Folding Forum

My Integrated Intel HD530 graphics started showing up as a GPU slot about a month ago. Searching through some of the forum entries, it looks like that GPU might have been added for testing. It's odd that went to non-beta users. Is there more information on if that test is working and if the Intel slot might be used soon? I'd love to get some more work done on those boxes. I'd love to see it happen. Anyone have more detailed info?

The current system of detecting GPUs means that even during testing/development, the iGPU would be detected for all donors who have the new version of the V7 client. Unfortunately, there's nothing that can be done about it apart from leaving your iGPU Slot as paused. There's no ETA as to if/when it will be supported. Current efforts of the primary FahCore_22 developer is focused on the Moonshot Projects while also developing new features and further optimizations for FahCore_22 version 0.0.12. Plans are to include better debugging and potentially supporting CUDA. Again, there's no ETA apart from soon.

There are plans to improving the GPU detection and making it easier to assign suitable WUs to the correct architecture, it will take a while for that system to be developed, tested, implemented and rolled-out.

Intel UHD 630 integrated GPU is also not working here on my computer : (

avatusxan wrote:Intel UHD 630 integrated GPU is also not working here on my computer : (

At this point, no Intel GPU is supported in production, only Nvidia and AMD.

Someday this might change. Don't hold your breath.

If you have an active slot for an Intel GPU, for now just pause it and add the parameter 'pause-on-start' to that slot and set it to 'true'.

For now a few Intel GPUs are enabled in the GPUs.txt file for internal use by the developers of the GPU folding core. They don't have a separate test system to use on WUs from actual projects, so they also show for regular folders. When or if the Intel GPUs are ready to be supported for general use will be sometime in the future, no idea how long that might be.

An alternate use may be for accelerating some calculations for the CPU folding core. Again that is something being investigated, which approach gets considered for release will depend on how well it works and integrating it into the existing systems. Both approaches will need modifications to the client and server software.

FYI, here's documentation from GROMACS about iGPU/dGPU acceleration: http://manual.gromacs.org/documentation ... ance.html#

However, as Joe_H has stated above, there's no guarantee that this will happen. If it does happen, it may require some significant changes on various levels to ensure that it will be a smooth process for the donors