NVIDIA Pushes 62MB Of GSP Binary Firmware Blobs Into Linux-Firmware.Git

[qarium]

Im literally using it now you absolute mongoloid and it is working perfectly fine for me, intel is addressing the sparse residency support and I think parts of it have even landed? not too sure on that one. However maybe if you had less of a hate boner for me, you could post something that actually makes sense for once.

no shit I get aggressive when you constantly actively lie about me

you did say this multible times: "Im literally using it now (...) working perfectly fine for me"
and then i find you in a Intel forum threat writing about features your driver does not have and you need this feature.

just remember i also did buy a intel ARC A380 and i had to accept that the driver for this card did not have important features we needed.
my mother in what computer we used this card did run in 1024x800 pixel mode for month because the driver and plain and simple not in fedora 36/37
for this fact alone everytime i check my mothers computer and her reports i can see what is the driver status of intel drivers.

"intel is addressing the sparse residency support and I think parts of it have even landed?"

sounds good but as long as it is not landet for sure i count this as a miss. like the 1024x800 pixel mode on my mothers computer.

"However maybe if you had less of a hate boner for me"

i do not have any hate for you... i just follow you and read what you write and by this alone i can keep trak of what intel does with their drivers.
and keep in mind i have a intel ARC A380 so its not because i hate intel. it is more like intel is at a junior partner level in the GPU market.
this means they are beeing observed ... and you just need to check the lastest phoronix.com performance benchmarks on that matter intel is absolute low-end you can nearly find any AMD or Nvidia card with less performance than the biggest intel card A770...

"no shit I get aggressive when you constantly actively lie about me"

i did not lie about anything i just follow you and read what you write and every time a intel forum topic pop up you report in the forum what feature intel needs to implement...

and this my friend contradicts to what you claim about the intel driver. if your intel driver is done and finish you would never report any missing feature in the forum in a intel topic.

[Quackdoc]

snip

you absolute do lie about me, you spin things I don't say and make it seem like I have said, or even implied them. I've always said that intel is missing features, For some people, these features are crticial issues, and for others they aren't, my point has always been, that I think the intel arc cards are a good value if they either do everything you need them to, or plan on getting them as an investment as to what they can be in the future.

For me and for many other people, sparse residency is the major feature that intel cards are lacking, (there is also an issue with gamescope when you run it without norbc debug option, seems like not everyone is hitting this though). However intel has shown that they are quite commited on working on these issues.

for the sparse residency support issue, It hasn't been merged yet, but is making steady progress https://gitlab.freedesktop.org/mesa/...requests/25512 and it is quite close to merge, it probably won't make the next mesa cycle unless they really haul ass. but I have no doubts it will make the one after that.

and then i find you in a Intel forum threat writing about features your driver does not have and you need this feature.

And i can spend all day complaining on forums about features polaris doesn't have like VK_EXT_image_drm_format_modifier on polaris, poor compute support via rocm, etc.

and this my friend contradicts to what you claim about the intel driver. if your intel driver is done and finish you would never report any missing feature in the forum in a intel topic.

again more lies about me. Never claimed that.

[avis]

qarium,

Are you alright mate?

NVIDIA driver largely shares its kernel driver code between Linux, FreeBSD and Windows.
AMD has a totally separate Linux only kernel driver.

Exactly what I said earlier and something which you continue to argue about. You've contrived so much BS you kinda lost the plot completely. LMAO.

Mind that in English there's an an action of proving, e.g. "to prove" and there's an object, e.g. "a proof" - you could use word forms properly.

Also, the first letter of a sentence is normally capitalized.

In addition, a period is placed at the end of sentences.

Lastly, punctuation marks such as , . and : normally follow words preceding them without spacing. You could use punctuation properly if you have a modicum of respect which you've never had considering how insulting most of your posts are.

[oiaohm]

NVIDIA driver largely shares its kernel driver code between Linux, FreeBSD and Windows.
AMD has a totally separate Linux only kernel driver.

Lets get details wrong. amd open source driver is not only for Linux.

https://wiki.freebsd.org/Graphics#DRM

bridgman typed.

You may be thinking about fglrx, where we shared substantial driver code between Linux and Windows (and a couple of other OSes). We got a whole lot of flak for sharing driver code across OSes, by the way, and in general our customers seem much happier with a Linux-specific kernel driver.

bridgman wrote "Linux specific kernel driver" and you have read this as "Linux only kernel driver" not the same thing Avis. A Linux specific kernel driver can be ported to other OS just like a Windows specific kernel driver can be. So that text does not say only.

There is a problem here avis why would customers not like the 3 way Linux, FreeBSD, Windows but the customers be happier with the Linux/BSD mix instead.

There must be issue why customers are happier with the other option.. What is something Unix classed as a feature for decades. We are talking the 1979 Chroot invention the feature that the historic base of jails and cgroups/containers.

Stable interface between userspace and kernel space for kernel driver is critical to be able to use userspaces out of alignment with each other. Yes the Nvidia driver requirement of the user space side of driver and the kernel space side of driver lining up does not match how Unix/BSD/Linux systems are designed like it or not.

The reality here is Nvidia has taken a Windows specific kernel driver and ported it to other platforms this leads to problems like not obeying Linux/Unix/BSD platform classed good behavior. . Linux/BSD/Unix platform classed good behavior is very close to each other.

Windows you don't have to consider the chroot/container problem serous-ally. Nvidia not having accelerated xnest/xepher is another sign of not supporting Linux/Unix/BSD classed good behavior.

By the way Nvidia new open source driver

https://developer.nvidia.com/blog/nvidia-releases-open-source-gpu-kernel-modules/

Yes the one mentioned here is not made by straight up open sourcing the Windows only bits. Instead there is a lot of rewriting to match Unix/Linux/BSD based norms.

AMD has been just a lot faster on the take up that taking a Windows driver and making it work on Linux and FreeBSD was a mistake.

fglrx had the same problems of not working will with chroot/containers/jails and not working with xnest/xepher and other common defects as Nvidia drivers have had this complete time.

[tabicat]

By the way Nvidia new open source driver

https://developer.nvidia.com/blog/nvidia-releases-open-source-gpu-kernel-modules/

Yes the one mentioned here is not made by straight up open sourcing the Windows only bits. Instead there is a lot of rewriting to match Unix/Linux/BSD based norms.

I think you mean "Linux" instead of "Windows" here. The Nvidia open source driver is just a rebuild of the proprietary driver, with Windows stuff and a few other things stripped out.

[qarium]

you absolute do lie about me, you spin things I don't say and make it seem like I have said, or even implied them. I've always said that intel is missing features, For some people, these features are crticial issues, and for others they aren't, my point has always been, that I think the intel arc cards are a good value if they either do everything you need them to, or plan on getting them as an investment as to what they can be in the future.
For me and for many other people, sparse residency is the major feature that intel cards are lacking, (there is also an issue with gamescope when you run it without norbc debug option, seems like not everyone is hitting this though). However intel has shown that they are quite commited on working on these issues.
for the sparse residency support issue, It hasn't been merged yet, but is making steady progress https://gitlab.freedesktop.org/mesa/...requests/25512 and it is quite close to merge, it probably won't make the next mesa cycle unless they really haul ass. but I have no doubts it will make the one after that.
And i can spend all day complaining on forums about features polaris doesn't have like VK_EXT_image_drm_format_modifier on polaris, poor compute support via rocm, etc.
again more lies about me. Never claimed that.

i don't think you can compare your polaris problem to the intel one because polaris is 6 year old EOL (end of life) hardware and the intel hardware is very new.

but there is a possibility that for Polaris/Vega the situation with the MESA/RADV/ACO driver stack will become better because now they have one competitor less with AMD abolish the AMDVLK for Polaris/Vega.

as far as i know ROCm can still be enabled for the Polaris series.

[qarium]

qarium,
Are you alright mate?
NVIDIA driver largely shares its kernel driver code between Linux, FreeBSD and Windows.
AMD has a totally separate Linux only kernel driver.
Exactly what I said earlier and something which you continue to argue about. You've contrived so much BS you kinda lost the plot completely. LMAO.
Mind that in English there's an an action of proving, e.g. "to prove" and there's an object, e.g. "a proof" - you could use word forms properly.
Also, the first letter of a sentence is normally capitalized.
In addition, a period is placed at the end of sentences.
Lastly, punctuation marks such as , . and : normally follow words preceding them without spacing. You could use punctuation properly if you have a modicum of respect which you've never had considering how insulting most of your posts are.

I vote to remove any windows shared code from linux for example AMDVLK we just don't need it.

[qarium]

Lets get details wrong. amd open source driver is not only for Linux.

https://wiki.freebsd.org/Graphics#DRM

bridgman typed.
bridgman wrote "Linux specific kernel driver" and you have read this as "Linux only kernel driver" not the same thing Avis. A Linux specific kernel driver can be ported to other OS just like a Windows specific kernel driver can be. So that text does not say only.
There is a problem here avis why would customers not like the 3 way Linux, FreeBSD, Windows but the customers be happier with the Linux/BSD mix instead.
There must be issue why customers are happier with the other option.. What is something Unix classed as a feature for decades. We are talking the 1979 Chroot invention the feature that the historic base of jails and cgroups/containers.
Stable interface between userspace and kernel space for kernel driver is critical to be able to use userspaces out of alignment with each other. Yes the Nvidia driver requirement of the user space side of driver and the kernel space side of driver lining up does not match how Unix/BSD/Linux systems are designed like it or not.
The reality here is Nvidia has taken a Windows specific kernel driver and ported it to other platforms this leads to problems like not obeying Linux/Unix/BSD platform classed good behavior. . Linux/BSD/Unix platform classed good behavior is very close to each other.
Windows you don't have to consider the chroot/container problem serous-ally. Nvidia not having accelerated xnest/xepher is another sign of not supporting Linux/Unix/BSD classed good behavior.
By the way Nvidia new open source driver

https://developer.nvidia.com/blog/nvidia-releases-open-source-gpu-kernel-modules/

Yes the one mentioned here is not made by straight up open sourcing the Windows only bits. Instead there is a lot of rewriting to match Unix/Linux/BSD based norms.
AMD has been just a lot faster on the take up that taking a Windows driver and making it work on Linux and FreeBSD was a mistake.
fglrx had the same problems of not working will with chroot/containers/jails and not working with xnest/xepher and other common defects as Nvidia drivers have had this complete time.

very good explanation thank you. i liked your post. i always had the feeling that this new opensource nvidia driver will not be the gold we waited for.

[bridgman]

It gives me a warm fuzzy feeling when my words are used to prove both sides of an argument

[qarium]

It gives me a warm fuzzy feeling when my words are used to prove both sides of an argument

i told you many times this AMDVLK driver confuse the users. we have a lot of noise just because people get confused by the driver option they have.

AMD should think about this fact: it is not only about how good hardware works or how good the driver works you can add a lot of unwanted noise and confusion by just offer 2 different open-source drivers. even if both use the same kernel part. and even if of course you can use both drivers at the same time each for what it is best for.

a good experience with hardware and software also has a psychology part and as you see a lot of people are just confused by to much options.

you can also see this in the news when AMD did abolish the AMDVLK driver for Polaris and Vega people become angry and have negative feelings even people who did never use AMDVLK they always use RADV instead but they do not know the different and believe they are not affected because AMD did abolish the opensource driver.

this is all nonsense... honestly and we have this kind of noise for many years now.

AMD maybe believe to really support the RADV driver with developers is more expensive this is maybe true but they never calculate the loss because of negative press and negative forum posts from people who just are confused in the AMDVLK vs RADV thing.

AMD should value this PR negative press/negative forum posts as money lost... and better abolish this AMDVLK driver.