Fusion Titles make Davinci 15.0 crash

I have the behaviour that davinci crashes instant if i drop a title template on the timeline.
If i load the title directly into fusion (dropping the settings file) and playing it in fusion directly it works. But if i drop it to the timeline, davincy freezes for some seconds and crashes.

I also attached the log.

I also enabled to run Fusion on CPU only!
GPU: GeForce GTX 750 Ti
Ryzen 2700X, 8 Core CPU
32 GB Ram DDR4 3200
OS: Linux
GPU: GeForce GTX 750 Ti

I use the proprietary drivers from Nvidia.
Version: 396.45
Linux-x86_64
Desktop: KDE
I also have 4 monitors attached to the Graphics-card
Linux kernel is 4.17.14


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

#TIME Sun Aug 12 13:30:23 2018 - Uptime 00:02:13 (hh:mm:ss)
#PROGRAM_NAME DaVinci Resolve Studio v15.0.0b.080 (Linux/Clang)

bin/resolve() [0x40ee299]
bin/resolve(_ZN7BtDebug14ReportSegfaultEiP7siginfoPv+0x6c) [0x40ed89c]
/usr/lib/libpthread.so.0(+0x123b0) [0x7fcb3e6dc3b0]
/usr/lib/libc.so.6(gsignal+0x10f) [0x7fcb3cbd3b5f]
/usr/lib/libc.so.6(abort+0x125) [0x7fcb3cbbe452]
bin/resolve() [0x4bc1701]
bin/resolve() [0x4fabfd8]
bin/resolve() [0x47fe9e3]
bin/resolve() [0x479ea17]
bin/resolve() [0x443bb40]
bin/resolve() [0x443f333]
bin/resolve(_ZN10ReactorGPU18SingleBoardManager11_threadProcEPv+0x6) [0x443d6f6]
/usr/lib/libpthread.so.0(+0x7a8d) [0x7fcb3e6d1a8d]
/usr/lib/libc.so.6(clone+0x43) [0x7fcb3cc97823]
Signal Number = 6

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

[Peter Chamberlain]

Do u still get the crash with a single UI monitor connected?

Will check that in a minute. Here is also The Diagnostics Menu Logfile

I disabled the other 3 screens using the system settings in KDE. Is that enough or should i phyiscally detattch them and reboot?

UPDATE: Physically detached 3 monitors. The one left is connected to HDMI. Rebooted System. Still crash.

This is probably the same as was previously been reported here :
viewtopic.php?f=32&t=77491

I had the same problem with beta8, but beta7 works fine. Could you try the same with beta7 ?
This might help Blackmagic pinpoint the problem...

Is it enough to just run the 07 install script over the current installation? Still crashed
Deleted old install. Did a new install with beta 7, still crashed. Running Fusion on CPU only.

Also have same problem and I reported. B7 working just fine though. You need to completely reinstall B8 in order to install B7.

I had the same issue with an NVIDIA card. I solved it by using the NVIDIA web drivers instead of the built in Mac drivers. Not sure if you are able to update ATI drivers or not, but if you can it might be worth a shot.


Sent from my iPhone using Tapatalk

Tried 15.0 today. Same.
Could it be my graphics card here? Has only 2GB RAM but i am not loading anything big here.
Doesnt sound very reasonable and the only thing i did was put a title to the timeline.

Digged a little around the logs.

CUDA did get out of memory so i tested openCL after. Both have out of memeory fails with like a simple text. Does Davinci initially need like so much vRAM even if you do nothing?

[0x7f88ee095ac0] | SyManager.Signals | INFO | 2018-08-14 21:18:13,799 | Main view page is changed to 16
CUDA error 2: 'out of memory', SimpleCudaMemCacheMgr.cpp:142.
[0x7f889c7b8700] | GPUManager | WARN | 2018-08-14 21:18:13,823 | Exception caught for allocate Image 31.6 MiB. Clearing pool to retry...
[0x7f88ee095ac0] | SyManager.Signals | INFO | 2018-08-14 21:23:30,358 | Main view page is changed to 2
CUDA error 2: 'out of memory', SimpleCudaMemCacheMgr.cpp:142.
[0x7f889c7b8700] | GPUManager | WARN | 2018-08-14 21:23:35,450 | Exception

Code: Select all

*** SetFilePath[969]: Unsupported data handler type rlhd and subtype  lru
*** SetFilePath[969]: Unsupported data handler type rlhd and subtype  lru
OpenCL error CL_MEM_OBJECT_ALLOCATION_FAILURE:
[0x7f9b29a85700] | GPUManager           | ERROR | 2018-08-14 21:51:14,387 | OpenCL error detected: CL_MEM_OBJECT_ALLOCATION_FAILURE error executing CL_COMMAND_NDRANGE_KERNEL on GeForce GTX 750 Ti (Device 0).

bin/resolve[0x4ba8539]
bin/resolve[0x4fa56f4]
bin/resolve[0x4fab83f]
bin/resolve[0x4800cd3]
bin/resolve[0x47a0517]
bin/resolve[0x443d860]
bin/resolve[0x446809d]
bin/resolve[0x4440da3]
bin/resolve(_ZN10ReactorGPU18SingleBoardManager11_threadProcEPv+0x6)[0x443f166]
/usr/lib/libpthread.so.0(+0x7a8d)[0x7f9b7734ba8d]
/usr/lib/libc.so.6(clone+0x43)[0x7f9b75911823]
[0x7f9b2d33c700] | DVIP                 | ERROR | 2018-08-14 21:51:14,389 | Exception caught while running GPU algorithms: clEnqueueNDRangeKernel (CL_MEM_OBJECT_ALLOCATION_FAILURE)
[0x7f9b2d33c700] | GPUManager           | INFO  | 2018-08-14 21:51:14,389 | Flushing GPU memory...


Code: Select all




I did much more testing and played around with resolution/bit depth, proxy cache and stuff. I achived to have a scene running where it did not crash.
What now is beeing outputted to the console of resolve is this: Warning: exception during GPU buffer upload

p.s what does it mean if a small red bar appears above the timeline when you drop in a fusion title? is that some kind of auto cache?

I think thats the problem. I now did the following, all in the playback Menu:

-Playback, Proxy Mode half-resoultion
-Render Cache-User
-Fusion Memory Cache off
-i also checked always run fusion on CPU works
-then i tested with openCL, works
-then tested cuda, works
to be save i deleted the render cache...all

now i dropped a fusion title in the timeline. no red bar and no crash.
if i now activated SMART cache the red bar pops up and instant crash.