Intel releases patches to add Linux Kernel support for upcoming dedicated GPU releases

Last week, Intel released a big patch series to introduce the concept of memory regions to the Intel Linux graphics driver, which is being added to the Intel “i915” Linux kernel DRM driver. Intel stated that these patches were in “preparation for upcoming devices with device local memory”, without giving out any specific details of these “upcoming devices”. It was in December 2018, that Intel had made its plans clear that it’s working on everything from integrated GPUs and discrete graphics for gaming to GPUs for data centers. Fast forward to 2019, Intel is now testing the drivers required to make them run.

Phoronix was the first to speculate that this device-local memory was for Intel’s discrete graphics cards with dedicated vRAM; expected to debut in 2020. Specifying their motivations behind the release of the new patches, Intel tweeted:

Amongst other features once implemented, the patches will allow a system to:

• Have different “regions” of memory for system memory as for any device local memory (LMEM).
• Introduce a simple allocator and allow the existing GEM memory management code to allocate memory to different memory regions.
• Providing fake LMEM (local memory) regions to exercise a new code path.

These patches will lay the groundwork for Linux support for the upcoming dedicated GPU’s. According to Phoronix’s Michael Larabel, “With past generations of Intel graphics, we generally see the first Linux kernel patches roughly a year or so out from the actual hardware debut.”

Twitter users have expressed enthusiasm towards this announcement:

You can head over to Freedesktop.org to have a look at these patches.

Read Next

Researchers prove that Intel SGX and TSX can hide malware from antivirus software
Uber releases AresDB, a new GPU-powered real-time Analytics Engine
TensorFlow team releases a developer preview of TensorFlow Lite with new mobile GPU backend support