This is the journey of getting head tracking for X4 Foundations on Linux. Browsing the in-game options I found key bindings to enable and reset a head tracker so I got my hopes high for head tracking for X4 on Linux and started reading.

tl;dr: The native build of X4 does not support this at all. It even lacks the symbols in the executable. There is hope with Wine though.

While seeking through the patch notes two things catched my eye (harhar). The notes suggest support for Tobi Eyetracker and somewhat earlier TrackIR. The Tobi one was marked as “Windows only” so I hoped for TrackIR and tried to understand what is involved into getting this particular head tracking to work.

Please do keep in mind that I never touched any head tracking before so I had to grasp the theory behind all this first.

So apparently there are not many possibilities to configure anything in-game. There is the additional start parameter -forcehmd that may have some effect here but that’s about it. So everything related must come from the outside, right?

Funny enough a Kerbal Space Program extension on GitHub provided me with the idea how TrackIR is supposed to function: so let’s clear that one up: The code looks for the registry entry “Software\\NaturalPoint\\NATURALPOINT\\NPClient Location” that points to the NPClient.dll (or NPClient64.dll) which in turn in loaded by the executable if found and accessed. And sure enough I found the same pattern by using the good old hexeditor on X4.exe as well:

You may suspect this already but none of that can be found in the X4 binary for Linux and this is the end of the story.

Appendix: It might be possible that a joystick look-around thingy may be abused for this but since I do not own such a joystick (yet) I wouldn’t know how to fake such signals.

A new hope

During my research on head tracking on Linux I stumbled over the website that mentioned the OpenTrack project to play Windows games with Wine on Linux. How cool is that? A quick check suggests that OpenTrack features (also) the output protocol “freetrack” for Windows and a “Wine freetrack glue protocol” for Linux and OSX.

Freetrack seems to be a implementation of what TrackIR does up to version 4. Looks like the company didn’t like that or was afraid that some tinkerers use self built head tracking devices and not buy their ~200$ hardware any more so they added encryption to their devices with version 5 leaving freetrack dead in the water. Or whatever. So if a game would only support protocol 5… it’ll probably no longer work. Anyway, let’s jump into OpenTrack and give it a try. Once all required developer packages are installed, that includes the wine-devel packages for i686 (and libevdev if you want virtual joystick support), this can be compiled:

git clone
cd opentrack
mkdir build
cd build
ccmake ../
make -j4
make install

The important part is to tick the SDK_WINE option or we don’t get the precious Wine Glue. Here, have a video:

And sure enough we find the following already familiar files in install/libexec/opentrack/

  • NPClient.dll
  • NPClient64.dll
  • TrackIR.exe

So what OpenTrack does is once it’s started and configured to use “Wine — Windows layer for Unix” as Output: It “injects” the key [Software\NaturalPoint\NATURALPOINT\NPClient Location] to the user.reg of the configured Wine/Proton prefix on start while the also started is used for the shared memory mapping – means from my understanding from the OpenTrack binary to some Wine process. TrackIR.exe is just a dummy that may also be run with Wine but does nothing. It’s apparently for games that check for a process with that name before they initialize head tracking features. Neat, huh?

For this to work you have to click on the little hammer symbol next to Output so make sure that your Wine Prefix is properly configured or OpenTrack may insert the registry key to the wrong Wine prefix. Sure enough it seems to come Steam and Proton ready as well but I did not try this because I made the mistake of buying X4 on GOG (I know this is getting old).

So what’s next? Oh yes, a Head Tracking device. Well bite me, I don’t even have any. Why did I go through all this trouble at all?

The head tracking device

Well why should I purchase such a device without knowing if I can even use it? Do you know what kind of device has a gyroscope built in however? Yes, every average smart phone has one and some are even really good (I hear.) So here comes the fun part: I strapped my mobile phone to my headphones while an app sends the tracking information via wifi and the UDP protocol to OpenTrack as input source.

So I will not bore you with the details how I configured and mapped this in Opentrack. It will be of no use for you anyway since this depends totally on your device and system but I can tell you that this is a lot of fiddling with the settings. Also know that a 3-point device can indeed be built for ~10$ with some LED and a webcam – there are various people demonstrating this on the net and you will find plenty on the usual video platforms.


I don’t like to have to install the game twice now, once native and once with Wine, but… it’s for science! Or so.

Yes, it’s a little clunky. This is probably because:

  • my phone sucks
  • my phone really sucks (seriously)
  • it’s UDP – order of packages is not guaranteed
  • it’s wireless and the access point is two walls away
  • …and I probably also messed up with the mapping and filter options 🙂
  • I even read that it’s not any better on Windows xD

So that’s it. Head tracking on Linux for X4 Foundations. Jumped the hoops 🙂

Getting into modding is a painful experience for a newcomer especially when on a clock (as in: Ain’t no time for this) so here is what I understood so far.

Finding resources

EgoSoft is one of the few companies that still hold on to a forum. They may be on social media but that’s just as news outlet. This is somehow also where the modding community resides, when it usually would find a place on it’s own “elsewhere”. Clever move and I applaud – at least I don’t have to join some private group on Facebook.

Sadly the forum has no search function to speak of especially not for a technical search or, gasp, a code search. So most mod developers usually host their stuff elsewhere – like scattered on GitHub, and only talk about their mods here and there. Connecting the dots (and users) is up to the initiate. In fact it helps slightly to utilize an external search engine that does a better job compared to the forum search itself. Go figure.

Eventually I stumbled by accident or luck over a Confluence installation that seems to act as some sort of Wiki and has indeed some pointers on modding for various X titles at – and I am still uncertain if this is intended to be public or not. Sadly the X4 modding articles in there are short and of course available in English only. That wouldn’t be an issue if Confluence wouldn’t stubbornly insist in trying to render a page in my native language first, informing me helpful every time that no such entry exists and makes me switch the language every single time back to English. Well, I guess it is a community driven documentation system so I could scratch my own itch and translate stuff. Thing is I should understand what I’d be talking about in the first place, no?

Turns out that most know-how for X4 modding can in fact be learnt from X3 and X Rebirth in particular. Both are precursors and partially sandboxes for X4, from my understanding, so a lot of the information does also apply to X4. Sometimes with slight differences.

The most helpful place however is an unofficial Discord channel where many of the mod developers hang out and they do seem to be really friendly to newcomers asking the same old questions over and over again. I won’t bother with the invite link as this one is subject to change all the time but it can indeed be found with the dreaded forum search.

The extension file formats (and formatting)

Anyway, let’s dive into some details. Mods, or as it should be called nowadays “extensions”, consist usually of so called MD Scripts, that’s short for Mission Director (and not Markdown) written in well formed XML Syntax, and LUA scripts (that seems to be the gaming industry standard – at least I keep hearing that). LUA itself is explained in great detail in the online documentation of LUA while game specifics are listed in varying detail on the Confluence mentioned above but that’s subject for another article. Also various assets may be floating around in the extension folders.

Now in theory XML and LUA are completely system independent so modding with different systems in mind should be a no brainer, right? Right?

Well, of course nothing is ever that easy. Thankfully the user CulunTse took the burden and wrote an article on all the caveats encountered so far: (Steps to make your mod work on Linux+Mac) – it’s for X-Rebirth but the gist applies for X4 as well. So when writing mods make sure to use lower case only and don’t use special characters at all. Best not even use a white-space just to be sure.

<rant>Again Windows f****s us all over being the only system that is fine with "A==a" being true. Not even JavaScript manages this. And since X4 is a game and the majority of gamers (and mod devs) play on Windows we have to suffer from this yet again. The average mod is simply not compatible and broken for e.g. Linux users. Guess who gets the blame. The worst part is that another generation of developers will not see any problem with this behaviour.</rant>

So now we learnt that an extension consists mostly of XML and LUA files, so how comes that mods downloaded from Steam, e.g. to learn from, are riddled with various TXT files instead rendering your linter of choice useless because it won’t automatically parse a text file? Well that seems to be a Steam Workshop limitation not allowing certain file… extensions. So developers started to rename their files when uploaded to Steam. Yay, more confusion for man and machine (as in mankind – c’mon, it’s a lame allusion!).

Why I mentioned Steam now? Well, learning from existing extensions is the way to go. Also since some simply don’t work for Not-Windows users it’s up to myself to debug. The places to get extensions is usually from NexusMods (no thanks, still angry that they lost my user data years ago – still used by scammers) and Steam. Avid readers of my ramblings may know my especially sour spot of having purchased X4 on GOG ( so downloading from the Steam Workshop is not as straight forward as it is for others. There are various so called “steam workshop downloader” that easy the pain somewhat.

But wait. The Steam Workshop file comes as DAT file. What is that again when I just talked about TXT files? Well, this is from my understanding a format by EgoSoft designed for Steam Workshop files for X Rebirth – also called a XRWS file. At least from – a tiny little project that helps unpacking the DAT files but has to be compiled before use. That’s usually a matter of issuing the make command after checking out the repository (or download the release file if you’re feeling lucky). Fair warning: It is somewhat rigid in the way the DAT file has to be named so you may have to adjust that by renaming. Ymmv.

Armed with that knowledge I was able to download and extract the awesome extensions_fireandsmoke_v107.dat extension that really spices up space fights with the effects we know and love from X Rebirth (or not). This one I could also drop in my user space folder under ~/.config/EgoSoft/X4/extensions/x4_fireandsmoke/ (again: case sensitive, important). The folder name matters because it is also hardcoded into the extension files itself and will fail to load various assets if changed. Interesting design choice.

Why this is noteworthy? Well apparently this does not work for all kinds of extensions. Some seem to work only when put in the game path /path/to/X4_Foundations/game/extensions/ where the game also stores official DLCs. Better keep that in mind. The gist seems to be that user space extensions are limited in functionality to prevent nefarious mods. Or so I hear. Maybe EgoSoft simply never got it working properly. There is more (conflicting) info hidden deep in that Confluence mentioned above.

One extension to rule them all (and a pipe)

So why go through all that trouble when an extension can simply be downloaded from e.g. GitHub? Well, I learnt the hard way: Also not as simple. For example many mods rely on one very important extension that can be found in the repository – a wild mix of various extensions and even a Pipe Server (more on that later). It’s the indeed impressive extension sn_mod_support_apis featuring a clever way to work around some UI modding limitations in X4, allowing lazy loading of further LUA scripts and even introduces a Pipe Server to interact with the game from the outer world – mostly used for more complex hot keys. A dream coming true and used by many other extensions as well.

Sadly it didn’t really work out of the box when checked out from GitHub and put in place at /path/to/X4_Foundations/game/extensions/sn_mod_support_apis/. There were various reasons for this. First of all: Case sensitive again. The XML files in the md/ folder must be lower case or X4 will simply ignore the files. Easily fixed though.

The next problem wasn’t that easy to identify and the reason for this is hidden in plain sight in this titbit of information from the synopsis:

A workaround is to load in custom lua files alongside the egosoft lua. This is done by editing one of a handful of ui.xml files in the ui/addons folders, adding the path to the custom lua file. These ui.xml files cannot be diff patched. The lua file must be given an xpl extension, and this xpl and the ui.xml must be packed in a “subst” cat/dat.


Where this arcane know-how was acquired from in the first place I do not know. The gist is that some XML files are happily read by X4 (and can even be hot reloaded) while some can not. The ui.xml falls into the not so much category and since I have no idea how to create a cat/dat file (yet) I had to scrape the “subst” files from a release (Steam Workshop, NexusMod, GitHub release, you name it). Without it’s simply not read and ignored and this is also why no single debug line will ever be logged to give the (weary) initiate a hint what may be wrong.

After that mods relying on sn_mod_support_apis started working (or throwing traces at least). Awesome! Onwards to Pipes! Or Not! Because this part is Windows only. Why? Well, the Pipe Server uses a LUA feature to load a library from disk providing that pipe feature. And that project is written in C, compiled as a separate DLL and relying on Windows, of course. That makes even sense and I really can not blame the author for scratching the own itch only here. See for details and to be fair the extension is written in a way that other features do still work so it’s not a total roadblock and in theory I can go back to be a happy gamer at this point.

Alas I want that Pipe feature, of course, so I have to come up with my own library at this point. It’s not a complex file but my C days are long past. So to spice this up I needed a crash course on how LUA is supposed to work. Script wise and all and I don’t think I was prepared for all this.

Diving into the unknown (What is LUA/JIT anyway?)

LUA is basically… ah frell, go and look it up yourself. In the end LUA scripts are interpreted by a VM. That would be LuaJIT (JIT – Just In Time) on Linux and this is why X4 is shipped with a file named Sadly this is not the particular version of LuaJIT. It stands for an ABI compatibility version. In theory at least. My first goal to get the idea was to grab the source and compile my own libluajit. Should be a no brainer, right? Little did I know when I checked out the project from

I run a hexedit on the distributed library to get an idea what version is used and came up with 2_1_0_beta3. The commit 8271c643c21d1b2f344e339f559f2de6f3663191 of the LuaJIT project is tagged with that version so I went with that first.

Compile went smooth, beside some warnings, but X4 would stop dead throwing a Fatal Error at me that I’ve never seen before. Well, let’s fast forward to HEAD and try again and this time the game started but became stuck in main menu with unresponsive entries and missing labels (some said “Processing…”). So obviously X4 is not running vanilla LuaJIT and since this is under MIT licence I don’t think they even have to provide modifications. What now?

Thankfully some fellow gamer on Telegram, who doesn’t want to be named, pointed me in the right directions due to experience with LUA. An article over at explains some major caveats with this and sure enough once I knew what I was looking for I found evidence here and here and here. (Yes, they have Jira as well – who would have guessed).

Lua engine was upgraded to LuaJIT 2.1 which comes with performance improvements as well as new language features (incl. some added Lua 5.2-specific features as well as some Lua 5.3 ones.

Just looking at the number of forks of LuaJIT on GitHub makes me dizzy so I went with the first recommendation that also addresses the memory issue and also edited the file src/Makefile enabling some LUA 5.2 features by commenting in the line: XCFLAGS+= -DLUAJIT_ENABLE_LUA52COMPAT

And guess what, X4 launched with this and also started an older save game of mine just fine. I guess this works so I’ll keep that in mind in case I need some monkey patching to try stuff.

Next on that list? Find out how to write a loadable C library for LUA and adapt that Pipe Server.

At least gaming itself is easy as pie on Linux in 2021. Modding? Now so much.

I was eager for the second expansion set of “Cradle Of Humanity” since I enjoyed the Split Vendetta expansion a lot. Sunk many nights into this sandbox so I pre-ordered the DLC on GOG (Good Old Games) and watched that countdown to release ticking down. Oh boy, was I disappointed when the timer reached zero and I got nothing while people who bought on Steam already enjoyed the DLC. As usual GOG leaves Linux players standing in the rain so I called it a day and checked again on the next day when I was presented with this:

X4: Foundations patched, Split Vendetta was not, Cradle Of Humanity nowhere to be seen.

So apparently they finally managed to roll out version 4.0 of the main game but missed the first DLC Split Vendetta. What could possibly go wrong. Cradle Of Humanity is still nowhere to be seen. This didn’t change until now, one day after release and the time of writing of this article. Oh GOG, I am so done with this. And I even expected this, joking around weeks before the release date that this will be two weeks later for GOG users. Again.

Curiously I checked the Downloads for Windows next and guess what: The DLCs were all there and also on version 4.0!

So I did what every Linux tinkerer would do. I checked out what is really in the DLC files by extracting the contents using innoextract. To my delight I could not find anything operating system related in there so I threw all the Windows DLCs into my version 4.0 base game folder and extracted both DLCs.

The required files are:

  • setup_x4_cradle_of_humanity_4.00_(64bit)(45636)-1.bin
  • setup_x4_cradle_of_humanity_4.00(64bit)(45636).exe
  • setup_x4_split_vendetta_4.00(64bit)(45636)-1.bin
  • setup_x4_split_vendetta_4.00(64bit)_(45636).exe
cd /games/linux/X4_Foundations/game
innoextract --exclude-temp --extract setup_x4_split_vendetta_4.00_\(64bit\)_\(45636\).exe 
innoextract --exclude-temp --extract setup_x4_cradle_of_humanity_4.00_\(64bit\)_\(45636\).exe

When I run the game now I found both DLCs registered in the Extensions menu and could start a new game as Terrain fraction. Whop whop, here we go.

No, I will never buy a recent game on GOG again. Especially not as pre-order. I wish I could migrate this to my Steam account. This is the DRM free revolution. As usual, the joke is on the paying customer.

tl;dr: Add PATH="${PATH}:/bin:/usr/bin:/sbin:/usr/sbin" to /etc/default/firehol when using backported version 3 of firehol on Ubuntu.

firehol – an easy to use but powerful iptables stateful firewall

man firehol

With this out of the way: When installing firehol on aging Xenial (Ubuntu 16.04) you want the backported packages by Andrey Galkin to get version 3 of firehol over version 2 in universe – especially when working with IPv6:

When done setting up your rules you may find out after a reboot that the systemd job involved will claim to have started firehol but eventually discover that your iptables are empty despite systemd claiming otherwhise and having set START_FIREHOL=YES in /etc/default/firehol:

● firehol.service - LSB: firehol firewall configuration
   Loaded: loaded (/etc/init.d/firehol; bad; vendor preset: enabled)
   Active: active (exited) since Fr 2020-11-27 15:43:51 CET; 2h 8min ago
     Docs: man:systemd-sysv-generator(8)
  Process: 31555 ExecStop=/etc/init.d/firehol stop (code=exited, status=0/SUCCESS)
  Process: 31574 ExecStart=/etc/init.d/firehol start (code=exited, status=0/SUCCESS)

This is especially weird when you run the startup /sbin/firehol start command manually and it succeeds just fine.

I had to dig deep to find out where the script is in fact falling flat. This was mostly because of old init script /etc/init.d/firehol redirecting the output of the starting process to /dev/null not showing the errors at all:

do_start () {
        # return
        #  0 000 if firewall has been handled
        #  1 001 if firewall could not be activated
        #  4 100 if FireHOL is disabled via /etc/default/firehol
        [ "$START_FIREHOL" = "NO"  ] && return 4
        /sbin/firehol start "$@" > /dev/null 2>&1 || return 1

Now we finally get a result and with INIT_VERBOSE=yes set we do indeed get some useful output:

Nov 27 17:59:38 firehol[27095]: /sbin/firehol: line 33: dirname: command not found
Nov 27 17:59:38 firehol[27095]: /sbin/firehol: line 33: cd: HOME not set
Nov 27 17:59:38 firehol[27095]: /sbin/firehol: line 33: basename: command not found
Nov 27 17:59:38 firehol[27095]: /sbin/firehol: line 36: dirname: command not found
Nov 27 17:59:38 firehol[27095]: Cannot access /install.config
Nov 27 17:59:38 firehol[27095]:!

And this is basically yelling at us that the PATH variable is not set because the script can not find and execute required commands. Sadly this fail is not catched or logged without verbose information and thanks to the /dev/null redirect at all.

At first glance I was going to blame systemd isolating the script from environment variables but that was too fast because setting it explicit changed nothing. To blame is the old set-up logic of the init script /etc/init.d/firehol right at the top not allowing /usr/bin where dirname or basename and others are found.


test -x /sbin/firehol || exit 0

[ -r /etc/default/firehol ] && set -a && . /etc/default/firehol

I compared the /sbin/firehol script of version 2 with version 3 and there is a subtle difference at the start in version 2 that is missing in version 3:

# ------------------------------------------------------------------------------
# ------------------------------------------------------------------------------

export PATH="${PATH}:/bin:/usr/bin:/sbin:/usr/sbin"

I’d argue that version 3 missing this is more correct because setting up the PATH is really the job of the system that is running the script. So basically SysVinit or systemd. Sadly that doesn’t help us here and fiddling with a maintainer provided file is a no go because this will be erased on the next update (if any). Luckily we can see from the init script /etc/init.d/firehol that it also sources the file /etc/default/firehol. This means we can set any additional environment variable here:

# FireHOL application default file
# sourced by the initscript `/etc/init.d/firehol'.


# To enable firehol at startup set START_FIREHOL=YES (init script variable)

After editing this file we finally get some more information and our iptables are piling up with rules again.

● firehol.service - LSB: firehol firewall configuration
   Loaded: loaded (/etc/init.d/firehol; bad; vendor preset: enabled)
  Drop-In: /etc/systemd/system/firehol.service.d
   Active: active (exited) since Fr 2020-11-27 18:17:41 CET; 1s ago
     Docs: man:systemd-sysv-generator(8)
  Process: 14337 ExecStop=/etc/init.d/firehol stop (code=exited, status=0/SUCCESS)
  Process: 14511 ExecStart=/etc/init.d/firehol start (code=exited, status=0/SUCCESS)

Nov 27 18:17:39 systemd[1]: Starting LSB: firehol firewall configuration...
Nov 27 18:17:39 firehol[14511]: Params
Nov 27 18:17:39 firehol[14511]: FireHOL: Saving active firewall to a temporary file...  OK
Nov 27 18:17:40 firehol[14511]: FireHOL: Processing file '//etc/firehol/firehol.conf'...  OK  (470 iptables rules)
Nov 27 18:17:41 firehol[14511]: FireHOL: Activating ipsets...  OK
Nov 27 18:17:41 firehol[14511]: FireHOL: Fast activating new firewall...  OK
Nov 27 18:17:41 firehol[14511]: FireHOL: Saving activated firewall to '//var/spool/firehol'...  OK
Nov 27 18:17:41 systemd[1]: Started LSB: firehol firewall configuration.

Personally I can’t wait for all init scripts to sink into oblivion because debugging this sort of errors is hard and a waste of time and usually revolves about problems solved already in many different ways before – each falling flat in some corner case.

I seldom dabble in the corporate hell of Windows devices but sometimes I have to “use” a laptop to access some VPN to do my magic job and I have no idea how anyone can work like this.

I’m talking about the full set here starting with BitLocker, Cisco AnyConnect (yuk), virus protection and gods know what else.

Every time I start this I get to wait for 2-4 hours until all the updates are done while I’m getting swamped with pop-ups from all kinds of pre installed software each in their individual fashion and style asking me to click, tap, accept, proceed or acknowledge something I’ve no idea about.

Speaking of I usually even have a hard time reading anything on this excuse of a display. For unknown reasons someone thought it’s a good idea to design a default theme with probably fifty shades of grey (I know about high contrast mode but that makes it worse).

I am only a user on such a device without any admin permissions. Why am I even bothered with all this? And while I wiggle my way through all the pop-ups overlapping each other stealing input focus again and again trying to get anything done… Reboot required. Now. Reboot and… repeat! There are more updates we didn’t know of before!

In between an occasional error pops up about something not being able to install something because of some error. The amount of provided information is killing me.

And it’s slow. So gorram slow. What is this thing doing with an i5 processor all the time? And why do I have to babysit it for updates at all?

Eventually I may be able to use the device only to be prompted to change my password due to reasons. And bite me, every time I have to figure out what new password may be fine because the prompt won’t suggest the password rules or anything.

At the end of the day I’m happy that I can use a system again that, as odd as it may seem, provides a much better user experience [to me]: A Fedora Workstation. It just works.

Today I scratched an itch I had with and . Every time I run it on my PC I have to drag around the window until it fills my 3 displays setup. It’s tricky because it’s a grown installation and the displays have different resolutions.

Gnome has smart borders auto-sizing windows when you come close to a border. Usually that’s awesome but in this case it’s not. wmctrl to my rescue!

Find out about current window position when satisfied: wmctrl -G -l -x

Use that information for a one liner script: wmctrl -x -r code.Code -e 0,0,109,5276,1136

This will do until I get a 4k display or learn how to auto-run this snippet on the launch of vscode (like I do this with RisingWorld to force semi borderless fullscreen) 🤣

There’s a weird issue with (snap) on that starts when using voice chat causing really bad lag and short freezes (input, rendering, everything) that became worse over time. My journal filled up with looping messages from appindicator causing this.

appindicatorsupport(at)[2514]: discord1, Impossible to lookup icon for 'discord1_12-panel'

Followed by a JS exception and trace:

JS ERROR: Exception in callback for signal: icon: Error: Argument 'filename' (type filename) may not be null

When I finally found the cause of this I went on looking for a solution and it seems like the unsung hero @3v1n0 fixed this long standing bug like 8 days ago:

Here is a more coherent report on this:

Fun thing is: I only have that indicator because Discord would eventually crash without trying to access this.

Now it’s patched and gone – back to 😁

Man, this is a 180° turn for me. When I started out with Linux the GPUs where usually troublemakers and I kinda got used to throw moar power at it to solve the problem. Spent nights fiddling with Elsa Winner or 3Dfx Vodoo or some ATI cards (that eventual became AMD). When laptops of mine could no longer be used because AMD simply dropped support for perfectly fine hardware I was really never again buying from it again.

The background for finally ripping out the heart of my Linux PC is basically this issue:

NVRM: GPU 0000:01:00.0: GPU has fallen off the bus.

This is followed by a frozen X server rendering all HID interfaces dead until reboot. It happened once or twice a week. NVIDIA support has no idea and while the card is still fine and up for any task I finally decided to get a more recent GPU hoping that the problem will be gone (and not be an issue from the mainboard).

So here I am in 2020 ripping out the heart of my Linux PC.

The decision to try AMD again after a decade was basically made because I read so much positive news on their open source drivers and general good support by Mesa nowadays. Since nothing about the old fglrx days is valid any more this is sort of a jump into cold water for me 🙂

I decided for the slightly older RX 5600 XT 14Gbps 6GB (THICC III Pro) edition by XFX that seems to be good for 1080p gaming and this is close to my main display resolution of 1920×1200. While I never heard of XFX before I was hooked by NO RGB and that tiny vBios switch it has offering a backup bios. That’s a feature I like in my mainboards as well.

Speaking of I heard a lot of confusion on said vBioses on this series so I digged deeper on this topic. Thankfully a lot of the legwork was already done for me by André Almeida who describes the process for Linux PC on after a lot of research in part 1.

With the help of the mentioned tool amdvbflash I was able to drag the following vBios information out of the GPU:

AMDVBFLASH version 4.71, Copyright (c) 2020 Advanced Micro Devices, Inc.

    Product Name is :    NAVI10 A1/A2 D1990301 XLE 6GB 300e/875m 
    Device ID is    :    731F
    Bios Version    :
    Bios P/N is     :    113-170WCNAVIXLE6
    Bios SSID       :    5710
    Bios SVID       :    1682
    Bios Date is    :    03/27/20 21:25 

The extracted rom of the active vBios (switch was set on position closer to power connector) has the sha1sum 9ce7ecc9625d7ff39b3b08c45916b6c2e3bf4a8c and is according to the flashing tool valid and signed. I understood it’s a bad idea to flash with an unsigned rom because the GPU will probably refuse to boot. I’d upload it to techpowerup that seems to collect such roms and allows hassle free downloading but their extract and upload tool seems to be for Windows PC only.

XFX has vBios roms for the 12Gbps variant on it’s website but currently none for this one so it seems it is up to date already.

Installing it was a breeze. I upgraded to Fedora Workstation 32 before changing the GPU to make sure I get more recent drivers and that was it. System booted up just fine and the card worked out of the box. Unlike NVIDIA I didn’t need to download a specific driver first or add some further repository. There’s this nice tool CoreCtrl that shows me a power consumption of only 14W while the card is in idle with zero spinning fans. That’s right – no noise! When not in use this card consumes next to no resources which begs the question how I’m going to heat my man cave from now on 😀

CoreCtrl in action

This surprised me as well. All the cool bits are laid out for me to play with at /sys/class/drm/card1/device/. I mean I’m not much into over-clocking but it’s all there. This brings me to benchmarking the new GPU. I’m not some YTer so you’ll only get the Unigine Superposition benchmark with basically irrelevant OpenGL (I know of no nifty Vulkan benchmark like this yet) for Linux PC gaming.

The tool picked up the wrong model (it really is a FX 5600 XT)

That’s rad! My old Titan X has it’s stronger side in other features but managed only a total result of 3055 without over-clocking on this benchmark.

I’ve not much gaming experience with it yet. I just made sure that Valheim (beta) and X4 Foundations works fine with it. There seems to be some sound issue with Discord + Fullscreen that I found so far but I solved this by switching to windowed mode on the games seeing no difference in FPS.

The one notable issue I have compared with NVIDIA is that I have to set the environment variable DRI_PRIME=1 or games will pick up the integrated GPU of the i5-8600K. This is probably because I’ve a display connected to it as well so it’s active. Going to play with this a little until I get the idea. Will need some adapters first tho 😅

Update: Just as guessed. Once I had all my adapters in place the integrated GPU was not needed any more and does no longer show up. No need to use DRI_PRIME=1 for each 3D application any more.