Mirrored_Repos/LookingGlass
1
0
Fork 0
mirror of https://github.com/gnif/LookingGlass.git synced 2025-04-24 23:56:26 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[doc] requirements: Modify language of DMABUF for iGPUS

Browse Source
This commit is contained in:
Jonathan Rubenstein 2025-02-16 00:54:42 +02:00 committed by Geoffrey McRae
parent b0227a8ff8
commit 32d1b8063e
1 changed files with 8 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
doc/requirements.rst
View File

@ -82,17 +82,16 @@ Linux kernel unless you use the open source NVIDIA drivers.
iGPUs should use DMABUF iGPUs should use DMABUF
^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^
While `DMABUF` with the `KVMFR module <ivshmem_kvmfr>` offers performance While `DMABUF` with the :doc:`KVMFR module <ivshmem_kvmfr>` offers performance
benefits for all users, for the often bandwidth-starved users with an iGPU on benefits for all users, for the often bandwidth-starved users with an iGPU on
their host it's considered necessary for a decent experience. their host it's considered a requirement for a decent experience.
When using a normal SHM file, many GPU drivers will copy incoming frames from When using a normal SHM file, many GPU drivers will copy incoming frames from
shared memory to an intermediary buffer, then upload it from that buffer to the shared memory to an intermediary buffer, then upload it from that buffer to the
GPU's framebuffer. The KVMFR module will instead use the GPU's copy engine to GPU's framebuffer. The KVMFR module will instead use Direct Memory Access (DMA)
download incoming frames directly from shared memory without using an to download incoming frames directly from shared memory, which may depending on
intermediary buffer. This is especially helpful to iGPU users as it frees up RAM GPU design eliminate the intermediary buffer. This is especially helpful to iGPU
bandwidth, which an iGPU already uses extensively. users as it frees up RAM bandwidth, which an iGPU already uses extensively.
An added benefit: since the upload is done with the iGPU's copy engine, the CPU An added benefit: since the upload is done with the iGPU, the CPU load is
load is reduced as the upload is done by the iGPU module rather than the reduced as the upload does not use processor cores.
processor cores.