How to install nVIDIA Graphics Driver in Linux?

I may have to look into it at some point. I think I am about to have the GeFarts driver working. I found how to install it with Yast and it is in the process. Will find out after I restart X.

 

OK, I got it figured out. I will have to play with it more next week when I return from my business trip. Time to hit the sack. Have to got to the airport in about 4 hours.

Not sure why there is not taskbar tray icon for the wired network adapter.







@Ultra Male

 

So the first thing to do is to clock back down to stock. It's possible that your system isn't fully stable at what you're running at and that something or other is causing that instability to manifest itself more severely under Linux. My current laptop at first showed data corruption errors under Linux, but Windows appeared to be fine. It appears that the problem was a hardware issue triggering off the network adapter, but it was clearly a hardware problem. I had a lot of difficulty reproducing it under Linux, but that may have been due to the fact I couldn't achieve the same level of network throughput under Windows. In the event, I did reproduce it under Windows, my motherboard was replaced, and no problem since.

As for your wired network connection, I'm very surprised. Just about every major distribution uses NetworkManager, which does all of this automatically and doesn't require configuration.

 

There might be a single icon for all networking (that's what KDE does, at any rate).

 

@Mr. Fox, that is brilliant! Having professional ties to Miguel de Icaza, I've used OpenSUSE in the years past, having only switched to Mint on my main machine in the past year. I'm still running Open SUSE Leap on an old file server and media server at home.

Having more exposure now, I like the 'apt-get' model a bit better, but YaST is still good too. If it works for you, then that is the first step. Best of luck to you and the x299 Dark in this experiment, test, or full blown conversion, whatever the case may be.

 

I don't have any icon. My desktop does not have WiFi, only LAN and there is no icon. Weirdly enough, when I was running the Live distro from USB prior to installing it had one for LAN. And, it worked automatically. I had to manually configure the network adapter on the installed version. It did not work at all after installation.

Thanks, bro. Mint has always been my go-to Linux OS. I love Mint, but I could not get it to run on this motherboard. It was locking up constantly and having weird kernal panic problems. About half the time it was freezing during ACPI detection, long before it even tried to load x-org. I did make it to the Mint desktop a few times and it locked up there as well.

I don't use the latest BIOS because it has way too much security filth for Intel ME and CPU micro-code and I care more about performance than security. But, since I have 3 BIOSes, I will flash the latest on one of the chips and see if Mint will work with the newest firmware. Mint does seem to be a lot easier to use than OpenSUSE. But, I do like this OS. I don't my rolling up my sleeves and using a terminal. I kind of like that. But, I don't like not being able to find the information I need to make something work and I only got this far through about 8 hours of trial and error, LOL.

I am quad-booting now. Windows 7, Windows 10 Enterprise 1809 with the full load of cancer (Store and all the standard payload of UWP feces), Windows 10 Enterprise LTSC 2019 (with none of the worthless UWP filth, Store or Cortana the digi-slut) and OpenSUSE. So, I have a working digital Swiss Army Knife PC where I can now use the best OS for benching whatever benchmark returns the best scores.

All of this will have to wait until next week, as I am on a business trip. I might install Mint on my ZBook 15 laptop while I am traveling this week. Of course, that will be contingent upon the hotel WiFi not being so miserably slow that I cannot download it.

 

Mint being based on Ubuntu LTS usually means the Kernels will be a generation behind the current leading stable Kernels. It wouldn't surprise me if this has been fixed in a later kernel, but since you can't boot it you can't really install a newer revision.

What performance issues are there? If you're referring to the Spectre mitigations, you're copping those anyway since they're patched in the Kernel, not in BIOS. Or now un-patched, since the v2 mitigations were reverted a little while back. Most newer kernels will note this fact into your syslog.

 

Done!

 

Try this acpi-osi gimmick to make Linux look like Windows 10 just like GTA's saloon. I just got 30 mins to 60 mins of extra battery life doing nothing. You're better off using Ubuntu especially Kubuntu. I use Xubuntu for minimal experience, its better to stick with GNOME 3.3 or Plasma for best experience. For Linux its better to revert BIOS to defaults and quickly install it and once its done and updated with all the stuff you need then tweak it, otherwise Linux will never work since it adds quirks or workarounds with some BIOS revisions. Its better to update MEI FW using win-raid for best performance and compatibility.

 

But, I also run Windows and I don't want any Meltdown or Spectre mitigations in my firmware or my OS. And, that is one reason why I do not allow Windows Updates as well. If it is baked into Linux kernel it won't really matter on the laptop since it is not a high performance product and I am not really doing any overclocking or benching with it. I am much pickier about that on my desktop because overclocked benching is the primary reason that it exists.

 

If you don't want those patches, you may be able to install older Intel microcode by hand for Spectre and compile your own kernel with PAGE_TABLE_ISOLATION=N for Meltdown.

I think the microcode versions from 20171117 backwards are free of the Spectre mitigation. See: https://metadata.ftp-master.debian....ocode/intel-microcode_3.20180807a.2_changelog

Edit: apparently it's even easier than compiling yourself. There are a couple of boot parameters:
https://kernelnewbies.org/Linux_4.15#Meltdown.2FSpectre

Edit2: there are now even more including nospec_store_bypass_disable:

Code:

        nospectre_v1    [PPC] Disable mitigations for Spectre Variant 1 (bounds
                        check bypass). With this option data leaks are possible
                        in the system.

        nospectre_v2    [X86] Disable all mitigations for the Spectre variant 2
                        (indirect branch prediction) vulnerability. System may
                        allow data leaks with this option, which is equivalent
                        to spectre_v2=off.

        nospec_store_bypass_disable
                        [HW] Disable all mitigations for the Speculative Store Bypass vulnerability

Spoiler: the rest of the options with more granular control

Code:

        spectre_v2=     [X86] Control mitigation of Spectre variant 2
                        (indirect branch speculation) vulnerability.

                        on   - unconditionally enable
                        off  - unconditionally disable
                        auto - kernel detects whether your CPU model is
                               vulnerable

                        Selecting 'on' will, and 'auto' may, choose a
                        mitigation method at run time according to the
                        CPU, the available microcode, the setting of the
                        CONFIG_RETPOLINE configuration option, and the
                        compiler with which the kernel was built.

                        Specific mitigations can also be selected manually:

                        retpoline         - replace indirect branches
                        retpoline,generic - google's original retpoline
                        retpoline,amd     - AMD-specific minimal thunk

                        Not specifying this option is equivalent to
                        spectre_v2=auto.

        spec_store_bypass_disable=
                        [HW] Control Speculative Store Bypass (SSB) Disable mitigation
                        (Speculative Store Bypass vulnerability)

                        Certain CPUs are vulnerable to an exploit against a
                        a common industry wide performance optimization known
                        as "Speculative Store Bypass" in which recent stores
                        to the same memory location may not be observed by
                        later loads during speculative execution. The idea
                        is that such stores are unlikely and that they can
                        be detected prior to instruction retirement at the
                        end of a particular speculation execution window.

                        In vulnerable processors, the speculatively forwarded
                        store can be used in a cache side channel attack, for
                        example to read memory to which the attacker does not
                        directly have access (e.g. inside sandboxed code).

                        This parameter controls whether the Speculative Store
                        Bypass optimization is used.

                        On x86 the options are:

                        on      - Unconditionally disable Speculative Store Bypass
                        off     - Unconditionally enable Speculative Store Bypass
                        auto    - Kernel detects whether the CPU model contains an
                                  implementation of Speculative Store Bypass and
                                  picks the most appropriate mitigation. If the
                                  CPU is not vulnerable, "off" is selected. If the
                                  CPU is vulnerable the default mitigation is
                                  architecture and Kconfig dependent. See below.
                        prctl   - Control Speculative Store Bypass per thread
                                  via prctl. Speculative Store Bypass is enabled
                                  for a process by default. The state of the control
                                  is inherited on fork.
                        seccomp - Same as "prctl" above, but all seccomp threads
                                  will disable SSB unless they explicitly opt out.

                        Default mitigations:
                        X86:    If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"

                        On powerpc the options are:

                        on,auto - On Power8 and Power9 insert a store-forwarding
                                  barrier on kernel entry and exit. On Power7
                                  perform a software flush on kernel entry and
                                  exit.
                        off     - No action.

                        Not specifying this option is equivalent to
                        spec_store_bypass_disable=auto

and for page table isolation:

Code:

        pti=            [X86_64] Control Page Table Isolation of user and
                        kernel address spaces.  Disabling this feature
                        removes hardening, but improves performance of
                        system calls and interrupts.

                        on   - unconditionally enable
                        off  - unconditionally disable
                        auto - kernel detects whether your CPU model is
                               vulnerable to issues that PTI mitigates

                        Not specifying this option is equivalent to pti=auto.

        nopti           [X86_64]
                        Equivalent to pti=off

Source: https://www.kernel.org/doc/html/v4.19/admin-guide/kernel-parameters.html

 

For anybody interested and stumbling onto this thread, "G-Sync Compatible" (ie FreeSync) mode works with the 418+ driver series if you're feeling brave.
https://www.phoronix.com/scan.php?page=article&item=gsync-compatible-linux&num=1

It should hopefully appear in the Graphics Drivers PPA (for *buntu users) in the near future.

Furthermore if you want to test G-Sync/FreeSync there's a simple variable framerate demo app available. It's basically the same as the 2D line test in the official test app.
https://github.com/dahenry/gl-gsync-demo

 

@Mr. Fox 19H1 has Retpoline that minimises Perf. impact for general usage w/ Spectre patch enabled on Win 10.

 

I got it figured out. Changed my BIOS over to Legacy and now running Windows 7, 10 and Linux in Legacy mode and no more issues. Linux Mint is now installed on my desktop. Now I just need to install Steam and play some stuff when time permits. CPU is running 5.0GHz and ram at 4000 CL16 and no issues. (I cannot see either of those clock speeds in Linux, but there's probably a tool I haven't found yet to show them.) I also found 'GreenWithEnvy' app for overclocking the GPU in Linux.

 

Kind of figuring it out...

 

At a pinch you can just use this command:
cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_cur_freq

There's also a tray app called "indicator-cpufreq" which should work on any GTK3 DWM (Mint and Cinnamon). Other DWMs should also have a matching applet (e.g. xfce4-cpufreq-plugin).
It tells you all the individual CPU clocks plus allows you to set the CPU governor (useful for benching).
The only useful governors are:
-performance = sets max clockspeed at all times. Don't leave it on this for general desktop use, but it can be useful for benching and gaming.
-ondemand = normal dynamic scaling
-powersave = basically the opposite of performance. Locks to base clock.

 

I stopped using xfce4-plugin in favor of tuned-adm to set between balanced,max performance and max battery mode.

Run Cb15 stock/extreme in Linux using Q4Wine or WINE with Win 8/10 compatibility you'll be amazed how much fast it renders.

 

Try searching for "lshw" and memory. There's an argument for the list hardware command that should give you what you want.

In case you want to mess with any undervolting, there might be a UI tool here worth looking at - http://forum.notebookreview.com/threads/linux-undervolting-gui.827363/

 

Thanks, will check that out.

I haven't figured out how to use WINE yet. I'm missing something.

Thanks. Appreciate the info. I might use that on my laptop (also running Mint) but I need to do the exact opposite on the desktop. But, I can do that in the BIOS. The BIOS on my laptop doesn't have anything at all for this kind of thing.

 

Install Q4Wine and right click exe and set Q4Wine to open EXE file and you need to wait for 30 sec to get it loading and you get a small window and press OK and Hit Run on Cinebench R11/15. R20 is Strictly Windows 10 only so I haven't been able to run it. You can tweak Wincfg,regedit, open explorer etc.. in nice easy to use GUI from Q4wine.

 

Not too bad... 50x32... also got that tray utility @Stooj mentioned. Now I need to see how it works at 52x32.

 

Within 2 years, I'm gonna freeze whichever version of W10 I'm using or else grab some LTSB keys from ebay to upgrade all my PCs and switch to Linux permanently.

 

I may stop my benching and switch to Linux permanently as well. There is getting to be too much stupid nonsense to deal with and I am getting fed up with the nonstop flow of crap we are seeing from Micro$lop, NVIDIA, Intel and AMD. Everything made by this insane clown posse of tech-tarded morons is either lackluster, broken, artificially crippled/capped by cancer firmware so they can meter performance (e.g. GeFarts vBIOS) or driven by cheap gimmicks... or a unwholesome combination of all of the above. I'm getting tired of giving my hard-earned money to these stupid garbage kings.

 

Amen to that. As a daily Linux user as my main computing rig, Window'$ has only been for games and benching. That's absolutely it. I don't even log into my email with M$ rubbish.

What's sad is that they have such a monopoly, providing users with the convenience factor that would take a major brain cleansing to shy away from...

I strongly encourage people to check out Linux Mint w/ Cinnamon. You can make it what you want and it is awesome. Heck, if you want to make it look like window'$ or O$X you can do that too..

There is Freedom in the computing world with Linux ...or remain prisoners to the goons...

It's great to see more focus on Linux @Mr. Fox ... I'll join you in this quest for sure ...

 

This isn't related to installing GeForce drivers in Linux, but it's still pretty cool because it is just one more example that shows Windows 10 performance sucks compared to Windows 7 and Linux. Somewhat off topic, but a good incentive to figure out how to run Linux and install GeForce drivers if you are fed up with the Windows 10 bloatware filth-o-rama.

Here is a comparison of Cinebench runs with Linux versus Windows 7 and Windows 10 with the same BIOS settings. As suspected, Windows 10 is the loser (status quo). Linux beats Windows 10 and comes very close to matching Windows 7. These are the best of 3 back-to-back runs (using Cinebench "Keep best score" feature).

 

So, now it's quad-boot, LOL.

Spoiler: Quadzilla

@Ultra Male

 

I just ran CPU-z from C drive and somewhat it runs/works.

 

Consider ditching Windows Boot Manager, and using systemd.boot instead.

 

I will have to Google that and find out what you are referring to and how to use it. I am having to figure stuff out as I go and information isn't robust like Windows. Does systemd.boot work with legacy BIOS, or UEFI? I asked because I switched everything over (reinstalled Windows as well) to legacy because I could not get Linux Mint to run on my X299 Dark motherboard with UEFI mode. It installed painlessly without any glitches at all in legacy BIOS mode.

I am still trying to figure out how to implement the boot parameters to disable Spectre and Meltdown that @ALLurGroceries mentioned in post #61, LOL.

 

Oh; it's amongst a choice of bootloaders that you can use to boot Linux, or any other OS.

About half a decade ago it used to be LILO; a lot of people use GRUB/GRUB2 now, but setup for that is insanely complicated. systemd.boot is a very simple, lightweight UEFI bootloader.

 

You replied as I was editing my post.

Well, I guess that isn't going to work out since it is EFI instead of legacy BIOS. openSUSE installed effortlessly in UEFI mode, but Linux Mint gave me nothing but fits until I changed the BIOS to legacy mode.

I can also just press F7 during POST and select the Linux SSD to boot with Grub2 and not even use Windows Boot Manager. When I reinstalled everything, I installed Linux on one NVMe SSD with all of the other drives disconnected to keep the two environments agnostic.

 

That might've been you probably had Secure Boot switched on. There's an option in the UEFI firmware to switch it off; have you considered that?

I personally leave Secure Boot switched on, as I don't want any funny business where I leave my laptop unattended (happens a lot in a university).

You can sign the kernel yourself using a key following instructions here:

https://blog.hansenpartnership.com/linux-foundation-secure-boot-system-released/

I had similar issues with Arch Linux, but the advantages of UEFI greatly outweigh those of legacy boot, so I spent the extra effort to make it work.

 

I never enable Secure Boot (ever, on any system, because I abhor it,) and I manually delete the keys in addition to disabling Secure Boot. I haven't ever identified anything beneficial about using UEFI other than having a higher display resolution while working within the firmware (BIOS) environment. Security isn't a big deal to me, especially when it reduces my benchmark scores or interferes with me using modded drivers or firmware. My Windows systems have driver signing permanently disabled in the BCD and UAC disabled in Windows. I view all of those things as inconvenient impediments. I also do no use antivirus software or Windows Updates because they lower my benchmark scores. I also tend to avoid BIOS updates, Intel Micro-code and Intel ME firmware updates for those same reasons. Unless the firmware does something specific to enhance overclocking or increase performance, I don't really want it.

Part of my contempt for this "security" filth is the fact that it has become a massive impediment to firmware modding. I will never forgive the Digi-Nazis for the draconian firmware signing bull crap that has screwed up pretty much everything in PC tech. End users should have the option of exercising absolute god-like control over every aspect of their PC if they want it, but we do not because of this filth. Understanding that security has a place for those that want/need it, I'd probably be more tolerant of this crap if all computers included an option of manually disabling firmware signing and flash protection.

 

Its better to use Secure boot in UEFI, its simply a gimmick and installing OS is really slick when BIOS actually plays nice with OS. Legacy Boot mode simply increases boot time. I am using Secure boot in BIOS but in Linux I disabled it using mokutils for proper working of nvidia driver, so BIOS and OS are happy seeing UEFI mode.
Firmware security bypass tricks are almost NIL these days on newer boards which is very sad.

 

I found a simple to use program called Grub Customizer and that allowed me to add Windows Boot Manager to the Grub2 menu and boot from the NVMe drive where Mint is installed rather than booting from the Windows drive. (Both are bootable independently from the other.) It also allowed me to set a Grub wallpaper and up the resolution to something nicer than the standard Grub VGA resolution.

 

Grub2 is very good. Once you've installed everything, go to terminal in Mint and type sudo update-grub so that all Win 10 installs are detected and entries are added.

 

I have all 3 Windows installations in the Boot Manager that Grub loads, so I do not need to have them listed in Grub. When Grub loads the Windows Boot Manager the boot screen lists all of them.

 

How do you disable driver signing permanently? The only way I know how is to reboot with advanced options and to select the "Disable Driver Signing" option but that only works on that particular boot. As for Intel Micro-Code updates - Do you know how to revert these? I downloaded InSpectre (a windows program which lets you see your OS/hardware protections against the Spectre and Meltdown vulnerabilities, and lets you disable the OS protections) - a few months ago it informed me that I had protections through Windows updates but not microcode, so when I disabled the protections I got about a 5% boost on CPU benchmarks on my Haswell machine (newer architectures are less affected). However at some point since then (I've since done fresh Windows installs and upgraded to 1.03.04 v2 Prema Bios on my Clevo P157SM), I appear to have received a microcode update. Now my CPU benchmark scores don't change significantly after disabling the Windows Spectre/Meltdown protections. If I could somehow delete the microcode patches I could get the 5% back.

 

Perhaps somewhat of a niche situation, but if you have an NVIDIA Optimus laptop and plan to use Linux with NVIDIA only for computing (not display purposes), then you can follow this guide step by step here: https://gist.github.com/hemenkapadia/9a36d4310b2bf6a945636f05d4a046c7

Note the suggestions I made in a comment at the bottom of that. If you can be bothered understanding exactly what all those steps are doing, you can adjust it to also use it for display purposes if you wish (but I personally dual boot and do my gaming in Windows).

Following that guide was the only way I managed to get everything I wanted working flawlessly on my Clevo P157SM. I tried other approaches but always ran into errors (some guides I followed were not specific to Optimus machines, sometimes I think I may have faced issues with making my NVIDIA card render the GUI desktop... honestly I don't know, I'm a linux novice). With this guide, I can do all my CUDA computing without hiccups and things like adjusting screen brightness with keyboard hotkeys works normally as well.