Rawtherapee forces computer shutdown

Software RawTherapee
1

My computer started to shutdown fast and unexpected when I try to export the raw pictures to any format using rawtherapee 5.2. This happened on my old Debian installation since upgrading from Debian 8 to 9 as well as on my current Ubuntu GNOME 17.04 installation. I also tried it with the LXDE desktop without success.

I’ve tried to uninstall the Nvidia legacy driver 340, but it was no successful. Also tried to export via rawtherapee-cli line-command, it turned off too.

This is my inxi -Fx output:

System:    Host: morais-R590-P-BE54P1 Kernel: 4.10.0-19-generic x86_64 (64 bit gcc: 6.3.0)
       Desktop: Gnome 3.24.0 (Gtk 3.22.11-0ubuntu3) Distro: Ubuntu 17.04
Machine:   Device: other-vm? System: LG product: R590-P.BE54P1 v: TBD
       Mobo: Intel model: N/A BIOS: INSYDE v: QL4L3G62 date: 04/15/2010
CPU:       Quad core Intel Core i7 Q 720 (-HT-MCP-) cache: 6144 KB
       flags: (lm nx sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx) bmips: 12769
       clock speeds: max: 1600 MHz 1: 933 MHz 2: 1466 MHz 3: 1066 MHz
       4: 1066 MHz 5: 1066 MHz 6: 1066 MHz 7: 933 MHz 8: 933 MHz
Graphics:  Card: NVIDIA GT215M [GeForce GT 335M] bus-ID: 01:00.0
       Display Server: X.Org 1.19.3 driver: N/A
       Resolution: 1600x900@59.98hz
       GLX Renderer: Gallium 0.4 on NVA3
       GLX Version: 3.0 Mesa 17.0.3 Direct Rendering: Yes
Audio:     Card-1 Intel 5 Series/3400 Series High Definition Audio
       driver: snd_hda_intel bus-ID: 00:1b.0
       Card-2 NVIDIA High Definition Audio Controller
       driver: snd_hda_intel bus-ID: 01:00.1
       Sound: Advanced Linux Sound Architecture v: k4.10.0-19-generic
Network:   Card-1: Realtek RTL8191SEvB Wireless LAN Controller
       driver: rtl8192se port: 3000 bus-ID: 02:00.0
       IF: wlp2s0 state: up mac: 00:17:c4:e9:68:3e
       Card-2: Qualcomm Atheros AR8131 Gigabit Ethernet
       driver: atl1c v: 1.0.1.1-NAPI port: 2000 bus-ID: 03:00.0
       IF: enp3s0 state: down mac: c8:0a:a9:62:68:0c
Drives:    HDD Total Size: 500.1GB (82.0% used)
       ID-1: /dev/sda model: FUJITSU_MJA2500B size: 500.1GB temp: 32C
Partition: ID-1: / size: 17G used: 4.0G (26%) fs: ext4 dev: /dev/sda5
       ID-2: /home size: 433G used: 370G (90%) fs: ext4 dev: /dev/sda6
       ID-3: swap-1 size: 10.11GB used: 0.00GB (0%) fs: swap dev: /dev/sda1
RAID:      No RAID devices: /proc/mdstat, md_mod kernel module present
Sensors:   System Temperatures: cpu: 67.0C mobo: N/A gpu: 59.0
       Fan Speeds (in rpm): cpu: N/A
Info:      Processes: 251 Uptime: 12 min Memory: 1755.6/3939.4MB
       Init: systemd runlevel: 5 Gcc sys: 6.3.0
       Client: Shell (bash 4.4.51) inxi: 2.3.8

I can’t even imagine what’s happening. I was using Rawtherapee 5.0.1 and upgraded do 5.2, and the problem still persists.

2

A couple of hypotheses:

  1. overheating your cpu
  2. running out of RAM and swap

What you can try:

  1. limit rawtherapee to a subset of the cores, taskset should do
  2. limit the amount of memory that RT can use, see ulimit for that
2 Likes
3

Furthermore:

In that order.

4

Well, I’ve started the exportation side-by-side with system monitor and it’s really the CPU. All the 8 cores are being used. I will search how to use taskset and see if it solves.

1 Like
5

I ran taskset to limit 4 cores to rawtherapee and it crashed the system again. But with only 2 cores it worked fine. Now I’m gonna make this limitation permanent. Thank you for help.

6

No cpu fan?

1 Like
7

@paulomorais1981 this is very likey a symptom of a larger problem with your CPU. I would investigate, maybe your fan doesn’t work properly as @heckflosse suggested…

8

Just thinking out loud, you might want to check your cpu cooling settings in BIOS, make sure you are not in quiet mode, etc.

9

Of course that’s a solution. But the better solution would be to get the power of all cores for rt by searching for the real issue and fixing it!

10

It is not unusual that Linux can’t determine the fan speed.

11

Yeah. You’ve made me think about it. I don’t understand about these things, but’s nothing that I can’t research. It has happened before, but less times. It just became more frequent. I will find a solution.

Thank you all guys.

12

I’ve used sensors to detect the CPU temperature, and I got this:

coretemp-isa-0000
Adapter: ISA adapter
Core 0:       +75.0°C  (high = +84.0°C, crit = +100.0°C)
Core 1:       +75.0°C  (high = +84.0°C, crit = +100.0°C)
Core 2:       +77.0°C  (high = +84.0°C, crit = +100.0°C)
Core 3:       +81.0°C  (high = +84.0°C, crit = +100.0°C)

acpitz-virtual-0
Adapter: Virtual device
temp1:        +77.0°C  (crit = +105.0°C)

nouveau-pci-0100
Adapter: PCI adapter
GPU core:     +0.90 V  (min =  +0.80 V, max =  +0.95 V)
temp1:        +65.0°C  (high = +95.0°C, hyst =  +3.0°C)
                       (crit = +105.0°C, hyst =  +5.0°C)
                       (emerg = +135.0°C, hyst =  +5.0°C)

Reading this, do you think the temperature is high?

For example, core 3 is 81ºC. Is that right?

Thx.

13

@paulomorais1981 Were these readings done with RT open? It hasn’t reached 84.0°C yet. If temps are usually on the high end, then it will wear down your hardware, especially if it is a laptop where everything is in close proximity. General info:

14

No, it was read with RT closed. But, anyway, I have never opened my hardware to clean it, and I distrust it’s the time. I’ve sent it to maintenance now, to clean it up, and then I will test it again. The fans are working, but maybe dirt is complicating the cpu cooling.

15

The temp is ok, the question is what does it do when it hits 85, does it keep rising or are fans kicking in? Cooling an i7 tower can be noisy. I usually setup cooling to have a minimum rpm that is quiet and keeps the idle temp warmish, then ramp up the fan speed to 100% once it hits 85. 100 degrees is not really a good max sensor temp, as it equals the max junction temp (the point at which the tiny integrated transistors melt down) for a i7-720QM. In this situation you figure a 5 degree drop between the case sensors and the die temp. Then you would definitely want to be already shut down at 95, and you might even give it some headroom and say shutdown when the case hits 90. This way, when I run an intense computation, I’ll hear the fans roar… otherwise they stay relatively quiet. The point at which you set the ramp up to 100% rpms depends how warm you want to keep your cpu, which balances with how quiet you want it. At the end of the day the cooling profile of your machine will definitely affect your cpu MTBF, and inversely, fan MTBF. If you’re on a laptop, you are really beholden to its fixed thermal solution. On a tower you can play with a variety of cooling methods that suite your use case.