Closed XinfengZhang closed 10 months ago
In Debian, "non-free" usually means that the source code contains binary blobs that cannot be read directly by humans. Similar to EU shaders, most linux-firmware are also known as "non-free".
IMHO for hardware features, it's pointless to sacrifice functionality for the title "free". One have paid for graphics card, he should be eligible for full drivers.
So users should only choose the "free" version if they are aware of the risks that some hardware features are not available in this driver. Otherwise the "non-free" one is recommended and you don't have to pay anything.
For some other distros, there is another matter. Some of them deliberately disable H.264 and HEVC codecs in the build stage to avoid patent issues.
Maybe the intel-media-va-driver-non-free package would have been better named intel-media-va-driver-proprietary, to indicate there is some closed-source code involved? I think everyone knows the packages are free as in "free beer", but in the Linux/OSS circles it usually indicates free as in "libre".
In Debian, "non-free" usually means that the source code contains binary blobs that cannot be read directly by humans. Similar to EU shaders, most linux-firmware are also known as "non-free".
IMHO for hardware features, it's pointless to sacrifice functionality for the title "free". One have paid for graphics card, he should be eligible for full drivers.
So users should only choose the "free" version if they are aware of the risks that some hardware features are not available in this driver. Otherwise the "non-free" one is recommended and you don't have to pay anything.
yes, fully agree
For some other distros, there is another matter. Some of them deliberately disable H.264 and HEVC codecs in the build stage to avoid patent issues.
yes, we enabled such build option , if they want to release a product without such codec support , such as #1643
which software version is included in the windows driver package? I think that's even worse than the difference between intel-media-va-driver and intel-media-va-driver-non-free.
Intel windows driver get new media-driver and onevpl features without a version bump in the readme files (txt/pdf) - open source media driver for linux and the other platforms has a clear changelog in github and you know what is included in every release/tag - but they are lagging a bit behind compared to the windows version
There's no such issue in Windows world. You are unlikely to get a detailed changelog for Windows driver as it is closed source. Only the MFX/VPL API version is documented.
As far as I know VPL and OpenCL runtime share code between Windows and Linux. UMD on Windows depends on DirectX/DXVA, so it should have a different version than the Linux media-driver.
In addition, WHQL and internal testing process also slow down the release speed of Windows driver.
You are unlikely to get a detailed changelog for Windows driver as it is closed source. Only the MFX/VPL API version is documented.
That's exactly what I mean, for example the big AV1 timestamp issue was fixed between 31.0.101.3802 BETA and 31.0.101.3959 BETA, changelog in both cases consists of many pages but it didn't change for MFX/VPL: Intel® Media SDK Runtime (21.0.1.35) Intel® oneVPL GPU Runtime (21.0.2.7) --> Intel® Media SDK Runtime (21.0.1.35) Intel® oneVPL GPU Runtime (21.0.2.7)
another example would be the calculation of ICQ values for AV1 Encoder, that changed dramatically with that commit: #https://github.com/oneapi-src/oneVPL-intel-gpu/commit/044994c38525e3016b0c9019f122dfde76fb01e6
that was between 31.0.101.3975 BETA and 31.0.101.4032 WHQL, changelog in both cases: Intel® Media SDK Runtime (21.0.1.35) Intel® oneVPL GPU Runtime (21.0.2.8) --> Intel® Media SDK Runtime (21.0.1.35) Intel® oneVPL GPU Runtime (21.0.2.8)
so basically the changelog documents for the windows driver only include "GAMING HIGHLIGHTS" for the current games on windows, everything else is not updated when a change happens - am I right?
We do receive some feedback for the confusion of free
, so the readme file https://github.com/intel/media-driver?tab=readme-ov-file#components-and-features has been updated as below. This issue will be closed. Please feel free to re-open it if any concerns.
About Ubuntu/Debian OSV, they provide [intel-media-va-driver-non-free](https://packages.ubuntu.com/disco/intel-media-va-driver-non-free) (Full feature build) and [intel-media-va-driver](https://packages.ubuntu.com/disco/intel-media-va-driver) (Free kernel build) two packages. Free here means open source kernel but not related to fee need to pay.
Which component impacted?
Not sure
Is it regression? Good in old configuration?
No, this issue exist a long time
What happened?
What's the usage scenario when you are seeing the problem?
Others
What impacted?
I suppose media driver should clarify that people should not pay anything to using intel-media-va-driver. word "free" has different meanings to different understanding, should not use standard open source community understanding to represent all. so , suggest some clarification in README. and from personal perspective, I would like people always use non-free, because it is "free to use..." and with more features.
Debug Information
No response
Do you want to contribute a patch to fix the issue?
None