Hacker News new | past | comments | ask | show | jobs | submit login
NVIDIA and nouveau (lwn.net)
196 points by pabs3 on Oct 6, 2022 | hide | past | favorite | 35 comments



It seems that the only benefit of moving the proprietary driver into firmware in the near term is to reduce friction for distros. Canonical or RedHat may be able to provide better out of the best experience to their customers. Doesn't change things too much otherwise. Will be a while before anything is mainlined.

Increasingly, open source for hardware just means shoving things into firmware. Holds true for CPUs, GPUs, and wifi cards.


"Just" shoving things into firmware at least means the closed-source code runs on a different CPU, often (but not always) isolated from main memory, often (but not always) with no network access, which the main CPU often (but not always) has the ability to power-cycle.


If it makes the hardware work to its design everywhere, reliably, and offloads complexity from the CPU, isn't that all good things? It's not like Nvidia is going to open source their silicon.


Valve as well.


I just got a Steam Deck. My hope was a PC gaming experience that worked with less than 5 minutes of fiddling to get a game running. I figured it would be disappointing. The UI is buggy and the first hour was rough. Once everything settled in, though, it's been a pleasant experience.

The nicest, best thing is that the suspend button just works flawlessly. I can imagine the only reason that's the case is because Valve had control of the kernel drivers and firmware. Maybe they had to sign NDA agreements with their vendors to do that work. If their changes trickle down into the vendors' firmware blobs for consumer facing hardware, they'll have compensated a bit for the technical debt of other companies, made the world a bit better.


I really wish nouveau were in a better state. I have a positively ancient laptop with a 8600M nvidia chipset that just will not reliably operate with the nouveau driver.

Instead, I have to use https://github.com/MeowIce/nvidia-legacy to patch and install the proprietary drivers.


Wow, thanks for this. I have an aging Thinkpad W510 which could benefit from this.


Some more discussion and clarifications about this at https://github.com/NVIDIA/open-gpu-kernel-modules/discussion...


>> GPU may allow creating a cross-platform compute stack

That was the intention of OpenCL.


except writing in opencl is a dogshit experience compared to cuda. khronos is incapable of producing an API that's even remotely pleasant to use, vulkan suffers from the same thing. it is one of the worst "design by committee" groups.


I think vulkan is intentionally to be like that. It's by design a `not so abstract` abstraction that allow you to extract most power from the device while keep your most setup the same.(you still need to handle different vendor device specifically) It never mean to be a `write once and behaves the same on every vendor device` API.

That isn't what vulkan mean to do. If you want a experience like that, opengl is the way to go.(At the cost of performance)


OpenGL is officially deprecated on all of Apple's platforms, and seems unofficially deprecated from Kronos; the last update to the standard was 4.6 in 2017. It will keep working (at least on most non-Apple platforms) for a long time still, but it's not going to necessarily give you access to everything your hardware can do.

That's not to say it's not the "way to go", but it's not in a great state.


The more ergonomic APIs with similar capabilities are Metal and it’s near-clone WebGPU.


I think the simplicity of metal(and the apple version webgpu, now webmetal) relies on the fact that it only has to support apple hardware (With much less edge case need to handle).

I would be more curious about if the whatwg version webgpu can still be simple like that and has similar performance like the original one apple proposed.

Besides that, I think DX12 is a more proper competitor about access low level primitive from devices from different vendor.


+1 to that. Nvidia isn't winning the AI accelerator race because it's making obviously superior products to AMD/Intel. It is winning because CUDA is easy to use and reason about.

OpenCL's API is terrible to use and even if you know all of it's kinks there is much more boilerplate which turns a small script into a tangled mess.


Reading this only reinforces my conviction to go with an AMD gpu for my next PC build (my last build was ~ 20y ago with an ATI 9700). Nvidia doesn't seem to be a good citizen in the linux ecosystem.


I recently did, it has been mostly a good transition.

The only thing I am slightly miffed about is I couldn't get recording working in HDR games. The colors come out completely ruined. This is on a windows pc.


> I couldn't get recording working in HDR games

Use latest obs with Color space Rec2100 in advanced tab in main setting and Color space Rec2100 selected in game recording setting.

That works perfectly on my machine when trying to record forza horizon 5 with hdr enabled.

(The didn't work for streaming though, And it is probably not possible because rtmp didn't have support for it)


firmware = simple, but not simpler than required to do a good enough job, hardware programming interface on complex chips.

You need software jingle room to fix things on complex chips, which is very true for SOC and GPU and now CPU, hence one of the main reasons for firwmare to be. But going beyond that sole purpose of presenting a hardware programming interface is a slippy slop, for instance the CPU user digital jails.

Many simpler chips can expose a hardware programming interface which does not require a firmware though.

I have not read on what is the nvidia hardware programming interface is since now we have open source code, but rumors say it is orders of magnitude simpler than for AMD GPUs (which is still kind of a mess).


Probably in the minority here but I've recently had a better experience with nouveau than the proprietary nvidia driver on a laptop with the external video outputs wired to the nvidia gpu. The proprietary driver doesn't work well with sway, and recently nouveau got support for powering down the discrete gpu when not used, meaning I still get good battery life instead of it draining my battery ~4x as fast as with it disabled.


I have a Kepler laptop gpu using nouveau and the 2d experience is great. Even after reclocking it’s only just as performant as my haswell i7 iGPU with worse OpenGL support. That being said all I play is Minecraft and Stardew so it doesn’t bother me much.

I will be happier in the future getting better support. Either that or getting a newer laptop!


I recently upgraded from a laptop that also had Kepler + i7 Haswell iGPU, to a laptop with an 11th gen Intel iGPU, and the graphics performance of the new laptop is way better than the old one, even compared to running the proprietary NVidia drivers on the old laptop. I'd definitely recommend upgrading.


thanks! honestly all i use this for is personal web browsing, minecraft, stardew, and a bit of brutal doom to blow off some steam :). for me its perfect and i got it maxed out at 32GB of ram and the quad core i7.

one day i'll upgrade, maybe when all the kids are in public school and my wife has graduated law school. right now i'm paying a pre school and a law school tuition on a single income (as well as mortgage etc.). i would be excited about an amd framework laptop. i guess i will have to wait until they get their usb stuff straight though.


I bought a Kepler GPU in 2012, and it's been my most valuable technology purchase ever. It can even handle 3D game development in Godot. I only run free software on my desktop, including firmware, and nouveau works really well. Now I see that Kepler is even in-scope for the potential next-generation drivers, which is really considerate of the developers toward all the people who have stuck with it to avoid signed firmware.


We have been asked a few times not to post SubscriberLinks to hackernews, lwn submissions tend to make their way into the public anyway, with a 2 week delay for non-subscribers.

Continuing to post these articles has some potential for financial harm to the project and may prevent such articles being written in future.


I don't remember ever seeing anyone ask to avoid posting LWN subscriber links to HN. Actually, a few months ago Jonathan Corbet stated the following on this site:

"It's nice to see LWN on HN for the second time in one day, but please remember: it is only LWN subscribers that make this kind of writing possible. If you are enjoying it, please consider becoming a subscriber yourself — or, even better, getting your employer to subscribe."[0]

It sounds like Mr. Corbet is fine with the occasional link as long as people keep LWN's business model and financial needs in mind. LWN is an excellent resource, I highly recommend subscribing.

[0] https://news.ycombinator.com/item?id=31851755#31852477


On the contrary, "we" have been repeatedly told by Jonathan Corbet, that posting to HN is perfectly fine. Stop making things up and creating Internet drama.


> It's nice to see LWN on HN for the second time in one day, but please remember: it is only LWN subscribers that make this kind of writing possible. If you are enjoying it, please consider becoming a subscriber yourself — or, even better, getting your employer to subscribe.

https://news.ycombinator.com/item?id=31852477

sounds like he is fine with it as long as it comes with subscriptions, which makes sense. Maybe my suggestion should have been to subscribe and I wouldn't have been flagged. :)

EDIT: can I please get an explanation of what I've done wrong? Is it too meta?


I didn't flag or downvote you so I'm only speculating, but I suspect it was killed/flagged for being misinformation. Many people just skim, and don't verify info before spreading it, so a comment like yours could cause it to spread and in this case may cause friction/drama. Again, just speculating.


> We have been asked a few times not to post SubscriberLinks to hackernews, lwn submissions tend to make their way into the public anyway, with a 2 week delay for non-subscribers.

Who is "we"? There's been numerous threads in LWN posts that state that posting subscriber links to hackernews is in fact encouraged by LWN. So you seem to be stating the reverse of what is general knowledge.


Whilst I largely agree, in this specific case the talk on which the article is based is public.


I thought whenever this is posted someone paid for it to be made public?

'The following subscription-only content has been made available to you by an LWN subscriber.'

Are those really only subscriber links?


The links are public, but hidden, so any subscriber that has a link can share it to any number of people.


The only people I’ve seen ask that have been normal subscribers, not contributors or even owners. So, I feel like you might need to source your statement.


How about, if noone posts these links, noone will read their articles, hence noone will care to pay in the future?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: