Einstein not running
16 Oct 2022 4:13:21 UTC
Topic 228385
(moderation:
But Milky Way is.
I just rejoined after being away for several years. Einstein isn't running. It appears under Projects, but not in the Tasks list. I have no notices or messages. When I check the status page it shows everything is running, and I am not savvy enough to know how to find sched_reply and a spotlight search doesn't find such a term except here.
"My machine is not getting work! Why not?
"There are a number of possible reasons. In most cases, the Einstein@Home scheduler sends a detailed text explanation to your machine. But the BOINC client does not always display these messages correctly or clearly. So if in doubt, look in the sched_reply.xml file, located in the directory where BOINC is installed, and search for lines that have the word 'message' in them."
Much appreciate any help. Thanks.
SBranca wrote: But Milky Way
)
Go to the bottom of this page and click on Applications and then Server Status, on the Applications page you will see that there are only a couple of Apps for the Mac like yours, then the Server Status page will tell you which Apps actually have any tasks available. I'm guessing you are choosing to run one kind of App and the Server doesn't have any of those to send out and you need to go your Account, Preferences, Project page and select the correct Venue for you Mac and then choose which Apps to actually get tasks from.
SBranca wrote:...Einstein
)
Your computers are not hidden so I was able to see that you have an M1 Mac.
If you look at the applications page I don't think there are any apps that will run on M1 Macs. There are apps that run on Intel Macs. I haven't paid much attention since I have no use for the new Apple architecture but I think there were some comments from the Devs to the effect that they didn't have the time or resources to develop a suitable app for that architecture.
BTW, if you look at your computer as listed on the website, there is a link you can click to see the results of the most recent scheduler contact. I took a look at that and it showed a 0.00 secs work request.
One thing that surprised me was that you have a host ID that well and truly pre-dates the advent of the M1 Mac. I guess that you must have been running an Intel Mac previously and you've carried over your previous ID to the new system. That would be why the scheduler message gives values for things like active frac, on frac, DCF. These must be from the previous machine since you don't have any current stats :-).
Cheers,
Gary.
Gary Roberts wrote: I don't
)
FGRP5 has an M1 Mac app. It’s the one that says Mac OS X running on ARM v1.14.
there is no GPU app for M1 though only this CPU app
_________________________________________________________________________
Thanks for your replies. I
)
Thanks for your replies. I think I figured something out. I did as suggested and went to Preferences, etc. and selected Gamma Rays Search Nos. 1 and 5, deselected the rest. Now Einstein is running. Thank you for that tip.
One other question. Do new tasks have to be added by clicking Update or are they sent automatically?
Re my host ID: I wouldn't know what is behind that, but it not only predates my current M1 but I go back to System 7 (shortly after the Big Bang).
I appreciate your help.
Steve
SBranca wrote:... went to
)
You can only get tasks for the CPU search - FGRP5. You may as well disable the GPU search, FGRBB1G. Both of these searches are supposed to be finishing at some point, so don't be surprised if FGRP5 tasks stop in the future. It doesn't seem to be particularly imminent though. There should be further announcements in 'Tech News' about this.
The function of the BOINC core client is to implement the preferences you have set, one of which is the amount of work to keep 'on hand'. As the science app completes tasks, the client will return the results and request replacements so as to keep your work cache topped up. There is no need to click the update button if the client is working correctly.
One small tip, check your settings to see what work cache size is currently set. There is a default but you may have changed it (long ago in an alternate universe) so you should take a look to see. Don't make it too large. For the type of work you are doing, a day or two should be more than enough. Outages in work availability are usually quite rare.
If you go to your account page on the website and click on the details link for your computer, you will see that the host ID (4202395) first joined the project on 24 Aug 2011.
Perhaps you might remember what machine you were using at that time? It must have worked for a while since the detailed information about it shows that it has contacted the project's servers a total of 2,964 times, a few of which would be very recent. For your current machine to have picked up that host ID, you must have transferred your old BOINC installation to your current machine - critical parts of it at least. That's not a problem since things seem to be working.
As a result of having a new look at your machine, I see something you might like to investigate. You have a single task only and it's not FGRP5. It's a different CPU search called Arecibo, large - which you said would be disabled. Is it running correctly and making progress? Do you actually have the FGRP5 search enabled?
If it's running correctly seems like you have two choices for CPU work after all :-). Maybe you should check your preferences carefully. If you really have disabled the Arecibo, large search, you should enable it again to keep getting those in future. There is a preference setting for allowing "non-preferred" work. Maybe you had that enabled so the scheduler chose that search rather than FGRP5 for some reason.
You have a multi-core CPU but only one task. That's fine if you deliberately chose that. Maybe you are running CPU tasks for other projects on the other cores.
There is a setting for the % of cores BOINC is allowed to use. It would be very useful to go through all the settings since what you used in 2011 may not be appropriate now. Depending on what type of machine you have and how tolerant it might be to the heat load from high intensity crunching, you would be wise to think about limiting the total number of cores in use, particularly if the machine starts to overheat and throttle itself.
Cheers,
Gary.
Gary, Thanks for such a
)
Gary, Thanks for such a thorough response. It's quite helpful. The 2011 machine was a Mac Book, probably five years old at the time. It was replaced with a mid-2012 machine which ran BOINC too. For the current machine I downloaded BOINC fresh but there may have been some files that came over with the data transfer from that Mac to this one.
(I can't get a screen shot of the app list to copy here) I selected only the two apps mentioned (ticked) thinking those were the ones that M1 handles. Maybe I misunderstood that. I don't see "Arecibo large". I have changed some preferences to allow more work - 2 days - and more processor capacity and availability.
Binary Radio Pulsar Search (Arecibo)
Binary Radio Pulsar Search (Arecibo, fast)
Binary Radio Pulsar Search (MeerKAT)
Gamma-ray pulsar binary search #1 (ticked)
Gamma-ray pulsar search #5 (ticked)
Gamma-ray pulsar binary search #1 (GPU)
Gravitational Wave search O2 Multi-Directional
Gravitational Wave search O2 Multi-Directional GPU
Gravitational Wave search O3 All-Sky
You have been very generous with your time. I'm a low-grade newbie so I really don't expect you to provide a full guided tour of the system! You've got way more important things to do. You have given me a bunch of info that I can work through for now, and I appreciate it. I'm happy to go exploring.
Many thanks.
SBranca wrote: ... I
)
The Arecibo, fast entry is the CPU search that would work for you. It used to be called Arecibo, large (if I remember correctly) and I wasn't aware it was now called "fast". I'm not running CPU tasks and haven't been paying attention to current names :-).
My recollection is that the plain Arecibo entry refers to very small tasks suited to mobile devices like phones and tablets and the one below it is a bundle of these small tasks (bundle of 8 I think) to give a larger workload for desktop type computers. 'large' would be more descriptive - I don't know why they are using 'fast' to describe something that will take 8 times longer to crunch. I guess it must mean 'fast devices' rather than fast tasks :-).
The gamma-ray pulsar binary search #1 is only done on GPUs these days. Many years ago there was a CPU version so the other entry you ticked really shouldn't be there as there aren't any tasks these days. You should un-tick that one.
The #5 search is where you'll get some tasks for the moment. The Devs have stated that it will be finishing at some point but you'll be able to continue getting tasks for Arecibo, fast if that happens.
Cheers,
Gary.
My work units are showing up
)
My work units are showing up a "completed, marked as invalid"... I have reset the project multiple times. Any suggestions? Until the unplanned project shutdown in late August, everything was going fine. For weeks, I've been crunching work units but they' all getting rejected. I have 3 separate computers running on this, all with the same result. Thx
Barry Barnett wrote: My work
)
The first thing I would check is to be sure your URL addresses are all changed to "https" instead of "http".
Late August is about when the changeover took place.
Just a thought...
Proud member of the Old Farts Association
Barry Barnett wrote: My work
)
Your invalid tasks seem to be running on the Intel gpu, is that the cpu also? And if so are you ALSO running cpu tasks on the same cpu? If so try running just the gpu tasks by suspending the cpu tasks and see if you get some to come thru as inconclusive or even valid. Gary Roberts hopefully will come along soon and give more conclusive help as he knows how to read the error files whereas I don't.