HMM Linux Kernel Support

Can a full list of the Linux Kernels that support HMM be provided as NVIDIA CUDA Toolkit 12.2 Unleashes Powerful Features for Boosting Applications | NVIDIA Technical Blog only states 6.1.24+ or 6.2.11+ (what about 6.3, 6.4, etc.)? Which executables in the extras/demo suite can be used to test HMM? Also, can the executables that are distributed with the demo suite linked against libglut be linked against and not so a symbolic link does not have to be added for it such as in /usr/lib/x86_64-linux-gnu? Additionally, why is the “options nvidia NVreg_OpenRmEnableUnsupportedGpus=1” not automatically added by the current driver .run installer to a config file in /etc/modprobe.d/ since it seems to be necessary to get the open kernel driver to work currently per XFree86/Linux-x86_64/535.86.05/README/kernel_open.html and given that a blacklist file is already being added there for the noveau driver? (The last url should have us. download nvidia com in front of it, but is being blocked because of some stupid one URL per post as this account is deemed a new user.

In general we don’t have a specific list of kernels, but anything after our upstream HMM patch set was applied should work: 6.1.24+, 6.2.11+, or 6.3+. For samples there’s no specific HMM sample, but you can use for example general vector add with malloc() and free() instead of CUDA-specific memory APIs.

Regardling that shouldn’t be the case, is that in a specific sample? We will investigate and get it fixed.

nbody, oceanFFT & randomFog

Thanks for sharing that

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.