Yes, I made some changes to the settings on my account page.
OK. You should now leave those settings as is. There is one setting that you haven't mentioned - resource share. When you support multiple projects, the resource shares you choose for each become quite important. As you have 4 projects and on the assumption that you probably left each one at the default value of 100, E@H is only entitled to 25% of each host's resources. This allocation of resources is on a per host basis. E@H can reliably supply work whilst other projects do have shortages from time to time. It is quite easy for the E@H share to be exceeded and so BOINC will tend (if possible) to favour other projects until they 'catch up'. This could easily be the reason why the BOINC client on a particular host may stop asking for more E@H work for a period.
Quote:
I am confused because I have one good laptop that has a CUDA compatible GPU and 12-13 other Laptops's that do not have CUDA capable GPU's
There should be no confusion. With the one set of preferences, BOINC is smart enough to manage each individual host (within reason) no matter how well endowed or under endowed a particular host is.
Quote:
Since I have a mixture of PC's with different OS's and hardware I am not sure what the best settings are.
You seem to be fixating on the E@H preferences. In many ways, as I indicated in a previous answer, your general computing preferences can be more important and you don't seem to be talking about those. The OS that a particular host is running is of no real concern as BOINC fully supports the ones you mention. If you really decide that you would like to give different preferences to differently endowed hosts, you should investigate BOINC's ability to have different 'venues' - the default venue plus 'home', 'work' and 'school'. You could assign each host to one of those venues. That would enable you to have 4 different classes of hosts, each with their own independently settable preferences. I'm not suggesting you actually do this. A single set of preferences would most likely be fine for all your hosts.
Quote:
With these current settings some of my laptops are no longer connecting or running tasks. Not sure why.
You seem to have a couple of hosts with tasks in their caches that don't seem to be doing anything. In fact, some tasks have expired. Are those hosts actually running? How many hours per day do you allow them to run? If you have 4 projects per host and a host doesn't run for sufficient hours, deadlines can easily be exceeded.
As a personal preference, I'm very wary about running BOINC on laptops, particularly older machines whose heatsinks may be clogged with dust and fluff. Such machines may give compute errors or freexe up or even shut themselves down because of the stress imposed by the science apps. Do all your machines run reliably without error? Do any have overheating problems?
Another thing you could consider. If you are supporting 4 projects equally on all your hosts, do it a bit differently. I imagine E@H and Rosetta are the most reliable whilst Seti and LHC are the most likely to have work shortages or actually be offline. So pair up a reliable project with a less reliable one. Put one pair on half your hosts and the other pair on the other half. You would still end up giving 25% of your resources to each project but it would be easier for BOINC to manage the work caches on each individual host. For your best machine, still have all 4 projects on it. That way it will be the 'link' between all the projects so that when you change a preference at one project, the link machine will make sure the change gets noticed by all projects. BTW, it's good practice to use one project as a 'master' and make all your general pref changes there. Project specific pref changes are made at the project concerned.
Please realise that running BOINC projects is very much a personal preference thing. There is no single 'best' way to do anything. You have plenty of alternatives and the only way to find out what is best for you is to experiment and observe what happens. The 'production rate' of any one machine wont be improved dramatically just by tweaking BOINC preferences. Hardware upgrading will have a more dramatic effect. Apart from upgrading memory, you can't really do much with laptops. The real fun is investing in the latest hardware of the desktop form and seeing how productive you can make it. These days it doesn't need to be hugely expensive. Maybe you could sell a few less productive laptops on ebay and invest the proceeds in say a 2500K or 2600K plus a GTX560 GPU or two or something like that. It would be a lot of fun seeing how productive you could make that :-).
OK. You should now leave those settings as is. There is one setting that you haven't mentioned - resource share. When you support multiple projects, the resource shares you choose for each become quite important. As you have 4 projects and on the assumption that you probably left each one at the default value of 100, E@H is only entitled to 25% of each host's resources. This allocation of resources is on a per host basis. E@H can reliably supply work whilst other projects do have shortages from time to time. It is quite easy for the E@H share to be exceeded and so BOINC will tend (if possible) to favour other projects until they 'catch up'. This could easily be the reason why the BOINC client on a particular host may stop asking for more E@H work for a period.
Thanks Gary, I guess I didn't think to adjust the resource sharing. I want to dedicate most of my laptops towards the E@H app. I will play with the settings tonight and see if I get any improvement.
You seem to have a couple of hosts with tasks in their caches that don't seem to be doing anything. In fact, some tasks have expired. Are those hosts actually running? How many hours per day do you allow them to run? If you have 4 projects per host and a host doesn't run for sufficient hours, deadlines can easily be exceeded.
As a personal preference, I'm very wary about running BOINC on laptops, particularly older machines whose heatsinks may be clogged with dust and fluff. Such machines may give compute errors or freexe up or even shut themselves down because of the stress imposed by the science apps. Do all your machines run reliably without error? Do any have overheating problems?
I let the hosts run 24/7 usually and only had one overheating issue, but since I adjusted settings on that laptop to only use 60% of CPU it no longer overheats. However, I do sometimes see errors with clients connecting to the server, and I also see I have a lot of failed tasks. I wonder if I should set all of the laptops CPU usage down to 60-70% and see if that resolves the connection and computing errors.
... I wonder if I should set all of the laptops CPU usage down to 60-70% and see if that resolves the connection and computing errors.
You should certainly consider this if tasks are giving compute errors. It shouldn't be necessary and shouldn't be related to difficulties in connecting to the server.
RE: Yes, I made some
)
OK. You should now leave those settings as is. There is one setting that you haven't mentioned - resource share. When you support multiple projects, the resource shares you choose for each become quite important. As you have 4 projects and on the assumption that you probably left each one at the default value of 100, E@H is only entitled to 25% of each host's resources. This allocation of resources is on a per host basis. E@H can reliably supply work whilst other projects do have shortages from time to time. It is quite easy for the E@H share to be exceeded and so BOINC will tend (if possible) to favour other projects until they 'catch up'. This could easily be the reason why the BOINC client on a particular host may stop asking for more E@H work for a period.
There should be no confusion. With the one set of preferences, BOINC is smart enough to manage each individual host (within reason) no matter how well endowed or under endowed a particular host is.
You seem to be fixating on the E@H preferences. In many ways, as I indicated in a previous answer, your general computing preferences can be more important and you don't seem to be talking about those. The OS that a particular host is running is of no real concern as BOINC fully supports the ones you mention. If you really decide that you would like to give different preferences to differently endowed hosts, you should investigate BOINC's ability to have different 'venues' - the default venue plus 'home', 'work' and 'school'. You could assign each host to one of those venues. That would enable you to have 4 different classes of hosts, each with their own independently settable preferences. I'm not suggesting you actually do this. A single set of preferences would most likely be fine for all your hosts.
You seem to have a couple of hosts with tasks in their caches that don't seem to be doing anything. In fact, some tasks have expired. Are those hosts actually running? How many hours per day do you allow them to run? If you have 4 projects per host and a host doesn't run for sufficient hours, deadlines can easily be exceeded.
As a personal preference, I'm very wary about running BOINC on laptops, particularly older machines whose heatsinks may be clogged with dust and fluff. Such machines may give compute errors or freexe up or even shut themselves down because of the stress imposed by the science apps. Do all your machines run reliably without error? Do any have overheating problems?
Another thing you could consider. If you are supporting 4 projects equally on all your hosts, do it a bit differently. I imagine E@H and Rosetta are the most reliable whilst Seti and LHC are the most likely to have work shortages or actually be offline. So pair up a reliable project with a less reliable one. Put one pair on half your hosts and the other pair on the other half. You would still end up giving 25% of your resources to each project but it would be easier for BOINC to manage the work caches on each individual host. For your best machine, still have all 4 projects on it. That way it will be the 'link' between all the projects so that when you change a preference at one project, the link machine will make sure the change gets noticed by all projects. BTW, it's good practice to use one project as a 'master' and make all your general pref changes there. Project specific pref changes are made at the project concerned.
Please realise that running BOINC projects is very much a personal preference thing. There is no single 'best' way to do anything. You have plenty of alternatives and the only way to find out what is best for you is to experiment and observe what happens. The 'production rate' of any one machine wont be improved dramatically just by tweaking BOINC preferences. Hardware upgrading will have a more dramatic effect. Apart from upgrading memory, you can't really do much with laptops. The real fun is investing in the latest hardware of the desktop form and seeing how productive you can make it. These days it doesn't need to be hugely expensive. Maybe you could sell a few less productive laptops on ebay and invest the proceeds in say a 2500K or 2600K plus a GTX560 GPU or two or something like that. It would be a lot of fun seeing how productive you could make that :-).
Good luck with whatever you decide to do!
Cheers,
Gary.
OK. You should now leave
)
OK. You should now leave those settings as is. There is one setting that you haven't mentioned - resource share. When you support multiple projects, the resource shares you choose for each become quite important. As you have 4 projects and on the assumption that you probably left each one at the default value of 100, E@H is only entitled to 25% of each host's resources. This allocation of resources is on a per host basis. E@H can reliably supply work whilst other projects do have shortages from time to time. It is quite easy for the E@H share to be exceeded and so BOINC will tend (if possible) to favour other projects until they 'catch up'. This could easily be the reason why the BOINC client on a particular host may stop asking for more E@H work for a period.
Thanks Gary, I guess I didn't think to adjust the resource sharing. I want to dedicate most of my laptops towards the E@H app. I will play with the settings tonight and see if I get any improvement.
You seem to have a couple of hosts with tasks in their caches that don't seem to be doing anything. In fact, some tasks have expired. Are those hosts actually running? How many hours per day do you allow them to run? If you have 4 projects per host and a host doesn't run for sufficient hours, deadlines can easily be exceeded.
As a personal preference, I'm very wary about running BOINC on laptops, particularly older machines whose heatsinks may be clogged with dust and fluff. Such machines may give compute errors or freexe up or even shut themselves down because of the stress imposed by the science apps. Do all your machines run reliably without error? Do any have overheating problems?
I let the hosts run 24/7 usually and only had one overheating issue, but since I adjusted settings on that laptop to only use 60% of CPU it no longer overheats. However, I do sometimes see errors with clients connecting to the server, and I also see I have a lot of failed tasks. I wonder if I should set all of the laptops CPU usage down to 60-70% and see if that resolves the connection and computing errors.
RE: ... I wonder if I
)
You should certainly consider this if tasks are giving compute errors. It shouldn't be necessary and shouldn't be related to difficulties in connecting to the server.
Cheers,
Gary.