Linus Torvalds Calls NVIDIA The Worst Company Ever

[soulman901]

Really bad troll is really bad.

Dude, seriously, that was beyond weak. You can do better'n that.

Who said I was trolling......
Linux sucks.

 

[nOrVow]

People like Torvalds believe everything should be open source and that proprietary ideas and information should simply not exist. That's a reasonable belief if a person is rational, level-headed and empathetic about that viewpoint (even if that seems pretty contradictory). It's not reasonable if your attitude is that of Linus's. If open source software is a religion, then he's the leader of the OSS equivalent to the Westboro Baptist Church.

He spends more time damaging public perception of the Linux community than any other single person in the Linux community, in my opinion. Rather than being a positive beacon, he chooses to be the negative beacon, festering negativity and resentment in the community. I don't hold the guy singularly responsible for the way many Linux people act (arrogant to an extreme degree), but he absolutely isn't helping.

You're talking about Richard Stallman. Linus is a pretty pragmatic person actually who sees the practical reasons for proprietary software (he even support the idea of DRM!), and endorses software developers to do as they please. There's a lot of things Linus doesn't like nor agree with about Richard. He doesn't even like using the word "free" that Stallman endorses so heavily, and goes with "Open" that Eric Raymond coined in Open source.

--I don't think there is an ideology, and I don't think there *should* be an ideology. And the important part of that is the "an" - I think there can be *many* ideologies. I do it for my own reasons, other people do it for _their_ own reasons. I think the world is a complicated place, and people are interesting and complicated animals that do things for complex reasons. And that's why I don't think there should be "an ideology". I think it's really refreshing to see people working on Linux because they believe they can make the world a better place by spreading technology and making it available to people more widely - and they think that open source is a good way to do that. That's _one_ ideology. I think it's a great one. It isn't really why I started doing Linux myself, but it warms my heart to see Linux used that way. But I _also_ think that it's great to see all the commercial companies that use open source simply because it's good for business. That's a totally different ideology, and I think that's a perfectly good ideology too. The world would be a _much_ worse place if we didn't have companies doing things for money. So the only ideology I really despise and dislike is the kind that is about exclusion of other ones. I despise people whose ideology is about "the one true ideology", and not following that particular set of moral guidelines is "evil" or "wrong". That's just small-minded and stupid, to me. So the important part about open source is not the ideology – it's just that everybody can use it for their own needs and for their own reasons. The copyright license is there to keep that openness alive, and to make sure that the project doesn't fragment into people who hide their improvements from each other and then have to re-implement each others changes - but it's not there to enforce some ideology.

http://marianoblejman.com/joomla/in...perating-system&catid=34:tecnologia&Itemid=54

He's made similar statements as such in many other interviews/videos as well.

 

[xorbe]

God damn it Linus, they actually provide working drivers for 99%, don't go burning bridges. Of course video hardware/driver is the most problematic -- it's also the most complicated add-in board.

 

[nOrVow]

Who said I was trolling......
Linux sucks.

Windows is an unholy shit wreck with an 1800s era kernel.

Keep trolling.

 

[8du8]

Linux on the desktop is free. It's also where it matters least. For enterprise and consumer electronics it is not free. Most definetly not for the former.

Linux all over is free, yes some distro's cost money, but to say all enterprise and all consumer electronics that use Linux have a license you need to buy is completely incorrect. Check your facts, because they are wrong.

 

[SkribbelKat]

Windows is an unholy shit wreck with an 1800s era kernel.

Keep trolling.

Windows wasn't very popular until the 1990's. Then again, I'm pretty sure people got upset in the 1800's when their abacus suffered from constant BSODs or was always getting hacked.

 

[Whatsisname]

I'm guessing he is alluding to the silly "free if your time is worthless" adage, as though windows servers somehow don't require administration details of their own.

Linux all over is free, yes some distro's cost money, but to say all enterprise and all consumer electronics that use Linux have a license you need to buy is completely incorrect. Check your facts, because they are wrong.

 

[dandragonrage]

Windows is an unholy shit wreck with an 1800s era kernel

But one that doesn't need to be updated every time I update my drivers... Windows has its own flaws, but it is better in many ways than Linux as well. There is no perfect OS, at least not currently or anytime soon.

 

[temujin987]

right. so, when will Linux take over the desktop OS market? oh, nm...

 

[nOrVow]

But one that doesn't need to be updated every time I update my drivers... Windows has its own flaws, but it is better in many ways than Linux as well. There is no perfect OS, at least not currently or anytime soon.

But one that doesn't need to be updated every time I update my drivers... Windows has its own flaws, but it is better in many ways than Linux as well. There is no perfect OS, at least not currently or anytime soon.

You don't need to update your kernel every time you update your drivers... If a certain chipset isn't supported by kernel version, you can either add kernel modules for it or update to a newer stable kernel entirely for that specific hardware, this is done once. I've never seen an issue of someone having to update their kernel everytime they update their drivers for any device... obviously you're doing something wrong. The kernel itself supports a vast amount of chipsets that are immutable on the motherboard.

Then you have user-space drivers, for removable devices (eg. NICs, GPUs, etc) that done via package manager, or built from source.. again, that doesn't require updating the kernel.

The only thing you would probably have to update constantly are xorg drivers - but that depends.

right. so, when will Linux take over the desktop OS market? oh, nm...

Right, so.. when will Microsoft stop being OEM extortionists? Oh.. nevermind.

 

[Azhar]

"I don't game."

"I like being outrageous at times."

"I like offending people, because I think people who get offended should be offended."

All Linus Torvalds quotes from the end of this video.

And us non-developers are supposed to respect this man? I think he's an arrogant jerk.

How dare Steve Jobs say all that? Who the hell does he think he is?

Oh wait..

 

[InorganicMatter]

Windows changes, browsers change, Adobe flash changes, games change. It's not a linux specific issue by any stretch.

High-level changes happen all the time and are relatively harmless. We're talking about kernel and breaking API changes. Something Linux does all the time, and Microsoft only does with every major Windows release.

Windows is an unholy shit wreck with an 1800s era kernel.

Windows from a user perspective is debatable, but anyone who has programmed a Windows driver can tell their kernel, driver, and plugin architecture is phenomenally stable and sensible. It's light years ahead of Linux's "hard code ALL the things!" and "break ALL the APIs!" system of updating.

 

[nOrVow]

High-level changes happen all the time and are relatively harmless. We're talking about kernel and breaking API changes. Something Linux does all the time, and Microsoft only does with every major Windows release.



Windows from a user perspective is debatable, but anyone who has programmed a Windows driver can tell their kernel, driver, and plugin architecture is phenomenally stable and sensible. It's light years ahead of Linux's "hard code ALL the things!" and "break ALL the APIs!" system of updating.

A lot of penetration testers, security analysts, and malware engineers would say otherwise.

 

[InorganicMatter]

A lot of penetration testers, security analysts, and malware engineers would say otherwise.

Are we talking about Windows 7/8, or 98?

 

[dandragonrage]

You don't need to update your kernel every time you update your drivers... If a certain chipset isn't supported by kernel version, you can either add kernel modules for it or update to a newer stable kernel entirely for that specific hardware, this is done once. I've never seen an issue of someone having to update their kernel everytime they update their drivers for any device... obviously you're doing something wrong. The kernel itself supports a vast amount of chipsets that are immutable on the motherboard.

Okay, you don't need to update the kernel EVERY time you update drivers in Linux, nor vice versa... but fairly often. Drivers ARE tied to at most a few kernel versions. You do often have to update drivers when you update your kernel because of changes in the kernel that seriously happen all the friggin' time in Linux because you have such fragmentation and integration with the video and kernel stuff. And again, it's even worse for developers. Want your hardware to work on Linux? You'd better get in bed (not literally.. well, I mean, I guess you could, but..) with Linus and convince him that your driver should be included in "his" kernel. And then you have to fix it yearly (or more) whenever they break stuff (again - very often).

Hell, I said Windows Vista+7 together have stuck to one driver architecture longer than Linux ever has, but Windows 7 alone - if not having been out longer than Linux ever has stuck with one driver architecture, has at least been out longer than Linux has recently (since 2.6) done so.

 

[nOrVow]

Are we talking about Windows 7/8, or 98?

Well.. the version doesn't really matter. Windows has several fundamental design flaws that are inherent amongst all iterations.

Okay, you don't need to update the kernel EVERY time you update drivers in Linux, nor vice versa... but fairly often. Drivers ARE tied to at most a few kernel versions. You do often have to update drivers when you update your kernel because of changes in the kernel that seriously happen all the friggin' time in Linux because you have such fragmentation and integration with the video and kernel stuff. And again, it's even worse for developers. Want your hardware to work on Linux? You'd better get in bed (not literally.. well, I mean, I guess you could, but..) with Linus and convince him that your driver should be included in "his" kernel. And then you have to fix it yearly (or more) whenever they break stuff (again - very often).

Hell, I said Windows Vista+7 together have stuck to one driver architecture longer than Linux ever has, but Windows 7 alone - if not having been out longer than Linux ever has stuck with one driver architecture, has at least been out longer than Linux has recently (since 2.6) done so.

Use kernel modules.. append certain updates for things that don't need to be updated. Use stable kernels. Do you need it? or do you want want all of the newest shit without considering it's consequences? You have a choice on which to or not to update to prevent breakage. Also, updating to the latest kernel RC is never a good idea if you favor stability - that's why we have mainline and stable releases. I swear this is the same type of complaining I hear from people who use rolling release distro's - yet favor stability, it doesn't make any sense. I've updating many lower levels devices on my PC and have yet to break a single thing - given i actually read and understood what i was doing.

Learn your package manager. You have nothing to complain about here.

 

[BladeVenom]

right. so, when will Linux take over the desktop OS market? oh, nm...

Android has already taken over smart phones, and is a strong number 2 on tablets. Linux is number one on supercomputers, servers, and probably embedded systems.

Linux is winning.

 

[dandragonrage]

Using kernel modules does nothing different. The ONLY difference is that the module isn't compiled into the kernel. Aside from that, they still work the same, and they are still subject to the constant breaking changes introduced into the kernel. Your solution is to just pick a kernel and stick with it for a long time? That is not acceptable to me. Businesses may be content running the ancient 2.6.18 kernel, but I would not even consider running such an old kernel. I can just imagine being on the phone with Microsoft, being told to switch back to Windows XP to get my hardware to work as if that were an acceptable solution.

P.S. I have been using Linux 2.6 since 2.6.0-test9 - just not as a desktop (well, I did briefly, but not for a while now). (I did have some 2.2 and 2.4 experience before that, though not a ton.)

Learn my package manager? My preferred Linux is Gentoo and I can all but guarantee you that I'm more familiar with Linux than you are based on your apparent limited understanding of the Linux driver "ecosystem." I've installed and run the majority of the distros of it at one point or another, too. Not that I am the best, but I do know what I'm talking about.

 

[torquer]

Nvidia, like all companies, puts its development dollars where they'll see the most profit. Linux on the desktop is a small market, so this girl's question about Optimus on her laptop points to exactly that. I don't think theres a good excuse for anyone having crappy drivers on any platform but when you only have X development staff with X development budget, you spend it where you'll see the biggest gain. Right or wrong that is good business sense.

If somehow we're talking about Tegra here, it seems odd to me that if they are the worst company in the world to deal with that they've been embraced so completely by the Android hardware development crowd. Sure there are lots of ARM makers, but within the Android tablet market, Nvidia largely owns the non-Exynos share.

All of that aside, if you're supposed to be a leader in the tech field, you shouldn't be acting like such an asshat. It dilutes the content of your arguments, burns bridges, and makes you look foolish. Theres a right way and a wrong way to deal with stuff like this, and this is the wrong way.

 

[nOrVow]

Using kernel modules does nothing different. The ONLY difference is that the module isn't compiled into the kernel. Aside from that, they still work the same, and they are still subject to the constant breaking changes introduced into the kernel. Your solution is to just pick a kernel and stick with it for a long time? That is not acceptable to me. Businesses may be content running the ancient 2.6.18 kernel, but I would not even consider running such an old kernel. I can just imagine being on the phone with Microsoft, being told to switch back to Windows XP to get my hardware to work as if that were an acceptable solution.

P.S. I have been using Linux 2.6 since 2.6.0-test9 - just not as a desktop (well, I did briefly, but not for a while now). (I did have some 2.2 and 2.4 experience before that, though not a ton.)

Learn my package manager? My preferred Linux is Gentoo and I can all but guarantee you that I'm more familiar with Linux than you are based on your apparent limited understanding of the Linux driver "ecosystem." I've installed and run the majority of the distros of it at one point or another, too. Not that I am the best, but I do know what I'm talking about.

Without modules you would need to rebuild and reboot the kernel anyway. It gives you that modularity without the risk, so it wouldn't matter if it's compiled in or not. There's nothing wrong with using a stable kernel for long time when it just works. Either you want new shit, and risk breaking your system or keep it stable w/ user intervention. If that isn't acceptable to you, then you have no reason to complain, it's a consequence you have to deal with based on user choice.

And since you're a such a Linux scholar (because being a Gentoo users makes you so.. right?), I'm sure you'd know what to do to mitigate the issue.

 

[rihno13]

This just in, nVidia is a horible company all around. They always play dirty, but they do make good hardware. And frankly, as long as their windows drivers work well, I could care less how it runs on Linux (or how easy they are to work with.)

 

[ZodaEX]

How often are you reinstalling? If you're doing it more frequently than the grace period for activation, then just don't activate.

I reinstall on my own machines usually 4 times a year, on two machines, so it's 8 re-installs a year average not counting when I upgrade other people's computers. Linux reinstalls are much much more convenient for me.

 

[torquer]

Also, watching UNIX nerds fight over which distro gives them the most e-peen = lolz.

 

[Kalabalana]

I love this Dunning Kruger forum, apparently nearly everyone not only has more knowledge concerning the issue at hand than Torvalds, but they seem to agree he is wrong!

Nvidia fan boyism? Severe egos? Trovalds might actually be wrong? You decide, I already have.

 

[TheMadHatterXxX]

For those of you don't want to watch the whole thing and only want to see him say fuck you, the question gets asked at 48:12, and he says it 48:57.

 

[devil22]

Well.. the version doesn't really matter. Windows has several fundamental design flaws that are inherent amongst all iterations.

I see Linux advocates post this daily since the first tech web forums, I still have yet to see it backed up with any kind of detailed explanation.

 

[JosiahBradley]

I see Linux advocates post this daily since the first tech web forums, I still have yet to see it backed up with any kind of detailed explanation.

Instead of producing a giant rant like I normally do, I will stay docile and give you just one small example that pisses me off:

Failure to support the basic idea of symlinks that have been around for the longest time on unix-like systems. And before you talk about junction points and the half posix compliant symlinks introduced with Windows Vista, just attempt to use one (dear god the suffering).

I could give you several technical objections to windows, just ask what area and I'll explain.

To the others in this thread, just wow on the Linux hate. And for the nVidia fanboys: AMD has been providing complete specifications for hardware minus trade-secrets for the R300 series and up. It takes them some time sometimes but they are dedicated to at least helping out.

And before I get marked a crazy Linux fanboy (I am), I still use Windows. I am a rather strong power user for both OSes, I just prefer to use Linux for all my platforms, except for gaming, which I enjoy Windows for. I also applaud apple for sticking to the unix paradigm, regardless of my stance on their company tactics.

 

[dandragonrage]

Instead of producing a giant rant like I normally do, I will stay docile and give you just one small example that pisses me off:

Failure to support the basic idea of symlinks that have been around for the longest time on unix-like systems. And before you talk about junction points and the half posix compliant symlinks introduced with Windows Vista, just attempt to use one (dear god the suffering).

?

C:\Users\Dan>mklink /?
Creates a symbolic link.

MKLINK [[/D] | [/H] | [/J]] Link Target

/D Creates a directory symbolic link. Default is a file
symbolic link.
/H Creates a hard link instead of a symbolic link.
/J Creates a Directory Junction.
Link specifies the new symbolic link name.
Target specifies the path (relative or absolute) that the new link
refers to.

 

[JosiahBradley]

?

C:\Users\Dan>mklink /?
Creates a symbolic link.

MKLINK [[/D] | [/H] | [/J]] Link Target

/D Creates a directory symbolic link. Default is a file
symbolic link.
/H Creates a hard link instead of a symbolic link.
/J Creates a Directory Junction.
Link specifies the new symbolic link name.
Target specifies the path (relative or absolute) that the new link
refers to.

Did you miss the part where I spoke about "Half" complaint symlinks? Windows forces you to determine if a symlink is a directory or file outright and only supports a link depth of 31.

A snippet right from wikipedia

Symbolic links are designed to aid in migration and application compatibility with POSIX operating systems. Microsoft aimed for Vista's symbolic links to "function just like UNIX links".[4] However, the implementation varies from Unix symbolic links in several ways. For example, Vista users must manually indicate when creating a symbolic link whether it is a file or a directory.[5] Vista has a limit of 31 symbolic links in a single path.[6] Only users with the new Create Symbolic Link privilege, which only administrators have by default, can create symbolic links.[7] If this is not the desired behavior, it must be changed in the Local Security Policy management console.

In Windows 7 and Windows Vista, when the working directory path ends with a symbolic link, the current parent path reference, .., will refer to the parent directory of the symbolic link rather than that of its target. This behaviour is also found in at least some POSIX systems, including Linux.

So sure you can make them, but they do not work properly and as intended.

 

[devil22]

For real? A link depth of 31 is a fundamental design flaw? First of all, that can probably be corrected if there was any kind of desire whatsoever for it. Second, I have no idea what you're talking about when you say fundamental design flaw, so how am I supposed to ask about an affected system? Here's one anyhow; how about restartable crashed drivers. Last I heard, linux doesn't even support it. A fundamental design flaw in Windows to have such a useful feature I'm sure.

 

[Ashbringer]

I see Linux advocates post this daily since the first tech web forums, I still have yet to see it backed up with any kind of detailed explanation.

#1 Digital signed drivers can cause problems. Microsoft charges to have them signed.
#2 UAC is garbage.
#3 SuperFetch causes more performance loss then it helps.
#4 Massive amounts of hard drive trashing, which just brings my hard drives death sooner then it should.
#5 Readyboost doesn't do anything.

Mind you I can create a bigger list with Linux, but those are the problems that I can think of at the moment for Windows 7.

 

[Red Falcon]

It's funny because they have one of the best drivers in the Linux world. He is just mad because they are closed source.

They are the most stable and highest performing as well.

The closed and open source AMD Drivers are both equally terrible, and the open source NVidia driver is a joke.

As much as I have respect for the man, you are totally correct and hit the nail on the head.
Getting mad/emotional and being unprofessional is not going to aid the situation.

Respect for Linus' honesty and courage though, and for standing up to those NVIDIASSHATS.

 

[ol1bit]

The critical thinking skills of most of the people responding to this thread really make me despair for the human race.



The only person in the thread who has understood the discussion so far.



Umm... a LOT of people, actually.

So the tegra sucks to bad? I disagree. the Asus Transform Prime excels with the Tegra 3.

The guy is an idiot and unprofessional. I could care less what he thinks. Go cry to AMD, see if they care.

 

[JosiahBradley]

For real? A link depth of 31 is a fundamental design flaw? First of all, that can probably be corrected if there was any kind of desire whatsoever for it. Second, I have no idea what you're talking about when you say fundamental design flaw, so how am I supposed to ask about an affected system? Here's one anyhow; how about restartable crashed drivers. Last I heard, linux doesn't even support it. A fundamental design flaw in Windows to have such a useful feature I'm sure.

When did I say fundamental design flaw? I just picked an annoying issue I had recently and one of the reasons I prefer unix like environments. As for the driver issue, In the 8 years I've used Linux I've never had a kernel panic during operation. However for the 8 years I've used Windows I've had more than I can count. I've used experimental kernel drivers in the past with better stability then full fledged windows drivers.

Uptime for my server at work is still 1.5 years and it has only ever been shut off for 10 minutes for a hardware change.

 

[Red Falcon]

Uptime for my server at work is still 1.5 years and it has only ever been shut off for 10 minutes for a hardware change.

Word, Linux ftw.

 

[jwcalla]

I love this Dunning Kruger forum, apparently nearly everyone not only has more knowledge concerning the issue at hand than Torvalds, but they seem to agree he is wrong!

Nvidia fan boyism? Severe egos? Trovalds might actually be wrong? You decide, I already have.

Linux is a train wreck on anything but servers and HPC / clustered systems. And the only reason it has a (slight) majority of the smartphone market is because Google absolutely curb-stomped the GPL when coming up with Android, the success of which has been 100% dependent on the ability to link in proprietary, closed-source binary blobs; the types of drivers Linus and FOSS fanatics get all in a tizzy about. (Well, in fairness, Linus is okay with it, but this outburst is just plain silly.) Google got all tons of shiite from the Linux and FOSS communities when developing Android.

The fact that you can have great desktop experiences on OS X and Windows and not on Linux is because of an inherent flaw in design / philosophy of Linux... not because of Microsoft shenanigans or Apple's magic pixie dust.

 

[devil22]

#1 Digital signed drivers can cause problems. Microsoft charges to have them signed.

Not a universal opinion, I prefer not allowing malware drivers.

#2 UAC is garbage.

Linux is garbage.

#3 SuperFetch causes more performance loss then it helps.

Any benchmarks to corroborate that, or are we to take your word on it?

#4 Massive amounts of hard drive trashing, which just brings my hard drives death sooner then it should.

Google's massive hard drive study showed that hard drive usage doesn't affect life span, you'd gain more by disabling drive sleeping to keep the temps uniform.

#5 Readyboost doesn't do anything.

Don't use it.

Mind you I can create a bigger list with Linux, but those are the problems that I can think of at the moment for Windows 7.

 

[jwcalla]

In the 8 years I've used Linux I've never had a kernel panic during operation.

Get a Realtek R8169 NIC and try to push GbE over the LAN.

100% guarantee of a system hang. Won't even afford you a kernel panic or log error or trace or anything... just an outright frozen system.

Reported years ago... never fixed.

The driver off the Realtek website works flawlessly of course.

 

[devil22]

When did I say fundamental design flaw? I just picked an annoying issue I had recently and one of the reasons I prefer unix like environments. As for the driver issue, In the 8 years I've used Linux I've never had a kernel panic during operation. However for the 8 years I've used Windows I've had more than I can count. I've used experimental kernel drivers in the past with better stability then full fledged windows drivers.

Uptime for my server at work is still 1.5 years and it has only ever been shut off for 10 minutes for a hardware change.

OK, I'll see your anecdote and raise you one of my own. Since I've been running Vista and Win 7, I've never had a software caused hard crash. The only crashes I've seen period were due to a failing GTX 260 and another failing PSU. I tried Kubuntu and Ubuntu not too long ago, both locked up hard as soon as I logged in. I eventually discovered through trial and error (because I couldn't find anything in google), that if I logged in as console and ran 'apt-get update' I could log in without a hard lock. However, Kubuntu was so buggy and unstable with the default apps that there were constant app crashes. KMix was one that I remember crashing everything I logged in. Ubuntu seemed a bit better but only played with it a few hours before I got bored and went back to Windows 7, which required a reinstall of Win 7, because of a bug in the Linux boot loaded installer that caused it to not see Windows installed on another hard drive when in UEFI mode and I eventually wasted so much time in google trying to figure out how to do this that I figured it would be easier just to nuke it and reinstall. My experience in Vista and Win 7 have not been that bad despite using them for years. Windows FTW.

 

[devil22]

Whoops,
everything=every time
Linux boot loaded = Linux boot loader.