LinuxQuestions.org
Review your favorite Linux distribution.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - Distributions > Slackware
User Name
Password
Slackware This Forum is for the discussion of Slackware Linux.

Notices

View Poll Results: Which driver(s) do you use with your NVIDIA cards?
nouveau (free and open-source) 17 15.74%
nvidia (proprietary) 93 86.11%
Multiple Choice Poll. Voters: 108. You may not vote on this poll

Reply
 
Search this Thread
Old 03-24-2012, 11:10 PM   #31
afreitascs
Member
 
Registered: Aug 2004
Location: Brasil
Distribution: Slackware_Cur-64_mult
Posts: 433

Rep: Reputation: 30

Hi

I use the nvidia proprietary driver. Please do not arrest me, I have nvidia permissions to use it. Not pirated software !
 
Old 03-25-2012, 11:27 AM   #32
jtsn
Member
 
Registered: Sep 2011
Location: Europe
Distribution: Slackware
Posts: 803

Rep: Reputation: 354Reputation: 354Reputation: 354Reputation: 354
Quote:
Originally Posted by cascade9 View Post
Give one example where FOSS drivers have dropped support for a GPU/video adapter.
Most chips from the PCI era (i. e. 3Dfx, C&T etc.) don't work anymore. The drivers are still there, but they aren't tested on real hardware. Another example is the transistion from XFree86 3.3 to 4.0, which killed the support for ISA cards. It is okay, to do that, because unmaintained drivers for extinct hardware are useless anyway.

In the foreseeable future the folks with well-running Matrox or VIA/S3 UniChrome AGP hardware will face similar degradation, because most of the Xorg developers now have Intel or AMD.

Quote:
Originally Posted by H_TeXMeX_H View Post
Although I must say that the performance that was provided by the nvidia blob was the best on Linux, and I seriously doubt anyone can prove me wrong on this, I must say that the stability of these drivers is not good. Even in the past I have had stability issues with certain version. With the newest version I get random crashes. Definitely not what I would call "rock stable". I would say the stability of these drivers is "capricious" at best.
I never had any crashes related to the nVidia Driver, because my Slackware boxes don't crash at all (knocking on wood). The bigger problem with reverse engineered graphics drivers is, that they can damage hardware, not just crashing your machine. So there is a expensive risk in activating them by default - as most distros do.

BTW: The vendor-support problem isn't restricted to graphics drivers. The r8169 driver in the 2.6.37.6 kernel from Slackware 13.37 doesn't work well with either my PCIe 8111E or my PCI 8169S NIC. Replaced it with the official FOSS drivers from the Realtek site and everything is fine.
 
Old 03-25-2012, 01:07 PM   #33
H_TeXMeX_H
Guru
 
Registered: Oct 2005
Location: $RANDOM
Distribution: slackware64
Posts: 12,928
Blog Entries: 2

Rep: Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269
Quote:
Originally Posted by jtsn View Post
I never had any crashes related to the nVidia Driver, because my Slackware boxes don't crash at all (knocking on wood). The bigger problem with reverse engineered graphics drivers is, that they can damage hardware, not just crashing your machine. So there is a expensive risk in activating them by default - as most distros do.
This is very rare and would probably be due to bad hardware design.
 
Old 03-25-2012, 01:09 PM   #34
Keith Hedger
Member
 
Registered: Jun 2010
Location: Devon,UK
Distribution: Linux From Scratch, Slackware64,
Posts: 836

Rep: Reputation: 167Reputation: 167
the nvidia driver 'cos its slightly better and faster and also because although the source is closed the driver **IS** packaged and maintained to run on linux
 
Old 03-26-2012, 06:03 AM   #35
jtsn
Member
 
Registered: Sep 2011
Location: Europe
Distribution: Slackware
Posts: 803

Rep: Reputation: 354Reputation: 354Reputation: 354Reputation: 354
Quote:
Originally Posted by H_TeXMeX_H View Post
This is very rare and would probably be due to bad hardware design.
Using software-controllable clock generators, voltage regulators and fan controllers is not "bad hardware design", it is industry standard hardware design and common everywhere.

Since there is no documentation for nVidia chips, the work on nouveau is based on snooping on the nVidia BLOB and guessing. If some guess goes wrong for a specific graphics card model, it can easily blow up a $300 nVidia card within a second. Or damage it just slightly, so it crashes randomly even with the official drivers.
 
Old 03-26-2012, 08:57 AM   #36
dugan
Senior Member
 
Registered: Nov 2003
Location: Canada
Distribution: distro hopper
Posts: 4,631

Rep: Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429
Quote:
Originally Posted by H_TeXMeX_H View Post
This is very rare and would probably be due to bad hardware design.
It's not "very rare". It's never happened.

The only thing I can think of that could even be analagous is a version of Mandrake that bricked optical drives. That was around a decade ago.

And before you say: "just because you don't remember it doesn't mean it didn't happen": yes it does. When bugs of this magnitude happen, the headlines are impossible to miss.

Quote:
Using software-controllable clock generators, voltage regulators and fan controllers is not "bad hardware design", it is industry standard hardware design and common everywhere.
All of these are outside the scope of what would go into a driver. Separate userspace overclocking software, yes, but not a driver. And not Nouveau.

Last edited by dugan; 03-26-2012 at 09:04 AM.
 
Old 03-26-2012, 01:02 PM   #37
jtsn
Member
 
Registered: Sep 2011
Location: Europe
Distribution: Slackware
Posts: 803

Rep: Reputation: 354Reputation: 354Reputation: 354Reputation: 354
Quote:
Originally Posted by dugan View Post
It's not "very rare". It's never happened.
Well, it already happened with the nVidia driver itself. Got a lot of Windows users with fried cards angry...

Quote:
The only thing I can think of that could even be analagous is a version of Mandrake that bricked optical drives. That was around a decade ago.
Nope, this hardware was defective from the beginning. It just interpreted official documented MMC commands wrong. Trying to write a FOSS driver by poking in unknown and undocumented registers is a completely different thing.

Quote:
All of these are outside the scope of what would go into a driver. Separate userspace overclocking software, yes, but not a driver. And not Nouveau.
Of course, to get your card out of text mode, you have to program the GPU clocks to the correct specified speed for the load profile and adjust voltage accordingly, also you have to monitor temperatures to get the fan to the right speed. This is exactly, what the nVidia driver does and has nothing to do with "overclocking", Nothing of this happens automatically by hardware design. And this is of course, what nouveau tries without knowing, if it's gonna work on your card model. It can work, it can get very loud (fan speed wrong) or it can just overheat your GPU and let hundreds of dollars worth of hardware go down the drain.
 
Old 03-26-2012, 01:09 PM   #38
dugan
Senior Member
 
Registered: Nov 2003
Location: Canada
Distribution: distro hopper
Posts: 4,631

Rep: Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429Reputation: 1429
Quote:
Originally Posted by jtsn View Post
also you have to monitor temperatures to get the fan to the right speed. This is exactly, what the nVidia driver does
Both of my NVidia cards are fanless, so you're obviously wrong. (Well, perhaps not entirely wrong, but I'm petty like this).

Curses to you, though; now you have me curious to look through Nouveau's source code.

Last edited by dugan; 03-26-2012 at 01:34 PM. Reason: I own two fanless NVidia cards. Not one.
 
Old 03-26-2012, 01:50 PM   #39
afreitascs
Member
 
Registered: Aug 2004
Location: Brasil
Distribution: Slackware_Cur-64_mult
Posts: 433

Rep: Reputation: 30
even the "drivers (proprietary) nvidia" can kill a video card.
And this in windows!

http://www.zdnet.com/blog/hardware/w...hics-card/7551

http://www.evga.com/forums/tm.aspx?m=216216&mpage=1
 
Old 03-27-2012, 03:14 AM   #40
Cheesesteak
Member
 
Registered: Jun 2008
Distribution: Slackware
Posts: 100

Rep: Reputation: 24
I just upgraded from Nvidia's 275.43 to 295.33, and so far I haven't had any issues. It feels like forever since there has been a newer release that played nice with my card, particularly when running KDE.
 
Old 03-27-2012, 07:18 AM   #41
H_TeXMeX_H
Guru
 
Registered: Oct 2005
Location: $RANDOM
Distribution: slackware64
Posts: 12,928
Blog Entries: 2

Rep: Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269Reputation: 1269
Quote:
Originally Posted by dugan View Post
It's not "very rare". It's never happened.

The only thing I can think of that could even be analagous is a version of Mandrake that bricked optical drives. That was around a decade ago.

And before you say: "just because you don't remember it doesn't mean it didn't happen": yes it does. When bugs of this magnitude happen, the headlines are impossible to miss.



All of these are outside the scope of what would go into a driver. Separate userspace overclocking software, yes, but not a driver. And not Nouveau.
I completely agree, it has never happened. However, with the criticism I have received lately I dare not say never. I mean it could happen and maybe I missed the headline.

Quote:
Originally Posted by jtsn View Post
Using software-controllable clock generators, voltage regulators and fan controllers is not "bad hardware design", it is industry standard hardware design and common everywhere.

Since there is no documentation for nVidia chips, the work on nouveau is based on snooping on the nVidia BLOB and guessing. If some guess goes wrong for a specific graphics card model, it can easily blow up a $300 nVidia card within a second. Or damage it just slightly, so it crashes randomly even with the official drivers.
I have never seen it happen, so if you have seen it somewhere, link it or post pictures ... otherwise it didn't happen.
 
Old 03-27-2012, 08:12 AM   #42
TobiSGD
Moderator
 
Registered: Dec 2009
Location: Hanover, Germany
Distribution: Main: Gentoo Others: What fits the task
Posts: 15,576
Blog Entries: 2

Rep: Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036
So the proprietary drivers killed some cards, the open (as far as we know it) haven't. I find that rather funny, can we draw the conclusion that the nouveau developers are better in writing reverse engineered drivers than the Nvidia developers in writing drivers with documentation?
 
Old 03-27-2012, 11:44 AM   #43
qweasd
Member
 
Registered: May 2010
Posts: 437

Original Poster
Rep: Reputation: Disabled
Quote:
Originally Posted by TobiSGD View Post
So the proprietary drivers killed some cards, the open (as far as we know it) haven't. I find that rather funny, can we draw the conclusion that the nouveau developers are better in writing reverse engineered drivers than the Nvidia developers in writing drivers with documentation?
My guess would be (and I could easily be wrong) that nvidia code has been "hijacked" by a few core developers who get to see the complete source. They are the indispensable ones who understand how it works, and they resist as much as possible either opening the code or training the replacement. Aside from the docs NVIDIA squeezed out of them at gunpoint, they probably have their own secret and much more complete docs which they won't ever share with anyone. This problem is endemic in closed-source shops.

How can anyone expect high quality review in this setting is beyond me.
 
Old 03-27-2012, 12:08 PM   #44
TobiSGD
Moderator
 
Registered: Dec 2009
Location: Hanover, Germany
Distribution: Main: Gentoo Others: What fits the task
Posts: 15,576
Blog Entries: 2

Rep: Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036Reputation: 4036
Quote:
Originally Posted by qweasd View Post
My guess would be (and I could easily be wrong) that nvidia code has been "hijacked" by a few core developers who get to see the complete source. They are the indispensable ones who understand how it works, and they resist as much as possible either opening the code or training the replacement. Aside from the docs NVIDIA squeezed out of them at gunpoint, they probably have their own secret and much more complete docs which they won't ever share with anyone. This problem is endemic in closed-source shops.

How can anyone expect high quality review in this setting is beyond me.
That can't be how it works. The driver developers are not designing the chips, the chip-designers are the ones in charge, they have to have the complete documentation.
 
Old 03-27-2012, 12:20 PM   #45
qweasd
Member
 
Registered: May 2010
Posts: 437

Original Poster
Rep: Reputation: Disabled
Quote:
Originally Posted by TobiSGD View Post
That can't be how it works. The driver developers are not designing the chips, the chip-designers are the ones in charge, they have to have the complete documentation.
One does not exclude the other. The hardware specs, like you say, are most likely in a well-understood form, but it does nothing to prevent the driver code from being hijacked. We are talking the same people who used to release an obfuscated "open source" driver. The x64 binary is circa 50 MiB now, and if you want to make a change, you either appease the gatekeepers, or you rewrite the whole thing. Again, pure speculation on my part, but also something that would not surprise me.
 
  


Reply

Tags
nouveau, nvidia


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
nvidia cards rehan999 Linux - Software 16 12-09-2009 02:51 AM
Sticky Request: How to get the nvidia driver for old RIVA TNT2 and other nvida cards darkhatter Suse/Novell 2 03-14-2006 04:05 PM
Those with NVidia Cards might want to.... Black_iCE_mdk Mandriva 1 06-09-2004 04:56 PM
nvidia fx5600 loads vesa driver instead of nvidia driver saburo62 Linux - Hardware 3 05-04-2004 03:59 PM
Nvidia Video Cards liguorir Linux - Hardware 2 08-30-2003 05:48 AM


All times are GMT -5. The time now is 09:08 AM.

Main Menu
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
identi.ca: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration