Intel Looking To Finally Upstream Linux Driver For Their Gaussian & Neural Accelerator

Found with mobile Intel CPUs across Tiger Lake, Ice Lake, and even Cannon Lake has been the Intel GNA accelerator. This Gaussian and Neural Accelerator is also found with Intel Gemini Lake processors and various development kits. The Intel GNA has been backed by an out-of-tree Linux driver while now the company is finally working to upstream their GNA support in the Linux kernel.

Intel's Gaussian and Neural Accelerator is a neural co-processor that can be used for offloading inference workloads. The GNA on modern Intel laptops has been used for tasks like noise reduction and speech recognition, assuming you setup the out-of-tree Intel GNA kernel driver and various user-space components like OpenVINO.

The existing Intel GNA Linux driver is an out-of-tree, open-source kernel module designed to work just on Ubuntu 18.04 LTS and tested with Linux 5.0 through 5.6 kernels. But getting this driver upstreamed would mean it becomes much easier for Linux users across all distributions to begin tapping the Gaussian and Neural Accelerator, assuming they have compatible user-space software to make use of the interfaces.

Sent out today were the initial set of 12 patches by Intel engineer Maciej Kwapulinski now that they are looking to upstream the driver. The open-source user-space code including samples can be found meanwhile via this GitHub repository for exercising this driver and its interfaces.

The Intel GNA accelerator driver is just under three thousand lines of code at this stage. The initial patches can be found on the kernel mailing list. We'll see how quickly now Intel is able to get the Gaussian and Neural Accelerator support upstreamed to allow for much more widespread use of the GNA on Gemini Lake and Cannon Lake / Ice Lake / Tiger Lake hardware available thus far.