Choose correct Preference set for your host (same as the 'Location' of host). Then at the bottom of the page there are those boxes for "GPU utilization factor of .... apps:" . Default value is 1. That should result in running only 1 WU at a time, per GPU. If the values are something else, try changing them to 1 and save. On the Boinc Manager click an 'Update' for Einstein project. Any new tasks (not yet downloaded) shoud then start running only 1 at a time.
Another way to change that:
Edit or make and save a file app_config.xml in your Einstein project folder. Put these lines in the file:
Then on the Boinc Manager click 'Options' and 'Read config files'. The effect should be immediate. Ps. '.5' instead of '1' on the gpu_usage would mean 2 tasks per GPU.
Running 2 tasks per GPU will often produce a little bit higher RAC. If there's currently no problem with generating too much heat, noise or electricity (taking a tiny bit more from the wall) running 2 tasks per 1050Ti should be fine.
Choose correct Preference set for your host (same as the 'Location' of host). Then at the bottom of the page there are those boxes for "GPU utilization factor of .... apps:" . Default value is 1. That should result in running only 1 WU at a time, per GPU. If the values are something else, try changing them to 1 and save. On the Boinc Manager click an 'Update' for Einstein project. Any new tasks (not yet downloaded) shoud then start running only 1 at a time.
Clicking "Update" in Boinc after changing the "GPU utilization factor" is not enough to get the new setting to take effect. One needs to download new tasks for the specific search before the change will take effect, but once it does it will apply to all previous cached work.
Thanks but that is what it has always been set to. I was trying to figure if there was somewhere else that it might have a setting to run 2 WUs at a time.
I am suffering severe lag on the computer when it is running the 2 WUs and was hoping to address that problem by only running 1 WU at a time on the GPUs.
I have already set it to only use 75% of the CPUs. I had a TR with 16/32 cores and 2 X 1050 ti.
Again, thanks for you input but it must be something else.
There are only two places to change how many GPU tasks run concurrently and others have already mentioned both. These are (1) the GPU utilization factor, and (2) the use of an app_config.xml file.
Have you ever used either of these?
If you have previously used app_config.xml but have since just deleted the file, that may well be the reason you can't seem to change back to running single tasks per GPU. Deleting the file doesn't undo critical changes that were made to the state file (client_state.xml) at the time you first used it. If this happens to be the problem, you could reinstate the file and edit both <gpu_usage> and <cpu_usage> to be 1. That will get you what you want.
If you don't want to reinstate the file, you have two options. You could manually edit the state file to fix the issue (not recommended unless you really know what you are doing) or you could reset the project (which would automatically fix the state file but cause all project files - apps, data, tasks, etc - to be deleted and downloaded afresh).
If you have never used app_config.xml, the above doesn't apply. The next question would be, have you ever used different preferences for different 'locations' (previously known as 'venues')?
If you have, please confirm the value for GPU utilization factor for FGRPB1G tasks for the venue your TR host is currently listed as being in. I'm pretty sure the venue is listed in the startup messages but you can easily see by checking your host details on the website.
It's very hard for others to see what you see so, without lots of information, it can be quite hard to get to the real problem. At least, if the above things can be ruled out, it might help narrow down the field a bit.
I did not have a app_config.xml file in Einstein. I have used it regularly in SETI and in SETI change it regularly to change variables to see which setting gives the best result.s. After I created the file I had Einstein read it and I and saw that it was read in. The strange thing is that it is the same on the 3 computers that I am using the GUPs on with Einstein. I just restarted Einstein on these computers 2 days ago and all came up running 2 WUs at a time (even though they were only running 1 on SETI).
Strange
Note: I had to do a reset on one of the computers and it did not make any difference. The reset was necessary as I qas getting an error when it was trying to get work. (it was a https file not found)
.... After I created the file I had Einstein read it and I and saw that it was read in.
I don't understand what you mean by, "I had Einstein read it". How do you know that, "it was read in"?
If you create an app_config.xml file, it will be the BOINC client (and not the project or a project application) that will read and interpret the file, either on startup or by clicking the 're-read config files' option in BOINC Manager while running. If you click this option, BOINC will check all project directories for the projects that you run.
If there is not one of these files in the Einstein project directory, BOINC will not do anything about running multiple GPU tasks for Einstein. However, if you've ever had a file with that name in the Einstein project directory in the past, it could be the cause of your problem.
Can you please check there is currently no file called app_config.xml in your Einstein project directory? Can you please advise if (even by mistake) you've ever had a file with this name in the Einstein project directory?
I did not have a app_config.xml file in Einstein. I have used it regularly in SETI and in SETI change it regularly to change variables to see which setting gives the best result.s. After I created the file I had Einstein read it and I and saw that it was read in. The strange thing is that it is the same on the 3 computers that I am using the GUPs on with Einstein. I just restarted Einstein on these computers 2 days ago and all came up running 2 WUs at a time (even though they were only running 1 on SETI).
Strange
Note: I had to do a reset on one of the computers and it did not make any difference. The reset was necessary as I qas getting an error when it was trying to get work. (it was a https file not found)
Go under Account, Preferences, Project and near the bottom make sure the las t 3 boxes have 1.0 in them, you will have to click 'save changes' at the bottom of the page if you make any. And you will have to do this for all the venues you are using, ie generic, home, school and work. That setting is an easy way to make Einstein crunch more than one workunit at a time, no cc_config.xml file needed, but it seems yours could be set in error.
It finally sunk in and I was able to see what was happening; I was missing the venue and I was only seeing one venue when I was changing (Or rather looking at the options) and forgot to look at the others). Thanks for the guidance and problem is solved.
One way to change
)
One way to change that:
https://einsteinathome.org/account/prefs/project/
Choose correct Preference set for your host (same as the 'Location' of host). Then at the bottom of the page there are those boxes for "GPU utilization factor of .... apps:" . Default value is 1. That should result in running only 1 WU at a time, per GPU. If the values are something else, try changing them to 1 and save. On the Boinc Manager click an 'Update' for Einstein project. Any new tasks (not yet downloaded) shoud then start running only 1 at a time.
Another way to change that:
Edit or make and save a file app_config.xml in your Einstein project folder. Put these lines in the file:
<app_config><app><name>hsgamma_FGRPB1G</name><gpu_versions><gpu_usage>1</gpu_usage></gpu_versions></app></app_config>Then on the Boinc Manager click 'Options' and 'Read config files'. The effect should be immediate. Ps. '.5' instead of '1' on the gpu_usage would mean 2 tasks per GPU.
Running 2 tasks per GPU will often produce a little bit higher RAC. If there's currently no problem with generating too much heat, noise or electricity (taking a tiny bit more from the wall) running 2 tasks per 1050Ti should be fine.
Richie_9 skrev:One way to
)
Clicking "Update" in Boinc after changing the "GPU utilization factor" is not enough to get the new setting to take effect. One needs to download new tasks for the specific search before the change will take effect, but once it does it will apply to all previous cached work.
Thanks but that is what it
)
Thanks but that is what it has always been set to. I was trying to figure if there was somewhere else that it might have a setting to run 2 WUs at a time.
I am suffering severe lag on the computer when it is running the 2 WUs and was hoping to address that problem by only running 1 WU at a time on the GPUs.
I have already set it to only use 75% of the CPUs. I had a TR with 16/32 cores and 2 X 1050 ti.
Again, thanks for you input but it must be something else.
Bill G wrote:... it must be
)
There are only two places to change how many GPU tasks run concurrently and others have already mentioned both. These are (1) the GPU utilization factor, and (2) the use of an app_config.xml file.
Have you ever used either of these?
If you have previously used app_config.xml but have since just deleted the file, that may well be the reason you can't seem to change back to running single tasks per GPU. Deleting the file doesn't undo critical changes that were made to the state file (client_state.xml) at the time you first used it. If this happens to be the problem, you could reinstate the file and edit both <gpu_usage> and <cpu_usage> to be 1. That will get you what you want.
If you don't want to reinstate the file, you have two options. You could manually edit the state file to fix the issue (not recommended unless you really know what you are doing) or you could reset the project (which would automatically fix the state file but cause all project files - apps, data, tasks, etc - to be deleted and downloaded afresh).
If you have never used app_config.xml, the above doesn't apply. The next question would be, have you ever used different preferences for different 'locations' (previously known as 'venues')?
If you have, please confirm the value for GPU utilization factor for FGRPB1G tasks for the venue your TR host is currently listed as being in. I'm pretty sure the venue is listed in the startup messages but you can easily see by checking your host details on the website.
It's very hard for others to see what you see so, without lots of information, it can be quite hard to get to the real problem. At least, if the above things can be ruled out, it might help narrow down the field a bit.
Cheers,
Gary.
I did not have
)
I did not have a app_config.xml file in Einstein. I have used it regularly in SETI and in SETI change it regularly to change variables to see which setting gives the best result.s. After I created the file I had Einstein read it and I and saw that it was read in. The strange thing is that it is the same on the 3 computers that I am using the GUPs on with Einstein. I just restarted Einstein on these computers 2 days ago and all came up running 2 WUs at a time (even though they were only running 1 on SETI).
Strange
Note: I had to do a reset on one of the computers and it did not make any difference. The reset was necessary as I qas getting an error when it was trying to get work. (it was a https file not found)
Bill G wrote:.... After I
)
I don't understand what you mean by, "I had Einstein read it". How do you know that, "it was read in"?
If you create an app_config.xml file, it will be the BOINC client (and not the project or a project application) that will read and interpret the file, either on startup or by clicking the 're-read config files' option in BOINC Manager while running. If you click this option, BOINC will check all project directories for the projects that you run.
If there is not one of these files in the Einstein project directory, BOINC will not do anything about running multiple GPU tasks for Einstein. However, if you've ever had a file with that name in the Einstein project directory in the past, it could be the cause of your problem.
Can you please check there is currently no file called app_config.xml in your Einstein project directory? Can you please advise if (even by mistake) you've ever had a file with this name in the Einstein project directory?
Cheers,
Gary.
Bill G wrote:I did not have
)
Go under Account, Preferences, Project and near the bottom make sure the las t 3 boxes have 1.0 in them, you will have to click 'save changes' at the bottom of the page if you make any. And you will have to do this for all the venues you are using, ie generic, home, school and work. That setting is an easy way to make Einstein crunch more than one workunit at a time, no cc_config.xml file needed, but it seems yours could be set in error.
Gary Roberts wrote:Bill G
)
That usually means they clicked to have Boinc 'read the config files' in the Boinc Manager under options.
mikey wrote:Gary Roberts
)
That is correct.
It finally sunk in and I was
)
It finally sunk in and I was able to see what was happening; I was missing the venue and I was only seeing one venue when I was changing (Or rather looking at the options) and forgot to look at the others). Thanks for the guidance and problem is solved.