Einstein@Home  Is Back up ATM
 
<app_config>
<app>
<name>hsgamma_FGRPB1G</name>
<gpu_versions>
<gpu_usage>1.0</gpu_usage>
<cpu_usage>0.1</cpu_usage>
</gpu_versions>
</app>
<app>
<name>hsgamma_BRP4G</name>
<gpu_versions>
<gpu_usage>1.0</gpu_usage>
<cpu_usage>0.1</cpu_usage>
</gpu_versions>
</app>
</app_config>
 
 
 

Some call that FM, but glad you made it.

A Week Break  Well Almost.
07/11/2019 23:00 (UTC) - 07/14/2019 22:59 (UTC)
Project to be determined

And the next sprint is...another GPU project?
 

07/11/2019 23:00 (UTC) - 07/14/2019 22:59 (UTC) 
Milkyway@home

1080Ti do pretty well on Milkyway. They can't do 8 at a time, but do ok 2 at a time
I bet the new 20 series cards should do well. Older AMD cards will do well too R9 280, 280x, 7970,  7990, so on.
Shame they are having so many GPU projects in this warm weather. 

Stefem
How much CPU resources should be dedicatet to GPU WU? the default settings is 20% but the GPU seems much faster than CPU

Not worth running CPU Tasks IMO.
Settings I am using this set the GPU to run 8 Tasks at once.
 
<app_config>
   <app>
     <name>milkyway</name>
     <gpu_versions>
         <gpu_usage>0.12</gpu_usage>
         <cpu_usage>0.1</cpu_usage>
     </gpu_versions>
   </app>
</app_config>
 
Setting the GPU to run 3 - 4 Tasks seems to fit best and less Validation inconclusive (343) issues. (Non-DP Cards)
<gpu_usage>0.3</gpu_usage>

mektacular
I'm not consistently getting Milking Way WU's...  Maybe 25% downtime?  I've setup my GPU's to process 4 WU's each to try to keep them busy...

Increase your cache of tasks to 0.5 and an extra 0.3 That will give you a cushion so you do not run out.
I haven't had an issue with that setting.
You can even et it to one day cache and just lower it the last day, or abort the leftover tasks if you don't want to finish them.

Thanks.  Always have the cache at 0 and 0 for primegrid.  Didn't think of changing it...

We are doing pretty good, in 7th, and we are so very close to the 6th place team.
Crank up the overclocks and lets pass them!!!!!

General Error

SQL ERROR [ mysqli ]
Out of memory (Needed 196607933 bytes) [5]
An SQL error occurred while fetching this page. Please contact the Board Administrator if this problem persists.

Computers. GRRRRRRR
cust water loop
i7-8086k, 
16gb G-skill DDR4 3600 
EVGA 1080Ti 
EVGA 850W P2
Samsung 981 250GB NVME. for the OS and some programs
Samsung 961 960GB  NVME  PCIE adapter card in PCIE slot. For games mostly and some important zip files.
OLD regular SATA SSD 256gb Kingston for the OS....
 
Running BOINC, Milky way, everything going good, no issue. I check on this systems today and this box had no work.
Hit update, no work. Check Milkyway site and they have work, my other boxes have work.
So I change the name of the box, uninstall MW, reboot, install MW in BOINC. Still getting no work.
I disable that NVME OS drive slot in the BIOS and hook up the old SATA SSD and boot up, go to BOINC MW not installed.
So I install MW in BOINC and I get tasks, 0.5 days worth like I ask for.
What gives? The SSD is the same as the NVME, all the settings are the same. 
Why did I stop getting tasks and then get more tasks with the other SSD drive? 
I have no idea why, do you?
 

bcavnaugh
No Errors showing?
Maybe Driver missing OpenCL
See if you can run a Collatz GPU Task.
Is the Firewall Blocking BOINC?
 
"So I change the name of the box" The Host Name?
Using a Config File?
 
<app_config>
  <app>
    <name>milkyway</name>
    <gpu_versions>
        <gpu_usage>0.3</gpu_usage>
        <cpu_usage>0.1</cpu_usage>
    </gpu_versions>
  </app>
</app_config>

It was working, then it did have a bunch of computational errors, then I rebooted and it was working fine again.
Yes to the app_config like the posted one. It did thousands of valid tasks. Not sure why it had the errors that one time.
I did back off the OC and slid the voltage slider up.
 
Right now it is crunching away fine. The tasks it is doing are valid, hate to take it back down during the sprint.
When this sprint is over I will remove the SSD and turn the NVME back on.
 
Thanks for the ideas.
 
 

I have no idea, when I looked at the BIONIC MGR. under tasks they were listed as computation error.
If I remember right it is the same driver I use on all my 10 series cards. 396 or 398 something like that.
Under the 4xx driver, higher than the 373

Kicking it Bill

gbaker3
Just running in the Event is enough.