I suggest you checkout the post before it where i quoted what you had posted previously.

BTW, all of my devices have either a 0 cache setting or a 0.1 + 0 cache setting set locally, depending on how fast the GPU tasks (if available) run.

But i'll summarise again- 0 & 0.01 is best, not the other way around.
And the numbers shown against your computers & the applications show your cache is over 1.5days- that's how long it's taking you to return Rosetta work (i can't see any details for WCG).
Try 0 + 0.01 to see if that reduces the turnaround times- it'll take a couple of days to see if it does.

---

Grant
Darwin NT

BTW, all of my devices have either a 0 cache setting or a 0.1 + 0 cache setting set locally, depending on how fast the GPU tasks (if available) run.

That is too short for Rosetta. The work units run longer than that, and they won't download at all. (That will happen on CPDN too, with their long work units.)
Try 0.1 + 0.5 days.

If it doesn't accept 0, then give it 0.01 & 0.01.
CPDN might not give you work, but Rosetta does with those values.

The only thing that will stop (or should stop) you from getting work is if the Task won't be returned before the deadline. Having zero cache means just that- no cache. You will still get work- but only have the work you are presently processing, When it's done another Task will be downloaded.



And you have the values around the wrong way to get the result people are usually after-

If you want a .5 day cache, you need to set it as 0.5 & 0.01 additional days.
4 day cache, 4 & 0.01
Having a large value for the Additional days means you will actually end up with less work in your cache than the Days value as BOINC will let it run down until it can fill the additional Days value without going over the Day + Additional days total value.
Set the Day value for the size of the cache you want, then 0.01 Additional days so the cache remains around that level, and not significantly below it before re-filling.

---

Grant
Darwin NT

The only thing that will stop (or should stop) you from getting work is if the Task won't be returned before the deadline. Having zero cache means just that- no cache. You will still get work- but only have the work you are presently processing, When it's done another Task will be downloaded.

That is so only if he is not running any other projects. But it appeared to me that he is.

---

I don't want to jinx things, but so far no errors on any of the new Tasks.

---

Grant
Darwin NT

The only thing that will stop (or should stop) you from getting work is if the Task won't be returned before the deadline. Having zero cache means just that- no cache. You will still get work- but only have the work you are presently processing, When it's done another Task will be downloaded.

That is so only if he is not running any other projects. But it appeared to me that he is.

Indeed, I have WCG running. My issue is that WCG appears to have a higher priority than Rosetta despite my resource settings. Rosetta has 200 resource share vs 100 for WCG.

When my devices complete a task, they seem to prefer downloading WCG tasks over Rosetta ones. I do NOT have a bunch of tasks piled up. With my current settings BOINC only downloads new tasks if the current ones are almost complete.

BTW, is there currently a shortage of Android tasks? It's been a really long time since my phone received a Rosetta task.

Double post please delete.

Please delete, dunno why there were this many duplicate posts.

Triple post please delete

[snip]

When my devices complete a task, they seem to prefer downloading WCG tasks over Rosetta ones. I do NOT have a bunch of tasks piled up. With my current settings BOINC only downloads new tasks if the current ones are almost complete.

WCG usually has more tasks available. Lately, Rosetta usually doesn't.

BTW, is there currently a shortage of Android tasks? It's been a really long time since my phone received a Rosetta task.

There are no Android or Linux or Windows Tasks, they can all be processed by the appropriate application.
However many of the current Tasks need over 1GB of RAM. Depending on your BOINC memory settings & the amount of available RAM on the Android device it may not be possible for it to process them, so it won't get any even if Rosetta is owed processing time on the device.

---

Grant
Darwin NT