Sometimes I get dotted coloured screen, too.

Here is a log just after freeze:

It seems related to:
/////////////////
6/18/17 7:40 PM kernel NVRM: GPU at PCI:0000:01:00: GPU-2ff1ac4a-cd6b-6004-afe3-22c95feda424
6/18/17 7:40 PM kernel NVRM: Xid (PCI:0000:01:00): 13, Graphics Exception: ChID 0001, Class 00005097, Offset 00001694, Data 00000100
6/18/17 7:42 PM kernel NVRM: Xid (PCI:0000:01:00): 13, Graphics Exception: ChID 0001, Class 00005097, Offset 00001694, Data 00000100
6/18/17 8:00 PM kernel NVRM: Xid (PCI:0000:01:00): 69, Class Error: ChId 0001, Class 0000502d, Offset 000005e0, Data 000c0004, ErrorCode 00000034
6/18/17 7:56 PM kernel NVRM: Xid (PCI:0000:01:00): 13, Graphics Exception: ChID 0001, Class 00005097, Offset 00001694, Data 00000100
6/18/17 8:00 PM kernel NVRM: Xid (PCI:0000:01:00): 8, Channel 00000003
6/18/17 8:00 PM kernel NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
6/18/17 8:00 PM kernel NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
6/18/17 8:00 PM kernel NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context

/////////////////

http://docs.nvidia.com/deploy/xid-errors/index.html

Please post your xorg.log.(see [1]) and 'pacman -Qs' for nvidia and libglvnd (when error occurs)
Last working packages/drivers?

[1] run `nvidia-bug-report.sh` - https://devtalk.nvidia.com/default/topic/522835/linux/if-you-have-a-problem-please-read-this-first/

Thank you Darek!
But nvidia blocked my county...
I didn't see useful information in xorg.log . I removed libglvnd, because have many problems with nvidia-340xx. I just create a symlink to /usr/lib/nvidia/libGL.so.

> I removed libglvnd, because have many problems with nvidia-340xx. I just create a symlink to /usr/lib/nvidia/libGL.so.
No wonder you have problems :)

Try this
1 remove the symbolic link you created
2 install mesa-noglvnd without mesa-libgl-noglvnd (see [1],[2],[*])
3 install/reinstall nvidia-340xx packages (340.102-8)
The nvidia-340 package contains a file which blacklists the nouveau module, so rebooting is necessary.
4 test

[1] https://aur.archlinux.org/packages/mesa-noglvnd/
[2] https://bbs.archlinux.org/viewtopic.php?id=225275
[*] some applications have a problem with 00-nvidia.conf (good example  FS#53623 )

I tested mesa-noglvnd and reinstalling nvidia. Unfortunately didn't solved.

Please add new logs after that.

log didn't changed (same old messages). I try to send a pic from capacitors. It seems 2 capacitors have brown head.

Hi, I am able to reproduce this bug (openbox WM)

It seems this is an NVIDIA driver bug

Summary:

Additional info:
* package version(s)
nvidia-340xx 340.102-8
nvidia-340xx-utils 340.102-8
mesa 17.1.2-1
libglvnd 0.2.999+g4ba53457-2
linux 4.11.6-1
xorg-server 1.19.3-2

* config and/or log files etc.
asus_laptop: M70Vm model detected
NVIDIA GPU GeForce 9600M GS (G96)

Not related to
-GLVND (libglvnd package)
-Nvidia-340xx - No matching fbConfigs https://bbs.archlinux.org/viewtopic.php?id=225275
- FS#53623 

Tested with: glxgears,es2tri,webgldemo [1]

[1] http://madebyevan.com/webgl-water/

Test #2 - my custom package
Xorg.0.log.old:
[ 189.805] (EE) NVIDIA(GPU-0): WAIT: (E, 0, 0x887d)
[ 191.807] (EE) NVIDIA(GPU-0): WAIT: (E, 0, 0x887d)
[ 195.810] (EE) NVIDIA(GPU-0): Failed to tear down EVO channel
[ 213.010] (II) NVIDIA(0): Setting mode "DFP-0:nvidia-auto-select"
[ 215.009] (EE) NVIDIA(GPU-0): WAIT: (E, 0, 0x887d)

I have some opengl errors in qtcreator ( FS#53380 ).
I thought it is related to hardware, but Why nouveau have not error?

Please report this upstream at [1] following the instructions at [2]. Thanks

[1] https://devtalk.nvidia.com/default/board/98/linux/
[2] https://devtalk.nvidia.com/default/topic/522835/linux/if-you-have-a-problem-please-read-this-first/

I'm going to close this bug. I'm pretty sure this is not a packaging bug unless you show me some evidence of that. You should report this directly to nvidia.