Message boards :
Number crunching :
New Computer, Unable To Get GPU WU's
Message board moderation
Author | Message |
---|---|
Send message Joined: 6 Mar 10 Posts: 4 Credit: 37,798,404 RAC: 0 |
I used to participate in this project until my old computer system gave up the ghost in January, and was unable to contribute again until building my new current system in September, and I have not been able to get a single Milkyway@home work unit since returning. I have tried resetting my project on BOINC several times, and I am able to use my nVidia 1060 GTX on Asteroids@home units - joined in lieu of being unable to resume Milkyway@home - with no issues at all. And yet every attempt I make at getting Milkyway@home work yields the same results: Sun 11 Nov 2018 07:46:56 PM EST | Milkyway@Home | Sending scheduler request: To fetch work. Sun 11 Nov 2018 07:46:56 PM EST | Milkyway@Home | Requesting new tasks for NVIDIA GPU Sun 11 Nov 2018 07:46:58 PM EST | Milkyway@Home | Scheduler request completed: got 0 new tasks I don't know what steps to take to try and solve my problem, as researching previous messageboard threads regarding similar issues seem to be solved with a "wait for it" approach, but seeing as how I've been waiting for more than a month at this point, I think there might be something else preventing me from getting any work units. Any help with this would be greatly appreciated. |
Send message Joined: 24 Jan 11 Posts: 708 Credit: 543,880,978 RAC: 126,061 |
Are you loading any OpenCL driver for MW? That is required. I think you loaded the Nvidia drivers through the ppa. The ppa install misses the OpenCL drivers sometimes unless you passed through the 390 series first. Try loading the OpenCL component separately. If you don't see the OpenCL component listed for your 1060 in the beginning startup of the Event Log, that is the reason you are not getting work. sudo apt-get install ocl-icd-libopencl1 |
Send message Joined: 6 Mar 10 Posts: 4 Credit: 37,798,404 RAC: 0 |
Thank you very much, Keith! That simple little fix did the trick for me. I had no idea that OpenCL was dropped from the current nVidia PPA drivers, nor was I aware in turn that Milkyway@home required them. Hopefully this information can possibly help someone else in a similar situation in the future. :) |
Send message Joined: 24 Jan 11 Posts: 708 Credit: 543,880,978 RAC: 126,061 |
A lot of projects use OpenCL applications now. Unfortunately, that is the component that is most often dropped by Windows. Anytime Microsoft updates the graphics drivers with its own drivers, it is missing the OpenCL component. There are massive number of posts in project forums about people wondering why a perfectly well running computer doing distributed computing all of a sudden isn't crunching gpu tasks anymore. The same is true with a lot of Linux distributions too. Ubuntu being the worst offender. The only way to be sure of getting the OpenCL component with Windows is to load the graphics drivers from the card vendor. But that only works with Nvidia. AMD now only offers the OpenCL component as a separate download outside the normal graphics drivers package. The ppa hasn't offered the OpenCL component in the package since the 390 series as far as I can tell. Thankfully the OpenCL driver hasn't changed since the 1.2 version. A good sanity check to be sure both the CUDA and OpenCL components are loaded is to run clinfo. That will print out all the graphics drivers capabilities it finds on the host system. sudo apt install clinfo |
©2024 Astroinformatics Group