DaVinci Resolve on Linux - Install issues

attempting to generate optimized media... I get a dialog box at 0% that never changes

Sounds to me like Resolve lacks the rights to write to the assigned path, or something else is holding it up.

By default, Resolve sets everything to /opt/resolve, including the disk database. Check the file paths and, preferably, set them to point somewhere else. You might want to copy that db aside, too.

Sounds to me like Resolve lacks the rights to write to the assigned path, or something else is holding it up.

By default, Resolve sets everything to /opt/resolve, including the disk database. Check the file paths and, preferably, set them to point somewhere else. You might want to copy that db aside, too.

I have looked at /opt/resolve... most of the directories in there are 0777, the files 0666. Also tested writing/touching files here and there... no obvious write permission issues that I can find. Do you know where Resolve builds optimized media by default?

Looking in project settings under "working folders", the cache files location is "CacheClip". Not sure where that is (or if relevant), but clicking browse brings up my resolve media directory (in my homedir)... implying that it is there? I definitely don't have permission issues there.

Looking In preferences under "media storage", the only location I have there is that same resolve media directory. There is a checkbox for "Direct I/O" that is empty... don't know what that means in this context yet.

Look into the manual. It has some ẃell-written passages on the media storage and optimized media that'll really help you along

The media storage is the first mounted file path, and it's where optimized media is created. We use a separate SSD, because it allows quick and easy access to the opt media.

You should relocate the Media Storage and the db on the grounds that it's unwise to keep them on your boot drive. After creating a new db, you can copy/paste or even drag n' drop your projects from one db to another. BMD approves backups.

Relocate the media storage and try again. If Cineform still freezes, restart Resolve and try a few other codecs. If the problem persists, you'll find the crash logs under /opt/resolve. What is Resolve doing when it freezes?

Look into the manual. It has some ẃell-written passages on the media storage and optimized media that'll really help you along

The media storage is the first mounted file path, and it's where optimized media is created. We use a separate SSD, because it allows quick and easy access to the opt media.

OK, thank you... read up a bit more, most I caught on earlier (admittedly skimmy) read. But the bit on CacheClip and hidden folders I had not. Looked in those, the permissions were fine, but I started to suspect the problem was a clash between the apparently autogenerated project/folder names in those and the current installation. Because... I had brought that folder over from my previous machine. Obivously room for trouble there, so I backed everything out. Did an uninstall/purge, wiped out /opt/resolve, and so on. Which I'd already done, but this time I also got rid of everything except my actual media/clips from my main media storage directory. In particular, CacheClip and .gallery.

Then installed and put everything back, without that cruft, building it's database fresh, etc. And voila... that worked. Generates optimized media like a charm now.

You should relocate the Media Storage and the db on the grounds that it's unwise to keep them on your boot drive.

Any particular to Resolve reason, or just in the time-honored system management sense? (With various traditional reasons behind it... separating OS & data, separating sets of simultaneously busily spinning disks, among them.)

I'll probably leave that first one where it is in my home dir, as it is on my fastest disk (Samsung 970 NVMe SSD)... makes sense as a cache location in my mind. I'll have other locations on other drives. I'm OK with the db where it is for now for the same reason, I'm pretty conscientious maintaining /opt, /usr/local, and /home separately from the rest of the OS.

Thank you for your help!

HiDPI again. I have installed 15.1.2.008 and it is up an running but on my 4K laptop screen it is completely and utterly unusable for me. I've googled and searched this forum for awhile but I can't determine if there is a GUI config option or command line flag, or an env var, that might enable HiDPI scaling?

OS is Kubuntu 19.04 with Kubuntu-CI/stable packages (Plasma 5.14.3), nvidia 390 on a GTX 1070 + Intel.

Glad that helped.

Any particular to Resolve reason, or just in the time-honored system management sense? (With various traditional reasons behind it... separating OS & data, separating sets of simultaneously busily spinning disks, among them.)

It's a bit both. For one, it's generally considered proper to keep session files away from /opt, sure. The other bit is that optimized media can clog up the boot drive. Our Mint runs on about 20Gb partition, so do the math. Btw, the same can happen on a Mac, as I once noticed...

Hello, hobbyist videographer here. Trying to get Resolve 15.1.2 to Run on Linux Debian with GTX 1050Ti card. I used Daniel Tufvesson's MakeResolveDeb successfully and the resulting deb installed cleanly. At startup after quick setup I get to see "Checking for Control Surface" for 5 seconds or so, and then crash, from ResolveDebug.txt

Code: Select all

[0x7f4029ddf700] | DtManager            | INFO  | 2018-11-13 20:23:43,345 | Dt Worker Thread Started
[0x7f406aa71c40] | DbCommon2            | INFO  | 2018-11-13 20:23:43,345 | Loading dblist file: /opt/resolve/configs/.dblist
[0x7f40245d6700] | IP                   | INFO  | 2018-11-13 20:23:43,345 | No GPUMapping parameter found for LsManager 3.
[0x7f40245d6700] | IP                   | INFO  | 2018-11-13 20:23:43,346 | Resolve will generate a default mapping.
[0x7f40245d6700] | IP                   | INFO  | 2018-11-13 20:23:43,346 | Using NVIDIA driver '390.87'
[0x7f40245d6700] | GPUManager           | INFO  | 2018-11-13 20:23:43,346 | Let There Be CUDA Light!
[0x7f40235d4700] | GPUManager           | INFO  | 2018-11-13 20:23:43,346 | Initializing GPU board 0 with context thread 0
[0x7f406aa71c40] | IO                   | INFO  | 2018-11-13 20:23:43,468 | Using DNxHR library v2.3.5.43r
[0x7f406aa71c40] | DVIP                 | INFO  | 2018-11-13 20:23:43,955 | DVIP Version : 2.0.0, Revision = 77485 (Release)


Some required information is missing for email crash report, the crash archive will not be updated


==========[CRASH DUMP]==========
Please send this to support:

#TIME Tue Nov 13 20:23:44 2018 - Uptime 00:00:02 (hh:mm:ss)
#PROGRAM_NAME DaVinci Resolve v15.1.2.008 (Linux/Clang)

../bin/resolve() [0x40e8789]
../bin/resolve() [0x40e7d0b]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x128e0) [0x7f40712918e0]
/lib/x86_64-linux-gnu/libc.so.6(+0x9cd36) [0x7f406f406d36]
/opt/resolve/libs/libFairlightPage.so(_ZN6SDebugC1EPKcS1_ibbNS_8CategoryENS_8LoadWhenEb+0x39) [0x7f3ffdc7a499]
/opt/resolve/libs/libFairlightPage.so(+0xbe6ff1) [0x7f3ffdf28ff1]
/opt/resolve/libs/libFairlightPage.so(+0xf80de6) [0x7f3ffe2c2de6]
Signal Number = 11

================================


Searching for this crash in this forum I see a few others having the same crash in that libFairlightPage.so function at startup, but I've found no reply on how to solve it. Any ideas?

Edit: tried 14.3.1 version, get the exact same error with that.

What CPU are you on, AMD or Intel? Some users have reported disabling internal graphics helped. Tried that yet?

There was a fix for what could be the same issue in the 15.1.2 release for Win/Mac. Still to see the affected Linux users confirming they got it, too.

If you need a working Linux Resolve now, you could try v12.5.6.

Just updated to Studio 15.2 - it is still crashing with most Fusion titles as repprted long ago. So, no fix for that. Still the same log entries I posted long ago.

Can anyone confirm this?

What CPU are you on, AMD or Intel? Some users have reported disabling internal graphics helped. Tried that yet?

There was a fix for what could be the same issue in the 15.1.2 release for Win/Mac. Still to see the affected Linux users confirming they got it, too.

If you need a working Linux Resolve now, you could try v12.5.6.

I'm using a Intel Xeon E5-2640, which lacks graphics. I think I've done the appropriate things with all the drivers and so on, and in the log the CUDA seems to be detected.

I tried the newest v15.2 today and got the exact same error.

Then I installed v12.5.6 as you suggested, and indeed that starts and shows the GUI, so it seems to work (haven't played around with it much yet though). I can have great use of version 12 for sure, but it would be nice with the latest since then I get sound , so it would be nice to be able to run that too...

Anyway, thanks for the help!

Sorry, I forgot to mention. v15b1 gave us System Audio. The preceding Linux versions only give audio out via BMD I/O (Mini Monitor, Intensity, etc). Also, do remember that in the pre-15 versions, you'll need to use US keyboard layout to save your grades properly (lofty coding, I know...).

Also, I wonder if it's the Xeon that's giving you a hard time. That's a dicey proposition, because some users have working Xeon configs. Only with Resolve, some versions and parts are more equal than others.

As you scroll down these posts, you'll notice people, suddenly, having issues with AMD and Xeon, after a point release.

Please keep in mind that I was able to make the beta version run on Manjaro (an Arch derivative), but the final version isn't going far, with an AMD.

Just recently bought a dongle to use the studio-version, but ended up with not being able to start it up at all in Linux. I'm using Antergos (Arch, basically).

My hardware is dual Xeon E5-2683v3, 32Gb of RDimm Ram, and two 980Ti's.

i tried every Version of DaVinci from 14.0 to 14.3 and it seems that this problem does not exist with the Versions 14.0 and 14.0.1. All Versions from 14.1 and higher do have this problem in my case! With 14.0 and 14.01 the software starts without any problem!

What you could now do is give v14.01 a go. It won't give you sound, but you'll experience the revamped playback engine

As of 15.2 AMD phenom CPUs (maybe also other non SSE4 enabled CPUs) gets this message:

Nov 14 22:15:27 resolver kernel: [ 1075.542832] traps: resolve[6204] trap invalid opcode ip:7f9ad6d8f518 sp:7ffff8f44670 error:0 in fusionscript.so[7f9ad6ba9000+3a5000]

So last resolve builds are for newer CPU models only

Thanks. My Xeon is indeed a bit old, but it does have sse4.2 so it's not stone age (just yet). I'll see if I can figure out if some missing instructions could be the problem, and I shall also test 14.0.1 and report back.

An idea for a feature in the software would be to actually detect the CPU and complain if it's too old instead of just crashing mystically. A log entry into the ResolveDebug.txt would be enough. Just a thought

(To have optional optimized code running when CPU instructions are available is the standard procedure in software development, you then support both simpler and more advanced/new CPU architectures, but I have come to realize that Resolve comes from a culture where software is sold together with hardware so supporting a wider range of hardware and/or OS versions is probably not on the map.)

The package on AUR is flagged out of date: https://aur.archlinux.org/packages/davinci-resolve/
How can i install it on Arch?

Anyone ideas on this one?
I'm running PopOS 18.04 LTS on a Dell XPS 15 9560. Resolve version is 15.2

I already tried clearing /opt/resolve/config but this does not solve the issue

Code: Select all

ActCCMessage Already in Table: Code= c005, Mode= 13, Level=  1, CmdKey= -1, Option= 0
ActCCMessage Already in Table: Code= c006, Mode= 13, Level=  1, CmdKey= -1, Option= 0
ActCCMessage Already in Table: Code= c007, Mode= 13, Level=  1, CmdKey= -1, Option= 0
ActCCMessage Already in Table: Code= 2282, Mode=  0, Level=  0, CmdKey= 8, Option= 0
PnlMsgActionStringAdapter Already in Table: Code= 615e, Mode=  0, Level=  0, CmdKey= -1, Option= 0
log4cxx: No appender could be found for logger (UI).
log4cxx: Please initialize the log4cxx system properly.
QSocketNotifier: Can only be used with threads started with QThread
New connection received.
Connection is closed.
15.2 (#033) Linux/Clang
Main thread starts: 73F94000
resolve: AppConfig.cpp:171: void AppConfig::LoadAllSiteInfo(): Assertion `m_SiteEnabledIdx > 0' failed.


==========[CRASH DUMP]==========
Please send this to support:

#TIME Thu Nov 15 11:24:17 2018 - Uptime 00:00:00 (hh:mm:ss)
#PROGRAM_NAME DaVinci Resolve v15.2.0.033 (Linux/Clang)

/opt/resolve/bin/resolve() [0x46675d9]
/opt/resolve/bin/resolve() [0x18b6a83]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x12890) [0x7f907de6d890]
/lib/x86_64-linux-gnu/libc.so.6(gsignal+0xc7) [0x7f907be8ee97]
/lib/x86_64-linux-gnu/libc.so.6(abort+0x141) [0x7f907be90801]
/lib/x86_64-linux-gnu/libc.so.6(+0x3039a) [0x7f907be8039a]
/lib/x86_64-linux-gnu/libc.so.6(+0x30412) [0x7f907be80412]
/opt/resolve/bin/resolve() [0x18a4190]
/opt/resolve/bin/resolve() [0x18a2966]
/opt/resolve/bin/resolve() [0x18a2238]
/opt/resolve/bin/resolve() [0x1d5ac21]
/opt/resolve/bin/resolve() [0x1d5a9ef]
/opt/resolve/bin/resolve() [0x18b588e]
/opt/resolve/bin/resolve() [0x189dce4]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7f907be71b97]
/opt/resolve/bin/resolve() [0x189cb3d]
Signal Number = 6

================================
resolve: AppConfig.cpp:171: void AppConfig::LoadAllSiteInfo(): Assertion `m_SiteEnabledIdx > 0' failed.


==========[CRASH DUMP]==========
Please send this to support:

#TIME Thu Nov 15 11:24:17 2018 - Uptime 00:00:00 (hh:mm:ss)
#PROGRAM_NAME DaVinci Resolve v15.2.0.033 (Linux/Clang)

/opt/resolve/bin/resolve() [0x46675d9]
/opt/resolve/bin/resolve() [0x18b6a83]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x12890) [0x7f907de6d890]
/lib/x86_64-linux-gnu/libc.so.6(gsignal+0xc7) [0x7f907be8ee97]
/lib/x86_64-linux-gnu/libc.so.6(abort+0x141) [0x7f907be90801]
/lib/x86_64-linux-gnu/libc.so.6(+0x3039a) [0x7f907be8039a]
/lib/x86_64-linux-gnu/libc.so.6(+0x30412) [0x7f907be80412]
/opt/resolve/bin/resolve() [0x18a4190]
/opt/resolve/bin/resolve() [0x18a2966]
/opt/resolve/bin/resolve() [0x18a2238]
/opt/resolve/bin/resolve() [0x18d6dbd]
/opt/resolve/bin/resolve() [0x18b6cce]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x12890) [0x7f907de6d890]
/lib/x86_64-linux-gnu/libc.so.6(gsignal+0xc7) [0x7f907be8ee97]
/lib/x86_64-linux-gnu/libc.so.6(abort+0x141) [0x7f907be90801]
/lib/x86_64-linux-gnu/libc.so.6(+0x3039a) [0x7f907be8039a]
/lib/x86_64-linux-gnu/libc.so.6(+0x30412) [0x7f907be80412]
/opt/resolve/bin/resolve() [0x18a4190]
/opt/resolve/bin/resolve() [0x18a2966]
/opt/resolve/bin/resolve() [0x18a2238]
/opt/resolve/bin/resolve() [0x1d5ac21]
/opt/resolve/bin/resolve() [0x1d5a9ef]
/opt/resolve/bin/resolve() [0x18b588e]
/opt/resolve/bin/resolve() [0x189dce4]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7f907be71b97]
/opt/resolve/bin/resolve() [0x189cb3d]
Signal Number = 6

================================
Aborted (core dumped)


@Dieter Scheel: Fusion titles work without crash here.

- Fusion tab: Resolve still crashing when merging a background node.
- Still no support for compressed audio files.

using Studio version with usb dongle.

@Marc

Thank you. I now set a missing symbolic link to the Cuda lib - now it works without crashing. Resolve is using hard-coded links to the Cuda library and won't find the libcuda.so on modern Multiarch Linux distributions.

Hello, hobbyist videographer here. Trying to get Resolve 15.1.2 to Run on Linux Debian with GTX 1050Ti card. I used Daniel Tufvesson's MakeResolveDeb successfully and the resulting deb installed cleanly. At startup after quick setup I get to see "Checking for Control Surface" for 5 seconds or so, and then crash

-------

I have found that I need to first use the resolve deb creator, install the .deb, then I have to install Resolve again with the original .sh installer, then it works. I don't know why this works or if it will work for you but it might be worth a try.

14.0.1 crashes in the same way as 14.3, 15.1.2 and 15.2 as reported above, that is in:
/libs/libFairlightPage.so(_ZN6SDebugC1EPKcS1_ibbNS_8CategoryENS_8LoadWhenEb+0x39)

12.5.6 works though and now I've had some time to play with it and it seems stable. I'll try some more with the more recent versions later and see if I can figure out something, at least *why* it crashes even if I don't get it running.

I thought I'd try this again as I've installed Pop!_OS 18.10 and have a later Nvidia driver version (410.57)

I followed the steps for MakeResolveDeb and installed the .deb, but I still have the same issue. When I click the Davinci Resolve icon nothing happens.

The MakeResolveDeb website said to run resolve from a prompt and check the output for error messages:

Code: Select all

$ /opt/resolve/bin/resolve
/opt/resolve/bin/resolve: error while loading shared libraries: libOpenCL.so.1: cannot open shared object file: No such file or directory


Hopefully this is the issue. The problem is that I'm completely new to Linux, so I don't know how to get this libOpenCL.so.1

Could someone be so kind to help a newb out

Install the OpenCL dev files via Terminal.

Code: Select all

sudo apt-get install ocl-icd-opencl-dev

Resolve wants both OpenCL and CUDA present before startup.

I don't know what that means, but it worked
Oh man, I've missed Resolve. Thank you!

If I don't have too many issues I'll probably just stay on this version as I don't want to break it
But if I want to upgrade to a newer version of Resolve, should I first uninstall the one I have installed, or isn't that necessary?

I don't know what that means, but it worked
Oh man, I've missed Resolve. Thank you!

If I don't have too many issues I'll probably just stay on this version as I don't want to break it
But if I want to upgrade to a newer version of Resolve, should I first uninstall the one I have installed, or isn't that necessary?

I'm glad you got it running. Since you used MakeResolveDeb for installing there is no need to uninstall the previous version when upgrading. Just make a new *.deb file and install it. MakeResolveDeb is prepared for this and the magic of the Debian package manager will take care of the upgrade. It is also possible to downgrade this way but be careful since the Resolve database and configuration files may not be backward compatible in all situations.

It's the man himself, tjänare. Your effort is much appreciated!

If I understand correctly, the Linux version of Resolve does not support mp4-files.
What's the recommended format to convert mp4's to for editing in Resolve, and what converter should I use, ffmpeg?

If I understand correctly, the Linux version of Resolve does not support mp4-files.

Only Resolve Studio supports h.264 under Linux.

What's the recommended format to convert mp4's to for editing in Resolve, and what converter should I use, ffmpeg?

Yes, that's the best option. I use ffmpeg and transcode everything to ProRes or DnxHD. Using h.264 directly for editing is not really recommended.

It would make things easier if you'd add your system info to your signature field.

What input parameters do you use for DNxHD?
I tried, but got an error

That's alright. With FFMpeg, Avid codecs have more mandatory parameters than, say, Prores. You need to set the frame rate, colour space, resolution and bitrate, according to Avid guidelines. Here's an example:

Code: Select all

ffmpeg -i <input file> -c:v dnxhd -vf "scale=1920:1080,fps=25,format=yuv422p" -b:v 120M -c:a pcm_s16le -ar 48000 output.mov

I also tend to set the audio sample rate to 48kHz, so that the output audio will definitely import properly to Resolve.

There's a good write-up on Avid & FFMpeg here, which will lead you further along.

That worked, thanks! And then I'm guessing I have to use a script to batch convert?

One last question, I want to end up with a network optimized mp4-file. Is it sufficient to export as Quicktime MPEG4 Video in Resolve with network optimized checked and then use ffmpeg to change from .mov to .mp4, or is it a better way?

You must seperate container (.mov .mp4 .mkv etc.) from the codec (h.264, h.265 etc.) used. Only the codec matters - a container may contain (hence the name) different codecs. If you encode your edited video to h.264 then you can use (or more specific: multiplex) either into .mov, .mp4 or .mkv container. The container depends on the further use or traget of the video.

In your example I would use a mp4 or mov container with network optimizations. I use Handbrake to encode from high-bitrate video to network optimized h.264 video. That gives me more flexibility regarding all the parameters. Resolve is quite limited when it comes to export formats.

I got 15.2 running on my setup. My Xeon CPU is fine, it was not that. I'm quite sure this is a real bug in Resolve, triggered by which combination of dynamic libraries that get loaded (and maybe in which order). Anyway, if of you have a crash in:

/opt/resolve/libs/libFairlightPage.so(_ZN6SDebugC2EPKcS1_ibbNS_8CategoryENS_8LoadWhenEb+0x39)

during startup, you are probably experiencing the same bug as I.

The following explanation is going to be a bit technical, I hope a developer at BMD will read and fix the bug. I'm a professional software developer, but not a specialist at debugging binaries (especially those without debug symbols) so I can't guarantee that the description of what happens is 100% correct, but hopefully it provides enough leads for a BMD developer anyway.

When libFairlightPage.so is loaded a function named SDebug::SDebug() is called repeatedly. I'm not sure what it does exactly, but it seems to be filling in data in the BSS, some sort of table of debug information. One of these entries is called "filter". By some reason instead of filling in the data in libFairlight.so's own "filter" symbol it picks the "filter" symbol from libncurses, which of course is read-only and you get a segmentation fault.

Resolve does not directly link to libncurses. So how does it get loaded? In my system the GL library loads swrast_dri.so which in turn links to libncurses. Probably the "fault" here is that the software rasterizer (swrast_dri) gets loaded at all. My guess is that on systems where Resolve works out of the box, the software rasterizer does not get loaded and thus not libncurses, and no symbol mixup can occur.

However it's clearly a bug that SDebug::SDebug() setup can't differ between libFairlightPage.so's own symbols and symbols in other libraries that happen to have the same name, so it would be nice if BMD could fix this, it would make the software more robust.

To get Resolve up and running in my system I simply created a copy of libncurses and renamed the filter symbol to something else by editing the binary directly. I'll look into why swrast_dri.so get loaded though (despite that I have nvidia drivers), as I think if I can avoid that I can get resolve running without this hack.

A quick look into why libncurses gets loaded which triggers a bug in libFairlightPage.so SDebug::SDebug to write to read-only memory during startup:

My Debian system (running the "bleeding edge" version, unstable) uses a vendor-independent libGLX.so (called libglvnd) which means you can have nvidia, mesa etc drivers simultaneously on the system without conflicts. LibGLX.so then automatically decides which driver to load, and it can actually load more than one. On my system it loads both mesa and nvidia for resolve, and when mesa is loaded, the software rasterizer gets loaded too which in turn loads libncurses, and we have a problem.

So why not just remove Mesa? Indeed if I do libncurses won't be loaded and we get past that problem, but then I get a crash later which seems to be due that the QT library needs some feature that libGLX thinks requires Mesa. Probably it would work if I had just native Nvidia drivers installed and didn't use libglvnd at all, but the Debian system doesn't allow you to set it up that way (or at least it seems to be really messy to do so), so I'm stuck with it.

So the work-around I'm using now is as described in my previous post: make a copy of libncurses with filter symbol renamed and place that in /opt/resolve/libs.

What to do when the render button is greyed out? Sometimes I get an error message about CUDA something something not being available, is that the issue? Nvidia gtx 1070.

I have done some testing of GPUs and drivers and want to share it with other ArchLinux users.

Intel igpu:
I have tested all opencl-intel providers and the problem seems to be that dr do not recognizes some of them as qualified opencl platform. I did all tests in dr v15.2.2.007 and with my i7-8700 (UHD Graphics 630 Desktop). The format is following: opencl-driver (file in /etc/OpenCL/vendors) | result

• compute-runtime-18.42.11702-1 (intel.icd) | Skipping unqualified OpenCL platform: Intel(R) OpenCL HD Graphics
• compute-runtime-bin-18.51.12049-1 (intel-gpu.icd) | Skipping unqualified OpenCL platform: Intel(R) OpenCL HD Graphics
• beignet 1.3.2+12+gfc5f430c-1 (intel-beignet.icd) | Skipping unqualified OpenCL platform: Intel Gen OCL Driver
• intel-opencl-5.0.r63503-1 (intel-gpu.icd) | No info about quality, dr crashed
  OpenCL error -1001, GPUPropertiesUtilUnix.cpp:338.
  terminate called after throwing an instance of 'ReactorGPU::GPUException'
  Possibly not suitable for my igpu.
• intel-opencl-runtime-1:18.1.0.013-1 (intel.icd) | No info about quality, dr crashed
  OpenCL error -1001, GPUPropertiesUtilUnix.cpp:338.
  terminate called after throwing an instance of 'ReactorGPU::GPUException'
  Possibly not suitable for my igpu.
  When starting in terminal, dr immediately spits out crash report. But it hangs at QProcess: Destroyed while process ("/bin/grep") is still running. Then after pressing ctrl+c to terminate process, it starts splash (why?) and after discarding to update configuration in dialog window, it still hangs in terminal and even ctrl+c does not help. You only can close terminal. And in process tree you can see DaVinciPanelDae zombie process.
  Then when you try to start resolve again, it says that Another DaVinci Resolve instance is already running
  and you can see a lot of DPDecoder zombie processes.

I got no success in any configuration. We will wait when bmd fix it. See here.

Amd igpu:
I have no such apu, so could not test. If you tested it, please share your results.

Intel igpu + Nvidia optimus:

• GPU model: GeForce 920M
  davinci-resolve-beta: 16.0b6-1, linux: 5.2-arch2-1, xorg-xserver: 1.20.5-2, nvidia: 430.26-9, opencl-nvidia 430.26-1, cuda: not installed
  
  • Using nvidia method: works normally
  • Using PRIME with nouveau method: not tested, because I am sure it wouldn't work with open source drivers
  • Using Bumblebee method: works partly
    bumblebee: 3.2.1-20, primus: 20151110-8, cuda: not installed
    It requires that some implementation of OpenCL (f.e. opencl-nvidia) to be installed (otherwise crashes), but actually falls back to cuda. If you manually set to use OpenCL instead of cuda, it will fail after relaunching. No matter you run it with optirun or primusrun. To get it back to working condition, remove ~/.local/share/DavinciResolve. Btw, it works on its own shipped cuda libraries, not archlinux's cuda libraries.
  • Using nvidia-xrun method: works normally
    nvidia-xrun-git: 0.3.79.gaf3b734-1, bbswitch: not installed

Note that dr will still crash for above methods if you additionally install proprietary implementation of amd opencl. Probably in future releases of dr it may be fixed by bmd.
update 2020 06 08: need to test with prime + nvidia, because proprietary driver now supports prime.

Nvidia gpu:

Tested configurations in the following format:
GPU model | linux version | xorg-server version | davinci-resolve version | drivers versions | result

• GeForce GTX 1060 | 4.19.4.arch1-1 | 1.20.3-1 | 15.2.1.005 | xf86-video-nouveau 1.0.15-3, opencl-nvidia 415.18-1 | fail - Aborted (core dumped).
  In ResolveDebug.txt I have: Updating display GPU information...
  OpenCL error -1001, GPUPropertiesUtilUnix.cpp:338.
  terminate called after throwing an instance of 'ReactorGPU::GPUException'
  terminate called recursively
• GeForce GTX 1060 | 4.19.4.arch1-1 | 1.20.3-1 | 15.2.1.005 | nvidia 415.18-4, opencl-nvidia 415.18-1 | success
• GeForce GT 630 | | | 15.2.2.007 | nouveau driver in use | fail - splash screen disappears after several seconds.
  No info about quality, dr crashed
  OpenCL error -1001, GPUPropertiesUtilUnix.cpp:338.
  terminate called after throwing an instance of 'ReactorGPU::GPUException'
• GeForce GT 630 | 4.19.4.arch1-1 | 1.20.3-1 | 15.2.1.005 | nvidia 415.18-4, opencl-nvidia-415.18-1 | success

It works fine only with proproetary driver. Install proprietary driver and blacklist/uninstall opensource driver.
Note, that even when using nvidia driver, dr will crash after few seconds of showing splash screen if you forget to update opencl-nvidia to match nvidia driver version. I mention it, because I was experimenting a lot, constantly installing and removing opencl and video drivers, and at that time repo update was made. Partial updates are not supported in Arch, so keep an eye for versions to be matched.
Bmd should at least prevent crashes when nouveau driver in use and opencl-nvidia installed.
Also with nvidia kernel module in use we got this in log:
Found qualified OpenCL board 0: platform - NVIDIA CUDA, vendor - NVIDIA Corporation, id - 0x7f57aac8cb30, device name - GeForce GT 630 id - 0x7f57aac8cbd0, global memory size (MB) - 2002, local memory size (KB) - 48, max global memory alloc size (MB) - 500
independently of the fact if cuda package installed. Looks like it always uses opencl on nvidia, but not opencl. I do not know if this is actually true, but made this note for developers.

AMD GPU:
This is when you have amd gpu only or amd + intel while amd is set as primary in bios.
As of dr v16.0 (#044), such configuration requires proprietary opencl implementation for amd and proprietary opengl implementation for amd.

Testing opencl implementations results in the format:
opencl-driver (file in /etc/OpenCL/vendors) | kernel driver in use | result

• opencl-mesa-18.3.1-1 (mesa.icd) | amdgpu | Skipping unqualified OpenCL platform: Clover
• opencl-amd 18.50.708488-1 (amdocl-orca64.icd and amdocl.icd) | amdgpu | Found qualified OpenCL board 0: platform - AMD Accelerated Parallel Processing, vendor - Advanced Micro Devices, Inc., id - 0x7feb0742de30, device name - Radeon RX 580 Series id - 0x7feb0bfa5610, global memory size (MB) - 7535, local memory size (KB) - 32, max global memory alloc size (MB) - 4048

Davinci resolve will crash if no suitable opencl-driver is installed.

Tested configurations

• GPU model: Radeon RX580
  davinci-resolve-beta: 16.0b6-1, linux: 5.2.1.arch1-1, xorg-server: 1.20.5-2, xf86-video-amdgpu: 19.0.1-1, mesa 19.1.2-1, amdgpu-pro-libgl: 19.30_838629-1, opencl-amdgpu-pro-orca: 19.30_838629-1
  Works normally.

I also would like to test it with Vega and later gpu cards, but I do not have them. If you have such hardware, please share your test results.

Intel (primary) + (AMD or NVIDIA)
This is amd hybrid graphics laptops. It also may be a desktop with intel + amd or nvidia when intel is set as primary in bios. On my desktop it automatically becomes primary if I turn on a desktop with monitor cable attached to intel and not attached to amd. Intel's opengl is suitable (you do not need proprietary amd opengl), but you still need to install proprietary opencl for amd or nvidia.

Tested configurations

• Radeon RX580 | linux 4.20.arch1-1, xorg-server 1.20.3-1, dr 15.2.2.007, xf86-video-amdgpu 18.1.0-1, mesa 18.3.1-1, opencl-amd 18.50.708488-1 | success - amdgpu (not pro) driver in use.
• Radeon R7 M265 | dr 15.2.1, opencl-amd 18.30.641594-1, radeon kernel module in use | fail - crash. In logs I see: amdgpu_device_initialize: DRM version is 2.50.0 but this driver is only compatible with 3.x.x.
• Radeon R7 M265 | dr 16.2.2-1, opencl-amd 20.10.1048554-1, amd kernel module in use | semi-success - started like normal, then showed error window "No OpenCL boards available on chassis 1", after clicking ok, another window appeared "CRITICAL_PREF: DaVinci Resolve could not find any OpenCL capable GPU's. Would you like to review and update your DaVinci Resolve Hardware Configuration? Please ensure that appropriate GPU drivers are installed and your DaVinci Resolve hardware is configured appropriately.", then after pressing ok, looks like it is normal. But actually it is not working. When you add any clip (even just an jpg or png image) to the timeline, it is not able to display it or play. Here is a youtube video (start watch at 2:33):
  

Other:

• no any opencl-driver (force removed dependency with pacman -Rdd) | Not info about quality, dr crashed
  OpenCL error -1001, GPUPropertiesUtilUnix.cpp:338.
  terminate called after throwing an instance of 'ReactorGPU::GPUException'
  This test was done while nvidia driver in use while gt630, uhd640, rx580 presented in system. In some intel opencl cases output looks the same. Means that dr not thinks that they are not qualified opencl platforms, but even do not sees them.
• pocl-git v1.2.r111.g94fba9f5-1 (pocl.icd) | Skipping unqualified OpenCL platform: Portable Computing Language
  But interestingly, while at least one icd loader exist, dr not crashes immediately with error -1001.
  And even more interesting, in that case (unqualified platform) it allows to use cuda (of course nvidia driver in use) while opencl-nvidia and cuda are not installed! I think bmd could check if nvidia gpu is presented and work with it using cuda while there is no any opencl icd instead of just crashing.

Currently, Davinci Resolve (15.2 at the moment of writing) in Linux has issue that it is not showing frame border with control buttons (minimize, maximize, close). It is not an install issue, but still a Linux only, so I decided not to create a new thread for this.

I have discovered a way how you can workaround this. At least in kde.
Press alt+f3 in the main window, go to More Actions -> Special window settings. A new window will appear.
At Window matching tab set Window types to Normal Window. Then in Appears & Fixes tab check No titlebar and frame checkbox and set it to Force No.
The same could be done for the second window (for dual screen mode), but its geometry also should be fixed.
And if you want to create a rule for splash screen, be aware that it is Dialog window, but not splash.

You can download a .kwinrule file and import it in your kde settings. Here it is: https://gist.github.com/Ashark/74034a4f ... acc7b4ce2e

When I open Resolve I first get this:

No CUDA boards available on chassis 1

Then this:

CRITICAL_PREF: DaVinci Resolve could not find any CUDA capable GPUs. Would you like to review and update your DaVinci Resolve Hardware Configuration?
Please ensure that appropriate GPU and CUDA drivers are installed and your DaVinci Resolve hardware is configured appropriately.

Then everything seems to be normal, but the render button is greyed out.

Edit: Not sure what the error messages are, but the render button stops being greyed out when I bring in some actual footage. I was just testing by putting a solid color generator on the timeline previously.

I was just testing by putting a solid color generator on the timeline previously.

It's the same here. If you add a title or other effect you can then add it to the render queue.

Are you absolutely sure your Cuda drivers are installed properly because I don't have those error messages on my system?

I haven't installed anything other than the Nvidia version of Pop!_OS. The weird thing is that I don't always get the error messages, they seem to appear if I've had my computer running for a long time without a restart.

I tried to open Resolve right now as I had a restart earlier today and now I got no error messages.

@Michael Hebo

Well, not much I can say - could be a system problem or a hardware failure (temperature of a component etc.). Or something is running in the background messing up everything. Maybe a memory leak somewhere. is there any other software (daemon) started who uses your NVidia card? Do you have any entries in your systems log about a hardware failure?

Maybe you could try a different, mainstream distribution like LinuxMint, Ubuntu, openSuse or even CentOS just to be sure it is not your hardware.

Hey there. Please let me know if I should make a new thread, but this is the only one I've found referencing BlackMagic RAW (.braw) on Linux.

I've got Resolve up and running on Ubuntu 18.04, as far as I can tell everything is setup and working well -- except for .braw .
I can open projects with it inside most of the time (created on Windows) but the footage is just black with some glitching. Trying to import footage directly while on Ubuntu just crashes the software usually.
Anything obvious I should try? Is this a known issue? Should I make a separate thread?

Hey there. Please let me know if I should make a new thread, but this is the only one I've found referencing BlackMagic RAW (.braw) on Linux.

I've got Resolve up and running on Ubuntu 18.04, as far as I can tell everything is setup and working well -- except for .braw .
I can open projects with it inside most of the time (created on Windows) but the footage is just black with some glitching. Trying to import footage directly while on Ubuntu just crashes the software usually.
Anything obvious I should try? Is this a known issue? Should I make a separate thread?

Hi Adam! I like your videos. Regarding braw, I have had the same experience. Only way I have found working is by disabling GPU for braw decoding.

Hi Adam! I like your videos. Regarding braw, I have had the same experience. Only way I have found working is by disabling GPU for braw decoding.

oh! I misread that setting. I'll give it a whirl, thanks.

when mesa is loaded, the software rasterizer gets loaded too which in turn loads libncurses, and we have a problem.

Thanks, that was a pretty solid look-up.

Seems to me like libGLX is triggering a glitch that can be avoided with propietary drivers. It is a bug, and it should be fixed. That driver solution is currently a bit too advanced for Resolve

hello im new to user and i hav installed Davinci resolve 15 studio in centos 7 operating system...

And now i am not able to run the Davinci resolve 15 studio

in Terminal when i hit the command

[admin@localhost ~]$ /opt/resolve/bin/resolve
ActCCMessage Already in Table: Code= c005, Mode= 13, Level= 1, CmdKey= -1, Option= 0
ActCCMessage Already in Table: Code= c006, Mode= 13, Level= 1, CmdKey= -1, Option= 0
ActCCMessage Already in Table: Code= c007, Mode= 13, Level= 1, CmdKey= -1, Option= 0
ActCCMessage Already in Table: Code= 2282, Mode= 0, Level= 0, CmdKey= 8, Option= 0
PnlMsgActionStringAdapter Already in Table: Code= 615e, Mode= 0, Level= 0, CmdKey= -1, Option= 0
15.2 (#033) Linux/Clang
Main thread starts: 499F6B00
[0x7f7a499f6b00] | Undefined | INFO | 2018-11-28 16:41:34,779 | --------------------------------------------------------------------------------
[0x7f7a499f6b00] | Undefined | INFO | 2018-11-28 16:41:34,780 | Loaded log config from /opt/resolve/configs/log-conf.xml
[0x7f7a499f6b00] | Undefined | INFO | 2018-11-28 16:41:34,780 | --------------------------------------------------------------------------------
Aborted (core dumped)


I needed help for the same to resolve my issue

Hello

hello im new to user and i hav installed Davinci resolve 15 studio in centos 7 operating system...

Well, how could we even begin to help you if you don't even post anything useful about your hardware and software?

Maybe put some of the information into your signature like some users, me included, do.

@Dieter Scheel: Fusion titles work without crash here.

- Fusion tab: Resolve still crashing when merging a background node.
- Still no support for compressed audio files.

using Studio version with usb dongle.

You might be interested by my recent post (regarding the Background node) here: viewtopic.php?f=21&t=79680&p=457115#p457528

In short, add a symlink with ln -s /usr/lib/x86_64-linux-gnu /usr/lib64 and it might fix the crash. (It works for me for this very specific issue.)

Hi,
It seem the treat for linux so is I'm wrong sorry for that.

My issue is the video doesn't show up.
(See the picture)

https://drive.google.com/file/d/1lXKldazZhwoWzs6xvQW1Lcq2obHSQ9b0/view?usp=sharing
the videos are in .mov

https://drive.google.com/file/d/1keUKZXZ8UPA2Vs_7LtX1orwmZ_rChkdk/view?usp=sharing
I start the project on Windows and I import it on linux version.
On windows, there are no problem.

Have you an idea ?
I am on linux Mint 19
Thx

Hi,
It seem the treat for linux so is I'm wrong sorry for that.

My issue is the video doesn't show up.
(See the picture)

https://drive.google.com/file/d/1lXKldazZhwoWzs6xvQW1Lcq2obHSQ9b0/view?usp=sharing
the videos are in .mov

https://drive.google.com/file/d/1keUKZXZ8UPA2Vs_7LtX1orwmZ_rChkdk/view?usp=sharing
I start the project on Windows and I import it on linux version.
On windows, there are no problem.

Have you an idea ?
I am on linux Mint 19
Thx

It looks like your media needs to be relinked. Did you perform a relink after project import?

Yes, they are relinked.
That's the problem.

Yes, they are relinked.
That's the problem.

Ah, okay. What codecs are you using? It may be that. The codec support is limited in Linux compared to Windows. Especially for the non Studio version of Resolve.

It's H264-MPEG-4 AVC.
Gosh !

Is there a way to fix it ?