Linux 5.9 Brings Safeguard Following NVIDIA's Recent "GPL Condom" Incident

[Zoll]

So no one cared to elaborate what actual impact would this have on us end users?

[oiaohm]

Well, I am of the opinion that the software should be made to run on the hardware. Not that the hardware should be picked to meet the software's arbitrary demands. The distinction might be subtle but it's an important one to make. The hardware enables you to do the things you do on your machine. The OS is just there to facilitate that hardware actually functioning as a well oiled machine.

Since I am from the field of setting up secure audit systems hardware chosen to match the OS.

And, well, again... you try finding an AMD GPU with decent performance that is completely passively cooled. You will not find one. Full stop, end of story. Been there, done that. I tried, I really, really tried. But the fastest passively cooled cards are, without fail, nVidia. And I have my reasons for requiring passive cooling on that front.

It simple really but when in AMD GPU cards its not always the least painful. Fastest passively cooled cards are not always Nvidia. AMD RX470/RX560 you have to use after market cooler/case and the do match up against the 1050Ti.

Fun part is noQaPLvxLF1c you would most likely looking for card with a passive cooler. Where I would be looking at HDPLEX 2nd Gen H5 Fanless Chassis or other equal fanless chassis. These fanless chassis equal remove what ever heatsink solution off the graphics card to connect up the cases heatpipes.

Issue with passive cooling on a graphic card is the increased heatsink weight causes increase sag and other issues of course the case based solutions don't have these problems.

noQaPLvxLF1c basically there are a lot AMD graphics cards that can be totally passive cooled. But I will give you that you don't get them as common as cards preassembled passive cool. Its the horrible order case/passive cooler and card then strip card so it can go into case or connect passive cooler.

Sorry to say Nvidia really does not have that much of advantage in passive cooled either. Lot of cases Nvidia cards with included passive cooler end up throttling lot more than if you strip that cooler off and mount them directly in a true built to be passively cooled case. Then why did you pay extra for the passive cooling heat sink that you were only going to strip off the thing anyhow. I will give you going AMD GPU you don't get the same level of options and are expected for passive cooling not to fear taking apart a GPU.

[skeevy420]

it's your invention. polaris support was solved in a week or two after launch. navi is an outlier

While I've always had stable running cards outside of some of the initial AMDGPU hiccups with my SI card, that can't be said for every GPU model used on Linux that was released sometime in the past 8 or so years. The 270 and 280 and their 3XX versions had a few different people on here talking about how they had issues for years after launch on Linux, ditto with the 290 and some of the Fury cards. While eventually they all got to the point to where they worked just fine OOTB, their first year on Linux wasn't the greatest and there are plenty of bug reports and forum posts from those days to corroborate that.

You gotta remember that half the problem is that they, AMD, were in the middle of writing a new graphics driver in those days when compared to now where AMDGPU might as well be considered feature complete when compared to what Radeon offered on Linux. That means that nowadays they're able to focus more on getting new cards supported and driver upkeep than they are trying to write a new driver and dealing with all the work that entails.

I was there for the entire transition from Radeon/Catalyst to AMDGPU and I can assure you that it wasn't all peaches and cream and that some models just did not get timely Linux support during those years. I can guarantee that almost every Linux user here that experienced those transitions can attest to how "well" it went for them. Most of us get it so while we'll discuss how crappy of times we had, we're not really coming from a place of malice because we know what we had and what we ended up with; the Radeon/Catalyst days to the time of AMDGPU or AMDGPU-Pro (or pieces) when necessary...or Radeon if you're still rocking an old GPU...that you should probably go ahead upgrade so you can join the rest of us in the AMDGPU era.

[Ximion]

So no one cared to elaborate what actual impact would this have on us end users?

None. It will have an impact on people developing open-source Linux code that depends on proprietary code (such as in NVIDIA's kernel module) though.
I highly recommend this LWN article which sums up what happened and why, and why the kernel is changed in this way now: https://lwn.net/Articles/827596/

Also, unlike the Phoronix article suggests, NVIDIA itself wasn't involved in this directly at all.

[Shevchen]

Polaris was a massive success among miners. Among gamers the GTX 1060 has been the most popular GPU ever released so far. It's still number one actually: https://store.steampowered.com/hwsurvey/videocard/

Polaris was also a success because at times (and again btw) it was the cheapest card you can get for stable 1080p 60FPS performance that doesn't require you to put in a months rent worth of money into a GPU. Aside from us rich kids who can go to the next store and just buy whatever we want, there are a lot of people on this planet who have troubles meeting months end. (esp. in current times)

GPUs are not a priority for those folks but once they reach a financial state where it is, they will be (and were) very value concious and thus the success on Polaris. Not that AMD made a lot of money on them because the margins were razor thin. Nvidia however is a different story. They were bought because Polaris (and Vega) was sold out - not because they were "better" for miners. Also: I'm happy the mining boom is over.

[loganj]

Only buy Intel and AMD, since 2012.

and look where that got you

[piotrj3]

I have headache from this article.

First, big minus to phoronix for not saying it was Facebook's patch, not Nvidia's patch.

Second, all people who say fuck Nvidia for Facebook's patch. Great.

Third, then tons of people literally lie, like for example Nvidia doesn't have proper VRR support on linux (they do, I use it, they even did release freesync support on linux before AMD), or that open sourcing driver is not that hard (WTF? you know that AMD open source stuff is literally written from scratch because they couldn't open source their existing stuff?) Nvidia would have to do the same.

Then people claim like fact some really serious AMD issues were fixed 1 year later is not a problem. It is huge problem. It is so huge that due to waiting time, it is more accurate to compare Vegas with Turing, and 5xxx serie with Ampere because it will be closer time frames of actual usability of card.

Then I see something like oiaohm said that it is not problem to passivly cool AMD, oh it is because green team according to Phoronix benchamarks etc is simply more energy efficient, even 12nm Turing is more efficient then 7nm RDNA1. Passivly cooling is limited by how much heat you output and here AMD does have problem.

Then I see Pal666 and I wonder why he is not banned yet for throwing slurs towards someone who posts random "novideo stuff" and links to google with "nvidia bsod". Oh great i can find you BSODs for Intel, AMD, Realtek. As person who sometimes help others, I didn't have anyone asking me to help with BSOD caused by nvidia in last 2 years, but plenty from AMD, even I got one from amd chipset.

Personally the only one with quality drivers is Intel who does it open source, ahead of time, and pretty much without issues for long years. I really hope Intel can make good GPUs, so there will be proper choice here, not closed & good green team and open & bad red team.

About security/backdoors etc. of open source vs closed source. I know people like Gynveal who fuzzed a lot of open source stuff and closed source stuff (pdf readers, stuff like ffmpeg etc.) and it doesn't seem like open source projects are higher quality in that aspect.

[Shevchen]

I said... PASSIVELY cooled.

Again: That is a "you" problem. If the demand on the market for a passive solution would be higher, AiBs would produce more of them. But as the demand is so low, that you only see a few models. And as AiBs want to make money, they produce those models on GPUs that are sold rather well instead of taking any risk and going with option #2. Thats not AMDs problem, nor Linux.

Then I see something like oiaohm said that it is not problem to passivly cool AMD, oh it is because green team according to Phoronix benchamarks etc is simply more energy efficient, even 12nm Turing is more efficient then 7nm RDNA1. Passivly cooling is limited by how much heat you output and here AMD does have problem.

Not true. AMD threw out their efficiency gains on 7nm to (again) throw the cards into redline power levels well beyond their sweet spot. A couple of MHz less and a bit of undervolting results in significant less power draw. A passively cooled card isn't designed to match the performance of a custom watercooled overclocking monster, but to hit the point of maximum efficiency without sacrificing too much performance. A lot of passively cooled cards are underclocked to meet thermal specifications of the cooling block.

If you disregard those parameters, you can make up any argument. Buts thats about it: A made up argument. I can build you a passively cooled 5700XT by tuning the parameters until its power draw is down to 1050Ti levels and it will perform in a region that is still well above the 1050Ti. As of such: Not AMDs problem. AIBs are responsible for designing those.

[piotrj3]

You are not aware then Nvidia also does have pretty good OC margins as well undervolting. Only undervolting monster was Vega, on RX5xx AMD can get much less from tweaking then nvidia and i heard 5xxx serie is same case.

2nd no OEM will go ever for hand written undervolting/underclocking attempts. Literally no one.

3rd. Comparing 1050TI to RX5700 is stupid, and i doubt it would be still more efficient. Now 5600 serie underclocked maybe can compete against it, but bring from Nvidia 1650 and 1660 serie.

[Paul Frederick]

This is how shit companies like nvidia should be dealt.

Nvidia was the first hardware manufacturer that supported Linux. Just in case you did not know.