Nouveau Developers Remain Blocked By NVIDIA From Advancing Open-Source Driver

[humbug]

LETS not forget now that AMD has not released their code for the close source Vulkan driver, basically AMD has been keeping their closed drivers, closed also, but they have the luxury of having a functional open-source option to work on.

Well nobody has forgotten the vulkan driver. But AMD is actively working on open sourcing it...

[duby229]

Well nobody has forgotten the vulkan driver. But AMD is actively working on open sourcing it...

That's what they said, but I just don't believe it anymore.

[RussianNeuroMancer]

Yeah, there are already plans to have better coverage on older hardware as well, just a matter of time when we are able to actually start working on this, because it would be quite an investment. There were some talks about CI and driver testing on XDC this year as well.
We really loved to reduce introducing regressions and have a broader way of testing to detect that important features are getting taken care of, as display stuff is for sure.

Good to know! Finally nouveau developers is looking into older hardware stability

By the way, is it possible to perform web-browser accelerated rendering stability tests too? Seems like right now accelerated rendering at least with Blink-based web-browsers is lead to bunch of stability issues, like for first hour or two it could be Ok, but then everything is slowdown, rendering artefacts start to appear on the screen, to the point that machine became barely unusable (due to slowness) or to GPU lockup.
If not, what is best way to report such issues? There is no exact steps-to-reproduce, it's simply something that happen every time if one sit and browse web with enabled hardware rendering in Blink-based web-browser on affected GPU (like one mentioned in my bugreport).

[karolherbst]

Good to know! Finally nouveau developers is looking into older hardware stability

By the way, is it possible to perform web-browser accelerated rendering stability tests too? Seems like right now accelerated rendering at least with Blink-based web-browsers is lead to bunch of stability issues, like for first hour or two it could be Ok, but then everything is slowdown, rendering artefacts start to appear on the screen, to the point that machine became barely unusable (due to slowness) or to GPU lockup.
If not, what is best way to report such issues? There is no exact steps-to-reproduce, it's simply something that happen every time if one sit and browse web with enabled hardware rendering in Blink-based web-browser on affected GPU (like one mentioned in my bugreport).

well, if there is no easy way to reproduce, then it is really hard to debug those things. Maybe a really long running apitrace might show something? Well it would be nice to get more people on board also for stuff like this, because not everybody in the Nouveau community is really interested in fixing those kinds of bugs, because waiting hours for something to happen is super annoying. Getting paid or being highly dedicated to fixing this exact problem would make a difference here.

It is more likely that things like that might be fixed randomly.

[oooverclocker]

it'd be no use contributing to something like the AMD RadeonSI driver - it's backed by full-time developers who are already familiar with the hardware, driver stack, etc., and have the hardware on hand. Not a lot of low-hanging fruit left there for a volunteer part-time contributor. (Not to mention that it uses LLVM for its compiler backend, which is an instant turn-off.)

Honestly, I have the most fun contributing to something that can really change things. I would be quite interested in better Vulkan drivers because once Linux performance is better than Windows performance in all disciplines it would really help the goals I like to achieve in my life because I think they are important for humans. So I can indeed imagine to contribute to Radeon drivers somewhere in the future when I don't have other projects. But I have also projects in my mind that would be more helpful.

When I look at Nvidia or Intel I don't feel like I should spend life time to improve their market share and revenue more in any way. That's just what I personally think is healthy for us.
Yeah perhaps I don't have the most fun in things I do but together with the results I feel more satisfied than when I just life for the fun moment itself.
I don't really care for brands, I look at balances and efficiency of an ecosystem. And I'm definitely ready to give up on statistics of my hardware or bank account to make the real world more the way I like it to be.

So it seems to me like there are always different points of view how people look at their lives.

[ssokolow]

Strictly speaking, nothing. There are some technical hurdles, but nothing strictly impossible. We do exactly this for some of the video decoding engine firmware. The technical hurdles have, however, yet to be surmounted. I just don't think there's a lot of interest in it, and the pool of people who can pull it off is dwindling.

Makes sense. Thanks.

[mitch074]

Stopping rumors is really hard, except for those rare cases where the reality is more interesting than the rumor

Agree that there does seem to be some potential for lowering voltages while maintaining reliable operation, but I'm not sure if all boards shipped that way would work in all expected conditions. I think the direction chosen was making it easier for user to undervolt instead (the whole WattMan thing) but not sure.

Well, I've used AMD boards almost exclusively since the HD4850, and even very early series (both my 4850 and RX480 were reference models dating back to the early production runs, I also got a 7770 GHz edition and a R7 270), and all of them accepted a lowered voltage reliably (in some cases, they were MORE stable with lower voltage). Not to mention CPUs...

So it may be a legend, but it's rooted in truth

I'd really like to know the syntax to reduce voltage for my Polaris, it reaching 89°C really does worry me - 0.02V less does reduce temperature by several degrees under Windows (using Wattman), I'd love to be able to do the same in Linux.

[Sonadow]

On the article, this was mentioned:

"Signed firmwares accessible publicly but not redistributable"

Can't the Nouveau guys tell us where these firmware images are, if they are accessible publicly? Surely we can get them ourselves and then manually drop them into /lib/firmware ? At least that will allow us to get hardware acceleration working, enough to render the desktop environment.

[aht0]

Originally posted by aht0 View Post
This thread makes funny reading.. All the indignant outcries, wide-eyed radicality.. like watching teenage kids bashing the 'hated authorities'..

Give it a decade or two and you might suddenly notice being the target yourself.. undeservedly in your own eyes.

I'd be curious to see how many of you would really care about open-source from the second it would threaten IP your economic well-being depends from. Then you would perhaps see the open-source die-hards as a pack of vultures trying to get for free what you worked your ass off to achieve.. All things are relative, like Einstein proved..
uhhh, well that might be true, but we simply ask to at least help us out with their signed firmware situation. It doesn't change a thing for the IP situation. Also they even published stuff with their NVGPU driver, which we asked for, but never got an answer. So "IP" doesn't really seems to be the problem here, does it?

uhhh, well that might be true, but we simply ask to at least help us out with their signed firmware situation. It doesn't change a thing for the IP situation. Also they even published stuff with their NVGPU driver, which we asked for, but never got an answer. So "IP" doesn't really seems to be the problem here, does it?

The main problem is the attitude. We could have a nice project with their having the stuff their need and we do our stuff all inside the same project.

Then instead of relentlessly bashing Nvidia tell hearty "Thank You" to Chinese scammers who started selling cheap(er) reflashed Nvidia FAKE cards in eBay/AliExpress. You could have bought "770Ti" from Chinese and received fucking "250 GTS" which would still claim itself to be a "770Ti" but support none of the extra features (like hardware encoding H.264) and perform like.. well "250 GTS".

Check out this for example http://www.ebay.com/itm/New-NVIDIA-G...g/201727165907
Want to take a stab at guessing what this card really is? It sure as fuck ain't 1050. It's probably GTX550 Ti with hacked BIOS.

With a problem like this, it's foregone conclusion that manufacturer does whatever it takes to protect itself and it's reputation. If it means complaints and whine of a 1% of PC users, then so be it. It sure as fuck does not want class-action lawsuit because it did not do everything in it's power to protect buyers of it's cards.

[duby229]

Then instead of relentlessly bashing Nvidia tell hearty "Thank You" to Chinese scammers who started selling cheap(er) reflashed Nvidia FAKE cards in eBay/AliExpress. You could have bought "770Ti" from Chinese and received fucking "250 GTS" which would still claim itself to be a "770Ti" but support none of the extra features (like hardware encoding H.264) and perform like.. well "250 GTS".

Check out this for example http://www.ebay.com/itm/New-NVIDIA-G...g/201727165907
Want to take a stab at guessing what this card really is? It sure as fuck ain't 1050. It's probably GTX550 Ti with hacked BIOS.

With a problem like this, it's foregone conclusion that manufacturer does whatever it takes to protect itself and it's reputation. If it means complaints and whine of a 1% of PC users, then so be it. It sure as fuck does not want class-action lawsuit because it did not do everything in it's power to protect buyers of it's cards.

So basically what your claim is that when you have cold symptoms it's better to cut your achilles tendon than it is to blow your nose.