Hi!
Thank You for the information, so no more windows wus in the near future?

---

Hans Sveen
Oslo,Norway

There are no new Windows based batches currently but no doubt there will be some later in the year.

Linux is the focus for the next batches (which of course you can run on Windows using WSL).

---

---
CPDN Visiting Scientist

I tried to run open ifs task on wsl1 but it crashed.

I wonder if we can unpin the other announcements posts and pin this one. Are the people with the right permissions no longer active on the forum?

Also coming up are two OpenIFS (linux) projects. These will be multicore and a high resolution configuration which need more time to get ready. They will also need machines with 64+GB of RAM.

My Linux machine has 128 GB of RAM.

How many cores needed for each work unit? Can I decide this, or do you? If me, how do I set it up?

---

In reply to kotenok2000's message of 9 Jan 2025:

I tried to run open ifs task on wsl1 but it crashed.

WSL1 doesn't have a full Linux kernel like WSL2, that may be the reason. I've successfully run a lot of Linux CPDN and other BOINC projects in WSL2 (Ubuntu). Very much recommend it, relatively easy to set up, uses minimal resources (compared to traditional VMs). With a little extra work you can control WSL2 BOINC client via the Windows BOINC manager.

Sadly I will have to duck out of these multicore IFS tasks - my computer can only take 32GB RAM. Any other IFS work in the pipeline that will work with this amount of memory?

In reply to Alan K's message of 10 Jan 2025:

Sadly I will have to duck out of these multicore IFS tasks - my computer can only take 32GB RAM. Any other IFS work in the pipeline that will work with this amount of memory?

Seconded regarding smaller tasks! I have several systems that either can't handle more RAM or would require a complete RAM replacement to get past my [previous] standard 32GB choice, but they all coped well with that test task Glenn packaged up a while ago.

I am currently replacing a failed system with a 64GB system with two spare memory slots and 128GB capability, so that one should be able to cope :-) -- the others won't ever be able to do these tasks, and I don't have another replacement needed at present (future failures excepted!...) -- and I was really looking forward to running more OpenIFS tasks...

Cheers - Al.

---

In reply to Alan K's message of 10 Jan 2025:

Sadly I will have to duck out of these multicore IFS tasks - my computer can only take 32GB RAM. Any other IFS work in the pipeline that will work with this amount of memory?

Yes, there might be some lower memory batches. We're still working out the requirements for the projects.

Regarding memory, yes the task actually only takes around 25-27Gb RAM depending on how much output is configured. So if it's a machine not used for anything else that'll be fine. But I was thinking of raising the RAM requirements assuming most PCs running these will also be used for other work at the same time. This isn't set in stone yet, so good to get opinions.

---

---
CPDN Visiting Scientist

I have a number of 32GB machines that would run CPDN dedicated so I would like that option to be available if the model can cope please

In reply to Vato's message of 11 Jan 2025:

I have a number of 32GB machines that would run CPDN dedicated so I would like that option to be available if the model can cope please

Would it be technically possible to implement this as an opt-in in user profile settings?
By default, these tasks would only be sent to hosts with 64+ GB of RAM (or however much you decide), but a user could lower the RAM threshold to 32+ GB.

Given these tasks would be opt-in, I feel so long as we make the expectation clear at the point of opt-in, it shouldn't be an issue to release them to 32GB machines.

In reply to wujj123456's message of 12 Jan 2025:

Given these tasks would be opt-in, I feel so long as we make the expectation clear at the point of opt-in, it shouldn't be an issue to release them to 32GB machines.

Not a bad idea. If there were too many chancing it the memory restrictions could be put back in or the maximum number of attempts could be set at 5 rather than 3 if this didn't delay the results coming in from good machines too much.

To keep the announcements thread clear.

Not sure where to ask this question, but will we see projects for Darwin/MacOS again? Would love to contribute some time again!

---

In reply to Robert Thomas Wyatt's message of 21 Mar 2025:

Not sure where to ask this question, but will we see projects for Darwin/MacOS again? Would love to contribute some time again!

There may be occasional batches of HADCM3S tasks. The issue is the recent iterations of Macs not supporting 32 bit working so these will only work on older kit. I don't know if there is an equivalent of the installing 32bit libraries on Linux. If there is, it is I suspect only ever going to be an option for those with a greater understanding of the OS as I am sure it wouldn't have Apple support. The other possibility is the OIFS code being ported for running on MacOS. I certainly would not rule this out in the longer term but it certainly isn't a priority right now. OIFS is 64 bit so not affected.

I can't remember which other projects are affected but I know there are at least two others using older 32bit code requiring 32bit libraries that don't exist.

Sorry I can't offer anything more hopeful at the moment.

There won't be any native macOS tasks for the foreseeable future. The problem is in maintaining codes to work across 3 operating systems; Windows/Linux/macOS. CPDN doesn't currently have resources available to achieve it.

Running a linux or Windows virtual machine on macOS to run CPDN tasks is perfectly feasible, even if it doesn't take full advantage of the powerful Apple chips.

---

---
CPDN Visiting Scientist

Thank you for the quick response! It's unfortunate but I understand. I'm able to contribute to five other BOINC projects right now, some do require Rosetta, but it only supports Intel Mac code. I wonder how difficult it is to contribute time to helping port your projects, is there a thread for this?

---

In reply to Robert Thomas Wyatt's message of 25 Mar 2025:

Thank you for the quick response! It's unfortunate but I understand. I'm able to contribute to five other BOINC projects right now, some do require Rosetta, but it only supports Intel Mac code. I wonder how difficult it is to contribute time to helping port your projects, is there a thread for this?

If you have the right expertise I can talk to CPDN about the possibility. There might be licensing issues with some of the code, I'm not sure.

To help you'd need to have a good working knowledge of C++ (fortran would be a plus), linux & macOS, git, Makefiles & building software on multiple platforms. It would be hands-on work.

---

---
CPDN Visiting Scientist

I don't have that expertise. Only have experience in helping to package fink projects. Kind regards.

---