ADRIConf Remains The Primary GUI Control Panel For Managing Mesa OpenGL/Vulkan Drivers

[bridgman]

(re: CoreCtrl) Or even fork it and make it "official" or something like you do with AMDGPU-PRO?

Not sure I understand. What third party code do we fork and make official in AMDGPU-PRO ?

[angrypie]

Not sure I understand. What third party code do we fork and make official in AMDGPU-PRO ?

https://gitlab.com/corectrl/corectrl

[qarium]

Not sure I understand. What third party code do we fork and make official in AMDGPU-PRO ?

just ship ADRIConfig and CoreCtrl and make it the official driver GUIs.
i bought 6 pices of vega64 in the past and i want to buy 2 big.navi cards for my threadripper workstations.
and i have windows-only friends who switch to linux and they are really annoyed/disturbed that there is no driver GUI.
I personally can perfectly unterstand that it is more imporgant to write the backend of the driver instead of writing a frontend GUI for the driver with limited functionality.
but people coming from windows they are really shocked.

anything from a AMD gui what only shows AMD logo and does nothing to just ship ADRIConfig and CoreCtrl would be better than what we have now.

you also should ship https://github.com/clbr/radeontop or better a graphical version of radeontop
i think radeontop is more important than ADRIConfig and CoreCtrl

but it would be nice to have it all

[_ONH_]

Whats the Point to apply the Logo to existing sw? 0.0, Short zero.

freesync should be togled in the display section of the DE of the game settings.

the problem of standalone App is the noobs won’t finde it.
For non noob’s the Adriconf schould be sufficient, if cleaned from 2-3 visible glitches.

from the user perspective the Apple way should be tanken.
Every single hw togle is integrated into system control of the de, beside display a point called „gpu control“. Then everyone can find ist easily, even if there is just a button which launches adriconf, which seems a better approach than i thougt at first glance. If together with powerupp integrated in that control Center tab proposed hooked into (Gnome 3, kde 4, mate, xfce, lxde) the worst part seems to get it in where ist belongs, to blend into. This means the backend should live in Mesa like adriconf generating all the switch for the installed hw, with the general Layout.
the Apearance should be generated fully inside the de. I am pretty sure, the Mesa dev community knows the People to hit on to get the hooks into the de, if they Seattle for an structure for the to keep maintenanc low with minimal duplication.

the Windows way is awfull every single hw vendor invents the weel anew.

[qarium]

Whats the Point to apply the Logo to existing sw? 0.0, Short zero.

freesync should be togled in the display section of the DE of the game settings.

the problem of standalone App is the noobs won’t finde it.
For non noob’s the Adriconf schould be sufficient, if cleaned from 2-3 visible glitches.

from the user perspective the Apple way should be tanken.
Every single hw togle is integrated into system control of the de, beside display a point called „gpu control“. Then everyone can find ist easily, even if there is just a button which launches adriconf, which seems a better approach than i thougt at first glance. If together with powerupp integrated in that control Center tab proposed hooked into (Gnome 3, kde 4, mate, xfce, lxde) the worst part seems to get it in where ist belongs, to blend into. This means the backend should live in Mesa like adriconf generating all the switch for the installed hw, with the general Layout.
the Apearance should be generated fully inside the de. I am pretty sure, the Mesa dev community knows the People to hit on to get the hooks into the de, if they Seattle for an structure for the to keep maintenanc low with minimal duplication.

the Windows way is awfull every single hw vendor invents the weel anew.

well yes. you are right but you miss one single and simple point: we can have both.

we can have everything we ned in control cenetr of the gnome desktop

and we can have a nice driver GUI from amd with a graphical radeontop +ADRIConfig and CoreCtrl

whats your problem with that ?...

[_ONH_]

but people coming from windows they are really shocked.

To translate what I understood from your writing.
„noob“:
we are shocked that the DE (whatever flavor you like, me: until lately, mine is whatever fedora deliver, since it was the only working dist back when I tried it first. mostelikely I will switch with the next fresh install to something else in their repository) is not like ugly „Winui“, broken to a point where every vendor needs an own GUI, pasted full of graphical poo and adds.

qaridarium and others:
Well, let’s see. How can we fix this in the linux ecosystem. Hm ... , We have an idea let’s pressure the hw vendor to copy the ugliness in Windows and generate vendor specific GUI (based on existing tools) with vendor logo.

me:
Theres a better route. Talk to the UI/DE they know how the DE User like it, give them the api to hook into it, and have a agnostic DE integrated interface.
Ah, you could even tell them there could be benefit to their project if they implement it. Helping getting a bigger userbase, which could result in more contributions.
What of you ask them to help you implement it/how you can plug the thing in reusing existing tool/structures within Mesa.git/Git*. Would it help to acquire some more costumers.
There is no much of point for Mesa involved contributors, to attract lots of user without the ability to make purposefull bug reports, having missconfigured the drivers, unable to get back to a workable system to fetch logs.

response:
Why can’t we habe both a sane and a ugly solution for it. The former Windows user are so accostumed to it, they might feel best to bath in the same ugliness as in Windows. Let‘s still copy it to sneak into them.

me: *facepalm* *sure, totally worth it.*

I hope that helps to understand me.
To sum it up, wasting paid time resources from intelligent people, which are aware of the issue at hand but don‘t see a profit setting odd priorities. The point of having linux, Mesa and gallium is to have vendor independent/common resource. Even Intel finally understood positive to hook in vendor independent structures to ease maintenance.

To propose to slap an amd logo into a tool is so ... wrong ... Inside the linux ecosystem. Yes, you can duplicate every effort like with the de none of the zillion results really took of (compared to the next derivative or win32/winui), meanwhile the projects which focused to integrate anyone contributing&maintaining took off and effectively have no viable alternative. If you want to have a GUI, Integration is key. If the tool is directly accessible itself, fine for me and all which wouldn’t need it but could like it for it’s comfort.

Your message is somewhat right, mostly everything but the marketing/vendor specific ugliness. Your recipient is not the best one to address, not the one with the capabilities, to take it to the finishing line.

[qarium]

To translate what I understood from your writing.
„noob“:
we are shocked that the DE (whatever flavor you like, me: until lately, mine is whatever fedora deliver, since it was the only working dist back when I tried it first. mostelikely I will switch with the next fresh install to something else in their repository) is not like ugly „Winui“, broken to a point where every vendor needs an own GUI, pasted full of graphical poo and adds.

qaridarium and others:
Well, let’s see. How can we fix this in the linux ecosystem. Hm ... , We have an idea let’s pressure the hw vendor to copy the ugliness in Windows and generate vendor specific GUI (based on existing tools) with vendor logo.

me:
Theres a better route. Talk to the UI/DE they know how the DE User like it, give them the api to hook into it, and have a agnostic DE integrated interface.
Ah, you could even tell them there could be benefit to their project if they implement it. Helping getting a bigger userbase, which could result in more contributions.
What of you ask them to help you implement it/how you can plug the thing in reusing existing tool/structures within Mesa.git/Git*. Would it help to acquire some more costumers.
There is no much of point for Mesa involved contributors, to attract lots of user without the ability to make purposefull bug reports, having missconfigured the drivers, unable to get back to a workable system to fetch logs.

response:
Why can’t we habe both a sane and a ugly solution for it. The former Windows user are so accostumed to it, they might feel best to bath in the same ugliness as in Windows. Let‘s still copy it to sneak into them.

me: *facepalm* *sure, totally worth it.*

I hope that helps to understand me.
To sum it up, wasting paid time resources from intelligent people, which are aware of the issue at hand but don‘t see a profit setting odd priorities. The point of having linux, Mesa and gallium is to have vendor independent/common resource. Even Intel finally understood positive to hook in vendor independent structures to ease maintenance.

To propose to slap an amd logo into a tool is so ... wrong ... Inside the linux ecosystem. Yes, you can duplicate every effort like with the de none of the zillion results really took of (compared to the next derivative or win32/winui), meanwhile the projects which focused to integrate anyone contributing&maintaining took off and effectively have no viable alternative. If you want to have a GUI, Integration is key. If the tool is directly accessible itself, fine for me and all which wouldn’t need it but could like it for it’s comfort.

Your message is somewhat right, mostly everything but the marketing/vendor specific ugliness. Your recipient is not the best one to address, not the one with the capabilities, to take it to the finishing line.

you have missunterstanding my person:i am linux only user for like 20 years and i do not care about driver GUI because i use linux anyway with or without driver gui.

you have second missunterstanding to because you are not forced to use the driver gui. if you are happy with the gnome control center options thats perfectly fine.

my argument and you just shot the one who delivers the message... my argument is people coming from windows are shocked about the status in linux. and the result is not to just use windows the result is maybe just use Nvidia who do have a Driver GUI.
this means it is complete insanity that amd does not put in a graphical radeontop +ADRIConfig and CoreCtrl into a driver GUI and put the AMD logo on it.

you see people do not care about your opinion they just buy Nvidia and get what they want.
but this for sure is a big mistake. because it hurts linux as a ecosystem.

[_ONH_]

Sure, to have not everything (a dau could safely config), in a gui hurts linux as an ecosystem. The lack of it is a mistake. But to copy a bloat sceme is wrong.
It is pointles to have upstream driver support and needin an external gui

[qarium]

Sure, to have not everything (a dau could safely config), in a gui hurts linux as an ecosystem. The lack of it is a mistake. But to copy a bloat sceme is wrong.
It is pointles to have upstream driver support and needin an external gui

well your point is not very strong because they just buy nvidia to get a driver gui...
also microsoft is very succesfull in making even more gui for gamers: https://www.pcgamesn.com/windows-10-task-manager
in windows you do not only have perfect driver gui you just have all what gamers want.

your point is like: just don't give a shit about gamers on linux and tell them to use windows and Nvidia.

and this is very sad. but yes "Technically" you are right but most people just don't care they want what they want and they want it now.

[_ONH_]

well your point is not very strong because they just buy nvidia to get a driver gui ...

Quite, possibly. But the point is if your contender sells poo for gold this doesn't mean you should do the same.

also microsoft is very successful in making even more gui for gamers: https://www.pcgamesn.com/windows-10-task-manager
in windows you do not only have perfect driver gui you just have all what gamers want.

No, in Window up to now there is no GUI hence the need of a vendor specific one.
If MS liked they could force NV and AMD to feed an API for an OS GUI within the next 12.x release timeframe, but they don't and moste likely won't.
But like you posted, even Windows came to the conclusion that the gamers need more GPU tools integrated in the DE and not have to mingle with vendor specific garbage. Looks they finally made one step, even if they are 1 - 2 decade behind what's reasonable.

your point is like: just don't give a shit about gamers on linux and tell them to use windows and Nvidia.
and this is very sad. but yes "Technically" you are right but most people just don't care they want what they want and they want it now.

No, my point is have it integrated in the 4 popular DE (starting with gnome/KDE trickling down into the others if successful) or every single relevant distribution install disk. This without adding a single new package to the install disk!

This would mean, get the gui's graphics definition into gnome-control-center/panels/*GPU* (the example is gnome since it is what I happen to use). If I would be bothered enough I would propose a RFC within the relevant gnome (I would know to program c) or kde (which seems to be using c++ ...) discussion channel. But I am not, if I was every single day going by without doing it would be a wasted day. To some degree I don't get why wether Intel nor AMD nor Valve seems to bother.

Keeping the config switch definition inside mesa (or wherever the feature is implemented upstream) in a stable representation form for gui to interface with for products enabled by iris and AMDGPU would prevent duplicating work between the zilion DE. Not just AMDGPU, I don't care which hw runs my os as long as I don't have to mingle with SW, which is not included on the install disk.

This way it would be granted/and forced to be available inside every single distribution ootb 1 or 2 release cycle out.

For the "now" part of your statement, sure this would mean to be a 6 - 12 months waiting period.

What you propose is to have a quirk (to be released by AMD with the cheap irrelevant logo, which ads no functionality). The time the (..)heads at AMD get through to give the issue its required funding, the proposed 6 - 12 month could have passed 2 - 3 times, without having it inside the distribution after this time. I wouldn't expect a move from it before Ubuntu 22.04 LTS at the earliest.

It would moste likely resulting in the same state there is with the pro driver.

Virtually no sane distribution for gaming is supported by this driver.
(Not that difficult since it is just advertised for 3 - 4 LTS distributions, whereas on the gpu side anyone concerned about performance would want to be at something as near as possible to upstream).

This is where moste, non long term linux people I know, would sidestep and go buying Nvidia. If they even were considering buying AMD beforehand.

Or it could even get worse like it is with ROCm. Breaking with update off the os and release, and no one at AMD giving a damn, that it breaks is less an issue than they don't seem to be interested to keep it in a working order or at least trace it downstream.
If I would be to release a SW I would put the .x Versions marked as development releases downstream in the distributions, where LTS I would like to support fork of, to be alert if some external dependencies gonna breaks in the future. It would probably give a hint I would like to maintain my sw longterm, but that probably would turn of the consumer of my software, to not use it at all.
Intel fails similar, with nano it seems but they seem to acting in a timely manner if alerted. Oh, no somehow they jumped to another ship their 3rd in just a few years.
What NV is concerned, consistency seems the biggest pro, not a silly controller center, while the contender sink the ship every other year.

EDIT:
Ups, they build Rocm 3.8 and one can finde the associated repo is somewhere buried inside the install instruction of the LTS releases. But they provide 2.0 inside my distribution, hm ... does it even run? Doesn't RH care the repo they fork their enterprise Linux of has probably rotting, unmaintained packages inside new releases (F33)? No, one nows. So no wonder no one uses AMD SW, if on first contact it leaves the impression to be not maintained?
Does AMD even have the expertises in-house to pull it through, to advise their newborn oss developer to behave like they do with general graphics? Or do they just drop sw like one is used to in CS space? Then why even bother to os it? Too many questions pop on the longer I think at it.

.

@Qaridarium
Do you think such a company hears your call? I mean as a company, not the available representative which sure care about the linux users and their user experience, but the company gives them other priorities to tackle first. I don't think so.